开云体育

Oh. Of course!

rogerw

what is safe??Where the green photograph?is taken is normally viewed by human operator's?bare eyes.

Everything X-Ray service tech does is safe, or he doesn't do it.

G

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 10:49:53 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

I guess it is safe?

wow!

rogerw

On 4/7/2022 10:34 AM, Geo Dowell wrote:

The photograph is of a model 07-451 X-Ray waveform detector. Every X-Ray Service Tech in Scotland carries on in his service case. You attach it via coax cable to an oscilloscope and turn on the X-Ray machine. The detector sends a waveform to the 'scope so its shape can be measured and examined.

If the machine happens to have an?(green screen)?X-Ray image intensifier, he can take a picture with a camera of what's inside the detector . A bunch of diodes. The gaps are where the junctions are, and you can see a faint outline of the epoxy packaging on each diode.

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:58:24 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

1) yes, could be multiple junctions.

2) You have an exray machine?

3) We used to "de-pot" plastic packaged IC chips with acid to expose the die in the package, to examine under a microscope, photograph, etc. trace circuits that way....... might be hard to do with an axial lead diode.... I don't know.

rogerw

On 4/7/2022 9:49 AM, Geo Dowell wrote:

I've always assumed the multiple junction idea, because in practice the higher voltage rectifier diodes get longer and longer. Ones for a 100kV stack are a good 6 inches long each. Don't know, just guessing. Will have to X-Ray one, but when it's out of the X-Ray machine how to do that?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:43:53 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well, the answer to your first question is: I am not sure, but I think it is probably just very lightly doped silicon pn junction. The light doping (and perhaps playing around with the doping profile - Its just fizzicks, ya' know!) gives higher reverse breakdown voltage. And the lighter doping also gives a higher bulk resistivity such that the extrinsic series resistance is higher than a LV diode, so higher forward drop at least is due to higher series resistance.

This is a "no-cost" contract. It is going to take an infinity of diodes!

rogerw

On 4/7/2022 9:32 AM, Geo Dowell wrote:

Question Roger, are HV (like 8kV) low current diodes a stack of normal PN junctions inside? or are they a single special PN junction? What I do measure is higher and higher Fwd V drop as the V (sub) max rating increases.

By the way, if you only need that many PN junctions to achieve thermal cancelling, try switching to Schottkey diodes. That will double or tripple the parts count. This is a "cost-plus-contract", right?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:06:12 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

?

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

?

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

开云体育

Geo,

With UnRegHV = 1300V and dropping resistor = 1.8Meg,

No load, 910.00V

Rload = 1G, 909.99V

Rload = 100Meg, 909.86V,

Rload = 10Meg, 909.01V.

If one were to measure on the other side of the 3.3Meg resistor going up the cable to the GM tube, a very different result would ensue. 1G would be very handy! But, if ne knows the numbers, one can calculate from the loaded measurement what the infinite input impedance
scope would have seen

rogerw

good question. I will hazard a MindSim answer, then will do the simulation.

As long? as the loading of the Zener String regulated node does not approach to close to the current flowing into the Zener, it will not drop the voltage too much.

(900+V/1G) << (1300V- 900+V)/1.8Meg? , or,

0.9uA << 222uA

So, the Zener String sees a very negligible change in current with a 1G loading at the anode of the string, since 222uA-0.9uA ~ 221uA is hardly any change. Even if you loaded it down to only 10uA in the Zener string you would not see a lot of change, as it depends on the regulation curve.

(By gosh! you just told me how I can measure the regulation curve over voltage with a fixed HV supply! Wallah!)

So, continuing Stream of Thought..... the more interesting case is not UnRegHV = 1300V but what if only 1000v? Then,

0.9uA << (1000V-900V)/1.8Meg? ,or,

0.9uA << 55.5uA

Still lots of room to load it and leave 10uA or more for the Zener String.

I actually think (I think, I am not sure) that the typical HV Generator is going to be current limited by the time it gets to maybe 20uA at some UnRegHV voltage well below the OpenCircuit voltage....... in that case:

0.9uA << 20uA (or whatever the current-limited number is. When I designed the Lionel and Vic HV Generators for lower power consumption, I shot for a 10uA output current.)

rogerw

On 4/7/2022 11:05 AM, Geo Dowell wrote:

I don't know, never saw that listed either.

Meantime do this simple sim for me.

+1300V DC to cathode of a Zener ( can be via? 1.8M Ohm res. or direct in this case only). Any Zener. On the other side of the Zener, a 1000 M Ohm load to ground(-).

What V does the infinite impedance scope probe show at the junction of anode and 1G load?

I am not good with simulators, but physically do this on the workbench every day.

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 10:46:43 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Ok, I got a bit curious...... If a Zener String of 900+V can change +/-40V over -25degC to +75degC, what did the original Corotrons do?

IE, a Victoreen GV3A Corotron?

It turns out that this Vic catalog does not give a spec on the basic GV3A used, but does give temp info on the "High Temperature" version:

? page 15

So, you can see an original 900V High Temp Corotron only varied 16V from -65degC to 150degC...... pretty impressive I think.

We are replacing these things with Zener Strings with relatively high positive tempco......

I think the ultimate economic answer is to regulate in the primary circuit, just as you suggested, and as ENi did, and depend on the fixed transformer secondary ratio to get the regulated HV, and to temperature compensate the primary along with an adjustable trim capability.

rogerw

On 4/7/2022 9:58 AM, Roger Whatley via groups.io wrote:

1) yes, could be multiple junctions.

2) You have an exray machine?

3) We used to "de-pot" plastic packaged IC chips with acid to expose the die in the package, to examine under a microscope, photograph, etc. trace circuits that way....... might be hard to do with an axial lead diode.... I don't know.

rogerw

On 4/7/2022 9:49 AM, Geo Dowell wrote:

I've always assumed the multiple junction idea, because in practice the higher voltage rectifier diodes get longer and longer. Ones for a 100kV stack are a good 6 inches long each. Don't know, just guessing. Will have to X-Ray one, but when it's out of the X-Ray machine how to do that?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:43:53 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well, the answer to your first question is: I am not sure, but I think it is probably just very lightly doped silicon pn junction. The light doping (and perhaps playing around with the doping profile - Its just fizzicks, ya' know!) gives higher reverse breakdown voltage. And the lighter doping also gives a higher bulk resistivity such that the extrinsic series resistance is higher than a LV diode, so higher forward drop at least is due to higher series resistance.

This is a "no-cost" contract. It is going to take an infinity of diodes!

rogerw

On 4/7/2022 9:32 AM, Geo Dowell wrote:

Question Roger, are HV (like 8kV) low current diodes a stack of normal PN junctions inside? or are they a single special PN junction? What I do measure is higher and higher Fwd V drop as the V (sub) max rating increases.

By the way, if you only need that many PN junctions to achieve thermal cancelling, try switching to Schottkey diodes. That will double or tripple the parts count. This is a "cost-plus-contract", right?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:06:12 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Roger and Geo:
?tried a simple series regulator
R6 is a dummy load of 100M( manually swept it down to 10k and seems to work)
Q2 is a cheapo 2n3904 and so are Q1 and Q3?
choice of Q2 depends on how much voltage needs to be dropped.?
If only 30V needs to be removed, ok.?
If 200V needs to be removed, need to pick a higher voltage tr for Q2
P

开云体育

good question. I will hazard a MindSim answer, then will do the simulation.

As long? as the loading of the Zener String regulated node does not approach to close to the current flowing into the Zener, it will not drop the voltage too much.

(900+V/1G) << (1300V- 900+V)/1.8Meg? , or,

0.9uA << 222uA

So, the Zener String sees a very negligible change in current with a 1G loading at the anode of the string, since 222uA-0.9uA ~ 221uA is hardly any change. Even if you loaded it down to only 10uA in the Zener string you would not see a lot of change, as it depends on the regulation curve.

(By gosh! you just told me how I can measure the regulation curve over voltage with a fixed HV supply! Wallah!)

So, continuing Stream of Thought..... the more interesting case is not UnRegHV = 1300V but what if only 1000v? Then,

0.9uA << (1000V-900V)/1.8Meg? ,or,

0.9uA << 55.5uA

Still lots of room to load it and leave 10uA or more for the Zener String.

I actually think (I think, I am not sure) that the typical HV Generator is going to be current limited by the time it gets to maybe 20uA at some UnRegHV voltage well below the OpenCircuit voltage....... in that case:

0.9uA << 20uA (or whatever the current-limited number is. When I designed the Lionel and Vic HV Generators for lower power consumption, I shot for a 10uA output current.)

rogerw

I don't know, never saw that listed either.

Meantime do this simple sim for me.

+1300V DC to cathode of a Zener ( can be via? 1.8M Ohm res. or direct in this case only). Any Zener. On the other side of the Zener, a 1000 M Ohm load to ground(-).

What V does the infinite impedance scope probe show at the junction of anode and 1G load?

I am not good with simulators, but physically do this on the workbench every day.

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 10:46:43 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Ok, I got a bit curious...... If a Zener String of 900+V can change +/-40V over -25degC to +75degC, what did the original Corotrons do?

IE, a Victoreen GV3A Corotron?

It turns out that this Vic catalog does not give a spec on the basic GV3A used, but does give temp info on the "High Temperature" version:

? page 15

So, you can see an original 900V High Temp Corotron only varied 16V from -65degC to 150degC...... pretty impressive I think.

We are replacing these things with Zener Strings with relatively high positive tempco......

I think the ultimate economic answer is to regulate in the primary circuit, just as you suggested, and as ENi did, and depend on the fixed transformer secondary ratio to get the regulated HV, and to temperature compensate the primary along with an adjustable trim capability.

rogerw

On 4/7/2022 9:58 AM, Roger Whatley via groups.io wrote:

1) yes, could be multiple junctions.

2) You have an exray machine?

3) We used to "de-pot" plastic packaged IC chips with acid to expose the die in the package, to examine under a microscope, photograph, etc. trace circuits that way....... might be hard to do with an axial lead diode.... I don't know.

rogerw

On 4/7/2022 9:49 AM, Geo Dowell wrote:

I've always assumed the multiple junction idea, because in practice the higher voltage rectifier diodes get longer and longer. Ones for a 100kV stack are a good 6 inches long each. Don't know, just guessing. Will have to X-Ray one, but when it's out of the X-Ray machine how to do that?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:43:53 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well, the answer to your first question is: I am not sure, but I think it is probably just very lightly doped silicon pn junction. The light doping (and perhaps playing around with the doping profile - Its just fizzicks, ya' know!) gives higher reverse breakdown voltage. And the lighter doping also gives a higher bulk resistivity such that the extrinsic series resistance is higher than a LV diode, so higher forward drop at least is due to higher series resistance.

This is a "no-cost" contract. It is going to take an infinity of diodes!

rogerw

On 4/7/2022 9:32 AM, Geo Dowell wrote:

Question Roger, are HV (like 8kV) low current diodes a stack of normal PN junctions inside? or are they a single special PN junction? What I do measure is higher and higher Fwd V drop as the V (sub) max rating increases.

By the way, if you only need that many PN junctions to achieve thermal cancelling, try switching to Schottkey diodes. That will double or tripple the parts count. This is a "cost-plus-contract", right?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:06:12 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

开云体育

1) Yes, PiN with an intrinsic layer is essentially "very very very lightly doped" and therefore HV. I should have thought of that, mind is getting rusty... thanks for bringing that up. PiN is also very low capacitance in reverse bias, often used in very high frequency applications.

Also, I once had an idea to use a large-area PiN diode in reverse bias as a radiation detector! (it should work. I don't know how well.)

2) Yes! you can use most any HV device for the cascode or for the series pass element, except a npn BJT will introduce new base current error and would need to be a darlington.

I recommend what Michael found at digikey:

One is depletion, one is enhancement, both are MosFet.

I purchased some of the cheaper depletion mode Mosfets. I only used the n-ch JFET in the simulation because it is also depletion mode (I could not find a depletion mode MosFet in the sim library to use.).

rogerw

Roger et al:
?Some hv diodes are PiN, there is an intrinsic band.?
Supposedly the IN4007 is one of the commonly available diodes which some people claimed had an "i" layer. Used them as low cost RF PiN diodes.

My mind is wandering again. I have to look at Roger's idea of a series regulator again. Is it possible to use some more common part in stead of the JFET pass transistor? in your circuit ?

The actual voltage drop across the pass transistor is small so some more generic part could be used ?

P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Roger et al:
?Some hv diodes are PiN, there is an intrinsic band.?
Supposedly the IN4007 is one of the commonly available diodes which some people claimed had an "i" layer. Used them as low cost RF PiN diodes.

My mind is wandering again. I have to look at Roger's idea of a series regulator again. Is it possible to use some more common part in stead of the JFET pass transistor? in your circuit ?

The actual voltage drop across the pass transistor is small so some more generic part could be used ?

P

开云体育

I guess it is safe?

wow!

rogerw

The photograph is of a model 07-451 X-Ray waveform detector. Every X-Ray Service Tech in Scotland carries on in his service case. You attach it via coax cable to an oscilloscope and turn on the X-Ray machine. The detector sends a waveform to the 'scope so its shape can be measured and examined.

If the machine happens to have an?(green screen)?X-Ray image intensifier, he can take a picture with a camera of what's inside the detector . A bunch of diodes. The gaps are where the junctions are, and you can see a faint outline of the epoxy packaging on each diode.

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:58:24 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

1) yes, could be multiple junctions.

2) You have an exray machine?

3) We used to "de-pot" plastic packaged IC chips with acid to expose the die in the package, to examine under a microscope, photograph, etc. trace circuits that way....... might be hard to do with an axial lead diode.... I don't know.

rogerw

On 4/7/2022 9:49 AM, Geo Dowell wrote:

I've always assumed the multiple junction idea, because in practice the higher voltage rectifier diodes get longer and longer. Ones for a 100kV stack are a good 6 inches long each. Don't know, just guessing. Will have to X-Ray one, but when it's out of the X-Ray machine how to do that?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:43:53 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well, the answer to your first question is: I am not sure, but I think it is probably just very lightly doped silicon pn junction. The light doping (and perhaps playing around with the doping profile - Its just fizzicks, ya' know!) gives higher reverse breakdown voltage. And the lighter doping also gives a higher bulk resistivity such that the extrinsic series resistance is higher than a LV diode, so higher forward drop at least is due to higher series resistance.

This is a "no-cost" contract. It is going to take an infinity of diodes!

rogerw

On 4/7/2022 9:32 AM, Geo Dowell wrote:

Question Roger, are HV (like 8kV) low current diodes a stack of normal PN junctions inside? or are they a single special PN junction? What I do measure is higher and higher Fwd V drop as the V (sub) max rating increases.

By the way, if you only need that many PN junctions to achieve thermal cancelling, try switching to Schottkey diodes. That will double or tripple the parts count. This is a "cost-plus-contract", right?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:06:12 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

?

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

?

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

开云体育

Geo,

Ok, I got a bit curious...... If a Zener String of 900+V can change +/-40V over -25degC to +75degC, what did the original Corotrons do?

IE, a Victoreen GV3A Corotron?

It turns out that this Vic catalog does not give a spec on the basic GV3A used, but does give temp info on the "High Temperature" version:

? page 15

So, you can see an original 900V High Temp Corotron only varied 16V from -65degC to 150degC...... pretty impressive I think.

We are replacing these things with Zener Strings with relatively high positive tempco......

I think the ultimate economic answer is to regulate in the primary circuit, just as you suggested, and as ENi did, and depend on the fixed transformer secondary ratio to get the regulated HV, and to temperature compensate the primary along with an adjustable trim capability.

rogerw

1) yes, could be multiple junctions.

2) You have an exray machine?

3) We used to "de-pot" plastic packaged IC chips with acid to expose the die in the package, to examine under a microscope, photograph, etc. trace circuits that way....... might be hard to do with an axial lead diode.... I don't know.

rogerw

On 4/7/2022 9:49 AM, Geo Dowell wrote:

I've always assumed the multiple junction idea, because in practice the higher voltage rectifier diodes get longer and longer. Ones for a 100kV stack are a good 6 inches long each. Don't know, just guessing. Will have to X-Ray one, but when it's out of the X-Ray machine how to do that?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:43:53 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well, the answer to your first question is: I am not sure, but I think it is probably just very lightly doped silicon pn junction. The light doping (and perhaps playing around with the doping profile - Its just fizzicks, ya' know!) gives higher reverse breakdown voltage. And the lighter doping also gives a higher bulk resistivity such that the extrinsic series resistance is higher than a LV diode, so higher forward drop at least is due to higher series resistance.

This is a "no-cost" contract. It is going to take an infinity of diodes!

rogerw

On 4/7/2022 9:32 AM, Geo Dowell wrote:

Question Roger, are HV (like 8kV) low current diodes a stack of normal PN junctions inside? or are they a single special PN junction? What I do measure is higher and higher Fwd V drop as the V (sub) max rating increases.

By the way, if you only need that many PN junctions to achieve thermal cancelling, try switching to Schottkey diodes. That will double or tripple the parts count. This is a "cost-plus-contract", right?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:06:12 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

开云体育

1) yes, could be multiple junctions.

2) You have an exray machine?

3) We used to "de-pot" plastic packaged IC chips with acid to expose the die in the package, to examine under a microscope, photograph, etc. trace circuits that way....... might be hard to do with an axial lead diode.... I don't know.

rogerw

I've always assumed the multiple junction idea, because in practice the higher voltage rectifier diodes get longer and longer. Ones for a 100kV stack are a good 6 inches long each. Don't know, just guessing. Will have to X-Ray one, but when it's out of the X-Ray machine how to do that?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:43:53 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well, the answer to your first question is: I am not sure, but I think it is probably just very lightly doped silicon pn junction. The light doping (and perhaps playing around with the doping profile - Its just fizzicks, ya' know!) gives higher reverse breakdown voltage. And the lighter doping also gives a higher bulk resistivity such that the extrinsic series resistance is higher than a LV diode, so higher forward drop at least is due to higher series resistance.

This is a "no-cost" contract. It is going to take an infinity of diodes!

rogerw

On 4/7/2022 9:32 AM, Geo Dowell wrote:

Question Roger, are HV (like 8kV) low current diodes a stack of normal PN junctions inside? or are they a single special PN junction? What I do measure is higher and higher Fwd V drop as the V (sub) max rating increases.

By the way, if you only need that many PN junctions to achieve thermal cancelling, try switching to Schottkey diodes. That will double or tripple the parts count. This is a "cost-plus-contract", right?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:06:12 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

开云体育

Well, the answer to your first question is: I am not sure, but I think it is probably just very lightly doped silicon pn junction. The light doping (and perhaps playing around with the doping profile - Its just fizzicks, ya' know!) gives higher reverse breakdown voltage. And the lighter doping also gives a higher bulk resistivity such that the extrinsic series resistance is higher than a LV diode, so higher forward drop at least is due to higher series resistance.

This is a "no-cost" contract. It is going to take an infinity of diodes!

rogerw

Question Roger, are HV (like 8kV) low current diodes a stack of normal PN junctions inside? or are they a single special PN junction? What I do measure is higher and higher Fwd V drop as the V (sub) max rating increases.

By the way, if you only need that many PN junctions to achieve thermal cancelling, try switching to Schottkey diodes. That will double or tripple the parts count. This is a "cost-plus-contract", right?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Thursday, April 7, 2022 9:06:12 AM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

开云体育

Geo,

My thoughts which did not make it into the previous note:

1) The largest Zener in my library appropriate to use was 36V. In fact I have used 100V, 200V, 300V, and 330V Zeners. (These last two are not sold as Zeners, per se, but as protection diodes. of course at above about 5V, they are all dominated by avalanche breakdown, not Zener.)

2) What is the tempco of three 330V Zeners in a string?? Elephino! (what do you get when you cross an elephant and a rhino? Elephino!) I guess I could take the time to do some heatgun experimentation..... not today.

3) There may be a better scheme for temperature compensation. I have been thinking about how to use a thermister.

4) none of this yet seems to practical.

5) this is all Stream of Consciousness, so far!

rogerw

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

rogerw

On 4/7/2022 3:36 AM, Geo Dowell wrote:

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

开云体育

Geo,

Sure, enclosed, but I have already modified the Zener String with a temperature compensation scheme that will be somewhat familiar to the discerning eye. I am NOT suggesting this circuit for a build, merely playing around with ideas!

So, if I measure the voltage of the top of that Zener String alone, it is a nominal 788V at 25degC, and 821V at 75degC, and 754V at-25degC. That is a +67V change over 100degC or about +0.67V/degC tempco. (parenthetically, the original full Zener String was about +/-40V.)? If I compensate that with the negative tempco of a Vbe which is about -2.2mv/degC,

I need 0.67V/2.2mV = 304 Vbes...... whew! That's a lot of diodes!

Instead the BJT circuit is multiplying the sum of Q1&Q3 Vbe by 129 = 258Vbe's. All I can say about the difference from calculated is that the actual tempco of the bjt's is in the model somewhere and differs from my Mindsim approximation. And btw, the pnp BJT Q2 is a local feedback trick that is "fortifying" Q1...... without Q2 you get weak knee, with Q2 you get a sharp knee. Q2 and R3 can be removed and the circuit still functions but regulation suffers.

Anyway, the second enclosure shows the regulation performance over temperature. When I expand the scale I can see the regulated voltage at 906V nominal 25degC, 908V at 75degC, and 905V at -25degC.

So, for one thing, this whole exercise begs a question: Who lets their GM Counter get THAT hot or cold? I think that is a philosophical question.... If a GM Counter fell in the forest, in the winter up north, and there was not a woman there to turn it on, does it really matter? Does it even exist???

rogerw

Roger, show schematic of sim?

Geo

---

From: "Roger Whatley" <rogwhat53@...>
To: "CDV700CLUB" <[email protected]>
Sent: Wednesday, April 6, 2022 9:28:41 PM
Subject: Re: [CDV700CLUB] Question and an interesting site for those that modify CDV700s and more.

Well I just simulated something I had not yet thought to do...... I simulated a Zener string.

Yeah, it worked...... ha!

Largest Zener I could find in the model library was 36V so I strung enough of those to make about 890V nominal with a 3.3Meg dropper resistor. Its regulation is very flat only rose about 3V from 1000V to 1200V input. That as good as my Penultimate Way-Good Cascode Shunt Regulator can do, and far less complex.

But guess what? over temperature it sucks big wind! +/-40V from -25 to 25 to 75degC! The Penultimate only varies about +/-3V over that temp range (overkill).

But the Good Enough Cascode Shunt Regulator, with its temperature compensation scheme, only varies less tha +/-5V. Regulation varies less than 10V from the knee to 950 to 1200V

On 4/6/2022 10:53 AM, Roger Whatley via groups.io wrote:

Peter,

Minor change, I noticed after I posted the note below that I had not specified the second BJT in the Darlington. The enclosed schematic is corrected and now the simulation predicts about dead-nuts on for the tempco of the output. Other than that, same comments as below. I think this is the best "good enough" cascoded shunt regulator.

rogerw

On 4/6/2022 10:40 AM, Roger Whatley via groups.io wrote:

Peter,Enclosed is a schematic of my suggestion below. That is an 8.2V Rohm Zener chosen for tempco to cancel the negative tempco of two Vbe's. It is slightly too positive in the result but quite "good enough." You willr just get much flatter regulation with the Darlington pair and that can be had in a single package.

Of course, depending on the Vcemax spec of your cascode MOSFET, you might have only one cascode instead of two..... That will not make a noticeable difference except to breakdown.

rogerw

On 4/6/2022 8:32 AM, Roger Whatley via groups.io wrote:

Peter,

I realize that circuit described below might be "good enough," but realize that making that one BJT a Darlington pair is a significant improvement to the regulation. You can even get the Darlington in one transistor package, of course, cheap.

As UnRegHV increases, the current in the BJT necessarily increases in order to regulate the loop. For every extra 0.1uA of extra base current required, the voltage drop in the feedback resistors (66M + 66M + 1M) goes up 0.1uA*(66M+66M+1M) = 13.3V and thus regulation suffers. The BJT current easliy goes from less than 10uA to over 200uA, thus the base current might go up by a worst-case of as much as 1uA, giving a 133V change in the output. So that is the motivation for the Darlington.

(In case anyone wonders, why not an FET instead of a BJT, that WAS the original circuit and it suffers worse circuit errors.....)

The next big error is that as the BJT current increases with UnRegHV increases, so does Vbe and Vz increase. Those are the motivations for subsequent circuit complications and resulting in the Penultimate Way-Improved version..... it is overkill, I know, to the requirement, but by simulating those improvements I gain insights as to just what the remaining errors were. And it was fun!

My take is that your present circuit, but with a Darlington Pair, will be "good enough" and if it were a purpose-built Darlington in one three-lead package, no more expensive or even board-space than a single BJT.

Of course, a single Zener voltage may be chosen to match the negative tempco of two Vbe's. Again, no extra expense or board space. (Off the top of my head, I am thinking about 8V Zener for +4mV/degC?)

rogerw

On 4/4/2022 10:33 PM, peter via groups.io wrote:

Roger:
?back to the circuit you posted in message #300
Simple cascode, with only 1 npn tr + 2 mosfets.
Since I found the 2SK3265, I removed the 2nd mosfet so its very basic circuit.
66Meg, two 1 Meg, 100k+ 2.2M trimmer , 5.1V zener, 2SK3265, 2SC3311
Set trimmer for 900V, Vary input from 950~ 1150, output only varied 5V or less.
Heated the circuit with a hair dryer, got it up from 25C to about 60C, voltage rose may be 5 or 6 volts. Good enough for me!
Time to make it small enough to stuff inside my CDV-700 !
It on the LHS of the attached jpg, circuit in message #300 is on the RHS
P

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Virus-free.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

-- 
rogerw
On the Catclaw
ΜΟΛΩΝ ΛΑΒΕ

The PanDemic is over, but the DemPanic goes on.
MisDisMal is a dictator's tool of repression.

Pic of the Si-3BG version, and the now standard arrangement. The bottom clip has been moved down all the way to the base, the 900 to 400V converter is simplified, miniaturized, and installed inside the black 3 pin base. This new profile, with its overwrap plastic insulating sleeve (not shown) plugs directly into an unmodified CDV-700 probe housing.

Same basic layout can be modified for the SBM-20 tube by using a longer PCB, requiring a shorter extension at the top, to just the plastic cap. The orange material is Kapton HV tape and secures the Teflon insulated ground wire going to the top clip. The whole adaptor slips into an insulating plastic sleeve when installed into the CDV-700 probe/