|
I have a mostly functional Tektronix 184 Time-Mark Generator that I am
restoring. It has six 7587 nuvistor tetrodes and they all appear to
work except for maybe the pair in the 500 MHz multiplier.
You can still pick up used but probably good and NOS 7587 nuvistors
for reasonable prices but how feasible would it be to replace them
with say JFETs with cascodes?
I will probably try it no matter what is said here just for the fun of
it but any advice would be appreciated.
|
I'm not familiar with the 184, but it seems that for a 500 MHz type circuit, it's low-Z RF anyway, so you could probably design in some bipolar transistor circuits to replace the tubes, with new bias and much lower operating voltage, and some changes to RF matching components - easier yet if they're class-C amplifiers.
UHF JFETs should work OK as amplifiers, but may not have enough power output available - it depends on the circuit RF power levels needed. Even for low power, you'll probably still have to significantly change the bias and operating voltage range. Big MESFETs would be way overkill, and probably unsuitable for the operating voltages and currents available in the 184.
With the vast number of choices in bipolar NPN RF transistors, it should be fairly easy to match the power of a 7587, as long as you can change enough other stuff to go with it.
Ed
toggle quoted message
Show quoted text
--- In TekScopes@..., David <davidwhess@...> wrote:
I have a mostly functional Tektronix 184 Time-Mark Generator that I am
restoring. It has six 7587 nuvistor tetrodes and they all appear to
work except for maybe the pair in the 500 MHz multiplier.
You can still pick up used but probably good and NOS 7587 nuvistors
for reasonable prices but how feasible would it be to replace them
with say JFETs with cascodes?
I will probably try it no matter what is said here just for the fun of
it but any advice would be appreciated.
|
I should have linked this in my original post. Here is a cleaned up copy of the 184 schematics that I made when I printed them out. The tetrodes are all on schematic 1: The full 184 manual is available here: The control grid DC biasing is all high impedance so unless I want to change that, I will have to stick with FETs. The screen grid DC biasing is not much lower. After further thought, I believe that a dual JFET cascode would work as a drop in replacement for each one if the anode supply is lowered. A source degeneration resistor can be added to lower the gain if necessary. The screen grid circuit voltage would need to be about half of the drain supply. On Wed, 02 Jan 2013 02:01:48 -0000, "Ed Breya" <edbreya@...> wrote: I'm not familiar with the 184, but it seems that for a 500 MHz type circuit, it's low-Z RF anyway, so you could probably design in some bipolar transistor circuits to replace the tubes, with new bias and much lower operating voltage, and some changes to RF matching components - easier yet if they're class-C amplifiers.
UHF JFETs should work OK as amplifiers, but may not have enough power output available - it depends on the circuit RF power levels needed. Even for low power, you'll probably still have to significantly change the bias and operating voltage range. Big MESFETs would be way overkill, and probably unsuitable for the operating voltages and currents available in the 184.
With the vast number of choices in bipolar NPN RF transistors, it should be fairly easy to match the power of a 7587, as long as you can change enough other stuff to go with it.
Ed
--- In TekScopes@..., David <davidwhess@...> wrote:
I have a mostly functional Tektronix 184 Time-Mark Generator that I am
restoring. It has six 7587 nuvistor tetrodes and they all appear to
work except for maybe the pair in the 500 MHz multiplier.
You can still pick up used but probably good and NOS 7587 nuvistors
for reasonable prices but how feasible would it be to replace them
with say JFETs with cascodes?
I will probably try it no matter what is said here just for the fun of
it but any advice would be appreciated.
|
Change the 125 volt plate supply on the switch to 12 volts. Leave the
screen open, it's not needed. Just put a pair of RF N-Ch e-mode fets to the
plate, grid, and cathode and tune away. Pick a fet that runs about 10 ma at
Vgs=0. It's worth a try if you can't find the tubes.
?
Good luck,
Tom
?
toggle quoted message
Show quoted text
----- Original Message -----
Sent: Wednesday, January 02, 2013 1:02
AM
Subject: Re: [TekScopes] Re: 7587
Nuvistor Tetrode Substitutions in the Tektronix 184 Time-Mark Generator
?
I should have linked this in my original post. Here is a cleaned up
copy
of the 184 schematics that I made when I printed them out. The
tetrodes are
all on schematic 1:
The
full 184 manual is available here:
The
control grid DC biasing is all high impedance so unless I want to
change
that, I will have to stick with FETs. The screen grid DC
biasing is not
much lower.
After further thought, I believe that a dual JFET cascode
would work
as a drop in replacement for each one if the anode supply is
lowered.
A source degeneration resistor can be added to lower the gain
if
necessary. The screen grid circuit voltage would need to be
about
half of the drain supply.
On Wed, 02 Jan 2013 02:01:48 -0000,
"Ed Breya" <edbreya@...>
wrote:
>I'm
not familiar with the 184, but it seems that for a 500 MHz type circuit, it's
low-Z RF anyway, so you could probably design in some bipolar transistor
circuits to replace the tubes, with new bias and much lower operating voltage,
and some changes to RF matching components - easier yet if they're class-C
amplifiers.
>
>UHF JFETs should work OK as amplifiers, but may not
have enough power output available - it depends on the circuit RF power levels
needed. Even for low power, you'll probably still have to significantly change
the bias and operating voltage range. Big MESFETs would be way overkill, and
probably unsuitable for the operating voltages and currents available in the
184.
>
>With the vast number of choices in bipolar NPN RF
transistors, it should be fairly easy to match the power of a 7587, as long as
you can change enough other stuff to go with
it.
>
>Ed
>
>--- In TekScopes@..., David
wrote:
>>
>> I have a mostly
functional Tektronix 184 Time-Mark Generator that I am
>> restoring.
It has six 7587 nuvistor tetrodes and they all appear to
>> work
except for maybe the pair in the 500 MHz multiplier.
>>
>>
You can still pick up used but probably good and NOS 7587
nuvistors
>> for reasonable prices but how feasible would it be to
replace them
>> with say JFETs with cascodes?
>>
>> I will probably try it no matter what is said here just for the
fun of
>> it but any advice would be
appreciated.
>>
>
|
Dunno if it'll be helpful or not but I'd had good results with LND150 depletion mode mosfets in various lower frequency tube circuits. They act a lot like a pentode in-circuit, with no screen connection, of course. Good to 450 volts.
Jamie
toggle quoted message
Show quoted text
--- In TekScopes@..., David <davidwhess@...> wrote:
I have a mostly functional Tektronix 184 Time-Mark Generator that I am
restoring. It has six 7587 nuvistor tetrodes and they all appear to
work except for maybe the pair in the 500 MHz multiplier.
You can still pick up used but probably good and NOS 7587 nuvistors
for reasonable prices but how feasible would it be to replace them
with say JFETs with cascodes?
I will probably try it no matter what is said here just for the fun of
it but any advice would be appreciated.
|
It looks like the required RF power levels aren't out of reach for JFET substitution of the 7587 tubes. The tricky ones will be the 500 MHz multiplier and the 100 MHz driver for the 200 MHz diode doubler.
None of the tubes use any cathode degeneration, and are only biased by grid current through the grounded 100k resistors, so dropping JFETs in their place means that they'll just run at Idss. For the 500 and 100/200 MHz circuits, the more, the better - in this case, high gain and nonlinearity are your friends. If you get a bunch of JFETs like U300, with Idss near the top of the range (>40 mA), they should be OK in those spots. If you come up short on power, you can try paralleling two devices with the same Idss for each spot. The other three tube spots have more modest power requirements, so lighter JFETs can be used there. For drain supplies, the +12V is the only other positive supply already available, which should work OK in place of the original +125V plate supplies.
One thing that will complicate matters is the disabling of certain sections when needed. The 100, 50, and 20 MHz amplifiers appear to be cut off when deselected, by grounding their screen grids. You would have to replicate this action in the new circuitry if you want to keep it operating the same as original.
Ed
toggle quoted message
Show quoted text
--- In TekScopes@..., David <davidwhess@...> wrote:
I should have linked this in my original post. Here is a cleaned up
copy of the 184 schematics that I made when I printed them out. The
tetrodes are all on schematic 1:
The full 184 manual is available here:
The control grid DC biasing is all high impedance so unless I want to
change that, I will have to stick with FETs. The screen grid DC
biasing is not much lower.
After further thought, I believe that a dual JFET cascode would work
as a drop in replacement for each one if the anode supply is lowered.
A source degeneration resistor can be added to lower the gain if
necessary. The screen grid circuit voltage would need to be about
half of the drain supply.
On Wed, 02 Jan 2013 02:01:48 -0000, "Ed Breya" <edbreya@...>
wrote:
I'm not familiar with the 184, but it seems that for a 500 MHz type circuit, it's low-Z RF anyway, so you could probably design in some bipolar transistor circuits to replace the tubes, with new bias and much lower operating voltage, and some changes to RF matching components - easier yet if they're class-C amplifiers.
UHF JFETs should work OK as amplifiers, but may not have enough power output available - it depends on the circuit RF power levels needed. Even for low power, you'll probably still have to significantly change the bias and operating voltage range. Big MESFETs would be way overkill, and probably unsuitable for the operating voltages and currents available in the 184.
With the vast number of choices in bipolar NPN RF transistors, it should be fairly easy to match the power of a 7587, as long as you can change enough other stuff to go with it.
Ed
--- In TekScopes@..., David <davidwhess@> wrote:
I have a mostly functional Tektronix 184 Time-Mark Generator that I am
restoring. It has six 7587 nuvistor tetrodes and they all appear to
work except for maybe the pair in the 500 MHz multiplier.
You can still pick up used but probably good and NOS 7587 nuvistors
for reasonable prices but how feasible would it be to replace them
with say JFETs with cascodes?
I will probably try it no matter what is said here just for the fun of
it but any advice would be appreciated.
|
Did you have a particular FET in mind? All of the enhancement mode FETs I am familiar with have way too high a gate to source threshold voltage. Small signal and power depletion mode MOSFETs used to be available which would have been perfect or at least worth trying but they are no longer commonly available. On Wed, 02 Jan 2013 02:21:17 -0500, "Tom Miller" <tmiller11147@...> wrote: Change the 125 volt plate supply on the switch to 12 volts. Leave the screen open, it's not needed. Just put a pair of RF N-Ch e-mode fets to the plate, grid, and cathode and tune away. Pick a fet that runs about 10 ma at Vgs=0. It's worth a try if you can't find the tubes.
Good luck,
Tom
----- Original Message -----
From: David
To: TekScopes@...
Sent: Wednesday, January 02, 2013 1:02 AM
Subject: Re: [TekScopes] Re: 7587 Nuvistor Tetrode Substitutions in the Tektronix 184 Time-Mark Generator
I should have linked this in my original post. Here is a cleaned up
copy of the 184 schematics that I made when I printed them out. The
tetrodes are all on schematic 1:
The full 184 manual is available here:
The control grid DC biasing is all high impedance so unless I want to
change that, I will have to stick with FETs. The screen grid DC
biasing is not much lower.
After further thought, I believe that a dual JFET cascode would work
as a drop in replacement for each one if the anode supply is lowered.
A source degeneration resistor can be added to lower the gain if
necessary. The screen grid circuit voltage would need to be about
half of the drain supply.
On Wed, 02 Jan 2013 02:01:48 -0000, "Ed Breya" <edbreya@...>
wrote:
>I'm not familiar with the 184, but it seems that for a 500 MHz type circuit, it's low-Z RF anyway, so you could probably design in some bipolar transistor circuits to replace the tubes, with new bias and much lower operating voltage, and some changes to RF matching components - easier yet if they're class-C amplifiers.
>
>UHF JFETs should work OK as amplifiers, but may not have enough power output available - it depends on the circuit RF power levels needed. Even for low power, you'll probably still have to significantly change the bias and operating voltage range. Big MESFETs would be way overkill, and probably unsuitable for the operating voltages and currents available in the 184.
>
>With the vast number of choices in bipolar NPN RF transistors, it should be fairly easy to match the power of a 7587, as long as you can change enough other stuff to go with it.
>
>Ed
>
>--- In TekScopes@..., David <davidwhess@...> wrote:
>>
>> I have a mostly functional Tektronix 184 Time-Mark Generator that I am
>> restoring. It has six 7587 nuvistor tetrodes and they all appear to
>> work except for maybe the pair in the 500 MHz multiplier.
>>
>> You can still pick up used but probably good and NOS 7587 nuvistors
>> for reasonable prices but how feasible would it be to replace them
>> with say JFETs with cascodes?
>>
>> I will probably try it no matter what is said here just for the fun of
>> it but any advice would be appreciated.
|
Hi David,
?
My brain fart there. I meant to say D-mode.
?
Sorry,
Tom
?
?
toggle quoted message
Show quoted text
----- Original Message -----
Sent: Wednesday, January 02, 2013 2:12
PM
Subject: Re: [TekScopes] Re: 7587
Nuvistor Tetrode Substitutions in the Tektronix 184 Time-Mark Generator
?
Did you have a particular FET in mind? All of the enhancement mode
FETs
I am familiar with have way too high a gate to source threshold
voltage.
Small signal and power depletion mode MOSFETs used to be
available which
would have been perfect or at least worth trying but
they are no longer
commonly available.
On Wed, 02 Jan 2013 02:21:17 -0500, "Tom
Miller"
<tmiller11147@...>
wrote:
>Change the 125 volt plate supply on the switch to 12 volts.
Leave the screen open, it's not needed. Just put a pair of RF N-Ch e-mode fets
to the plate, grid, and cathode and tune away. Pick a fet that runs about 10
ma at Vgs=0. It's worth a try if you can't find the tubes.
>
>Good
luck,
>Tom
>
> ----- Original Message -----
> From:
David
> To: TekScopes@...
> Sent: Wednesday, January 02, 2013 1:02 AM
> Subject: Re:
[TekScopes] Re: 7587 Nuvistor Tetrode Substitutions in the Tektronix 184
Time-Mark Generator
>
> I should have linked this in my original
post. Here is a cleaned up
> copy of the 184 schematics that I made when
I printed them out. The
> tetrodes are all on schematic
1:
>
>
>
>
The full 184 manual is available here:
>
>
>
>
The control grid DC biasing is all high impedance so unless I want to
>
change that, I will have to stick with FETs. The screen grid DC
>
biasing is not much lower.
>
> After further thought, I believe
that a dual JFET cascode would work
> as a drop in replacement for each
one if the anode supply is lowered.
> A source degeneration resistor can
be added to lower the gain if
> necessary. The screen grid circuit
voltage would need to be about
> half of the drain
supply.
>
> On Wed, 02 Jan 2013 02:01:48 -0000, "Ed Breya" <edbreya@...>
>
wrote:
>
> >I'm not familiar with the 184, but it seems that
for a 500 MHz type circuit, it's low-Z RF anyway, so you could probably design
in some bipolar transistor circuits to replace the tubes, with new bias and
much lower operating voltage, and some changes to RF matching components -
easier yet if they're class-C amplifiers.
> >
> >UHF JFETs
should work OK as amplifiers, but may not have enough power output available -
it depends on the circuit RF power levels needed. Even for low power, you'll
probably still have to significantly change the bias and operating voltage
range. Big MESFETs would be way overkill, and probably unsuitable for the
operating voltages and currents available in the 184.
> >
>
>With the vast number of choices in bipolar NPN RF transistors, it should
be fairly easy to match the power of a 7587, as long as you can change enough
other stuff to go with it.
> >
> >Ed
> >
>
>--- In TekScopes@..., David
wrote:
> >>
> >> I have a
mostly functional Tektronix 184 Time-Mark Generator that I am
> >>
restoring. It has six 7587 nuvistor tetrodes and they all appear to
>
>> work except for maybe the pair in the 500 MHz multiplier.
>
>>
> >> You can still pick up used but probably good and
NOS 7587 nuvistors
> >> for reasonable prices but how feasible
would it be to replace them
> >> with say JFETs with
cascodes?
> >>
> >> I will probably try it no matter
what is said here just for the fun of
> >> it but any advice would
be appreciated.
|
On Wed, 02 Jan 2013 17:55:38 -0000, "Ed Breya" <edbreya@...> wrote: It looks like the required RF power levels aren't out of reach for JFET substitution of the 7587 tubes. The tricky ones will be the 500 MHz multiplier and the 100 MHz driver for the 200 MHz diode doubler. Coincidently those are the stages that have been giving me trouble. I have to break out one of my faster oscilloscopes to see what is going on with them. It is possible that the 500 MHz multiplier just needs tuning but so far the output from the 200 MHz diode doubler has been low and not real symmetrical. Swapping V40 and V30 made no difference but maybe both are weak. The diodes in the doubler do not test as obviously bad in circuit but I may try and find higher performance replacements. None of the tubes use any cathode degeneration, and are only biased by grid current through the grounded 100k resistors, so dropping JFETs in their place means that they'll just run at Idss. For the 500 and 100/200 MHz circuits, the more, the better - in this case, high gain and nonlinearity are your friends. If you get a bunch of JFETs like U300, with Idss near the top of the range (>40 mA), they should be OK in those spots. If you come up short on power, you can try paralleling two devices with the same Idss for each spot. The other three tube spots have more modest power requirements, so lighter JFETs can be used there. For drain supplies, the +12V is the only other positive supply already available, which should work OK in place of the original +125V plate supplies. I guess I should go through and measure the operating points via the grid voltages and maybe plate currents to get a better idea of what is going on. For the drain supplies, I was thinking of just adding a low quiescent current series regulator to the 125 volt plate supply and dropping it directly to a voltage compatible with JFET operation. Normally I would eschew such a wasteful configuration but the original design does essentially the same thing via the plate resistance and quiescent current of each stage anyway which may be why the design only enables each stage as needed. One thing that will complicate matters is the disabling of certain sections when needed. The 100, 50, and 20 MHz amplifiers appear to be cut off when deselected, by grounding their screen grids. You would have to replicate this action in the new circuitry if you want to keep it operating the same as original. I noticed that every stage is disabled when not in use either via the screen grid or in the case of the 500 MHz multiplier, both the screen grid and anode supply. That is one of the reasons I wanted to try using two JFETs or a JFET and bipolar per stage in a cascode configuration which would allow the screen grid circuit to disable each stage as designed. Cascodes would also offer much higher performance at least as far as gain at high frequencies. The higher needed voltage is obviously not a problem in this case.
|
You may want to try swapping tubes from the other spots too, although it seems they're all biased for the same operating conditions, so should wear at the same rate - but if they are only on when needed for their specific function, then it all depends on the history of use. I think the only one that's always on then, would be the 10 MHz oscillator, so it should have the most wear.
I looked up the 7587 and see that it's rated for 20 mA max cathode current, but there were no charts showing what it should be near zero grid bias - unless the grid current gets pretty high, it doesn't seem that it will make much into 100k. It would be interesting to see what it is if you measure it and the actual plate currents.
I think your cascode version with a JFET and an NPN would be best, since you can then assure cutoff by pulling the base to ground. The double-JFET form won't positively cut off unless you can pull the gate negative, making it more complicated. You will have to watch out for the voltage ratings of the NPN to make sure it can handle the cutoff conditions.
I'd recommend around +15V or so for the bases (via small R so they won't oscillate on their own), and about +25V for the collector supply. Those should work with many types of VHF JFETs and small signal RF NPNs. For parts I'd recommend something like 2N5109 or MPSH10 for the NPN, and a J300 with around 15-25 mA Idss.
Ed
toggle quoted message
Show quoted text
--- In TekScopes@..., David <davidwhess@...> wrote:
On Wed, 02 Jan 2013 17:55:38 -0000, "Ed Breya" <edbreya@...>
wrote:
It looks like the required RF power levels aren't out of reach for JFET substitution of the 7587 tubes. The tricky ones will be the 500 MHz multiplier and the 100 MHz driver for the 200 MHz diode doubler. Coincidently those are the stages that have been giving me trouble. I
have to break out one of my faster oscilloscopes to see what is going
on with them. It is possible that the 500 MHz multiplier just needs
tuning but so far the output from the 200 MHz diode doubler has been
low and not real symmetrical. Swapping V40 and V30 made no difference
but maybe both are weak.
The diodes in the doubler do not test as obviously bad in circuit but
I may try and find higher performance replacements.
None of the tubes use any cathode degeneration, and are only biased by grid current through the grounded 100k resistors, so dropping JFETs in their place means that they'll just run at Idss. For the 500 and 100/200 MHz circuits, the more, the better - in this case, high gain and nonlinearity are your friends. If you get a bunch of JFETs like U300, with Idss near the top of the range (>40 mA), they should be OK in those spots. If you come up short on power, you can try paralleling two devices with the same Idss for each spot. The other three tube spots have more modest power requirements, so lighter JFETs can be used there. For drain supplies, the +12V is the only other positive supply already available, which should work OK in place of the original +125V plate supplies. I guess I should go through and measure the operating points via the
grid voltages and maybe plate currents to get a better idea of what is
going on.
For the drain supplies, I was thinking of just adding a low quiescent
current series regulator to the 125 volt plate supply and dropping it
directly to a voltage compatible with JFET operation. Normally I
would eschew such a wasteful configuration but the original design
does essentially the same thing via the plate resistance and quiescent
current of each stage anyway which may be why the design only enables
each stage as needed.
One thing that will complicate matters is the disabling of certain sections when needed. The 100, 50, and 20 MHz amplifiers appear to be cut off when deselected, by grounding their screen grids. You would have to replicate this action in the new circuitry if you want to keep it operating the same as original. I noticed that every stage is disabled when not in use either via the
screen grid or in the case of the 500 MHz multiplier, both the screen
grid and anode supply. That is one of the reasons I wanted to try
using two JFETs or a JFET and bipolar per stage in a cascode
configuration which would allow the screen grid circuit to disable
each stage as designed.
Cascodes would also offer much higher performance at least as far as
gain at high frequencies. The higher needed voltage is obviously not
a problem in this case.
|
Last time the subject of Nuvistor replacement with JFETs came up I searched the Supertex page for the depletion mode MOSFETs they make and did not find anything. I see now though that they are still in production and available although you sure would not know that from Supertex. I think the LND150 will have too small a current rating and the next larger device they make has too much capacitance. On Wed, 02 Jan 2013 17:30:51 -0000, "Jamie" <jamietyson@...> wrote: Dunno if it'll be helpful or not but I'd had good results with LND150 depletion mode mosfets in various lower frequency tube circuits. They act a lot like a pentode in-circuit, with no screen connection, of course. Good to 450 volts.
Jamie
--- In TekScopes@..., David <davidwhess@...> wrote:
I have a mostly functional Tektronix 184 Time-Mark Generator that I am
restoring. It has six 7587 nuvistor tetrodes and they all appear to
work except for maybe the pair in the 500 MHz multiplier.
You can still pick up used but probably good and NOS 7587 nuvistors
for reasonable prices but how feasible would it be to replace them
with say JFETs with cascodes?
I will probably try it no matter what is said here just for the fun of
it but any advice would be appreciated.
|
The heaters are always on though so I am not sure how much difference having the tubes in cutoff most of the time would make. I have a full set of characteristic curves which show just above 20mA at Vg1=0 volts and the data I have about biasing suggests that grid biasing will yield about 10mA but nothing matches the precise conditions Tektronix used so I will just have to measure it. Thinking about a dual JFET cascode gave me a headache. On one hand, the top JFET cannot be in cutoff without negative bias and on the other hand, the bottom JFET is hardly going to conduct with a Vds of 0 volts. As Vds on the bottom JFET increases, the top JFET will have negative bias. At worst I think this is one of those cases where even though the current will not drop to zero, it will be close enough for engineering purposes. Calculating it is pretty straightforward but it depends too much on the individual transistors. Using an NPN on top neatly solves the problem and is what I originally had in mind although I will test it both ways. I will have to be careful about the Vce on the NPN since RF transistors tend to have relatively low ratings. That will ultimately limit the supply voltage. On the other hand, the cascode configuration relaxes the transistor performance requirements and allows them to operate in the best case configuration. On Wed, 02 Jan 2013 22:06:30 -0000, "Ed Breya" <edbreya@...> wrote: You may want to try swapping tubes from the other spots too, although it seems they're all biased for the same operating conditions, so should wear at the same rate - but if they are only on when needed for their specific function, then it all depends on the history of use. I think the only one that's always on then, would be the 10 MHz oscillator, so it should have the most wear.
I looked up the 7587 and see that it's rated for 20 mA max cathode current, but there were no charts showing what it should be near zero grid bias - unless the grid current gets pretty high, it doesn't seem that it will make much into 100k. It would be interesting to see what it is if you measure it and the actual plate currents.
I think your cascode version with a JFET and an NPN would be best, since you can then assure cutoff by pulling the base to ground. The double-JFET form won't positively cut off unless you can pull the gate negative, making it more complicated. You will have to watch out for the voltage ratings of the NPN to make sure it can handle the cutoff conditions.
I'd recommend around +15V or so for the bases (via small R so they won't oscillate on their own), and about +25V for the collector supply. Those should work with many types of VHF JFETs and small signal RF NPNs. For parts I'd recommend something like 2N5109 or MPSH10 for the NPN, and a J300 with around 15-25 mA Idss.
Ed
|
It just occurred to me that adding a couple of diodes or a low voltage zener in series with the source of the top JFET would also assure cutoff. I will test using a 2N7000 as well since I keep those in my parts drawer. On Wed, 02 Jan 2013 22:06:30 -0000, "Ed Breya" <edbreya@...> wrote: . . .
I think your cascode version with a JFET and an NPN would be best, since you can then assure cutoff by pulling the base to ground. The double-JFET form won't positively cut off unless you can pull the gate negative, making it more complicated. You will have to watch out for the voltage ratings of the NPN to make sure it can handle the cutoff conditions.
. . .
|
I added a 10 ohm current shunt to the anode circuit of V20, the 20 MHz amplifier, and used it to test all of the tubes. The anode currents varied from 4.8 to 5.7 milliamps, the gate voltages varied from -10.3 to -12.6 volts, and the screen grid voltages varied from 72.2 to 76.0 volts except for one 7587 which had a screen voltage of 63.5 volts for some reason. That one also had the highest anode current but not the highest grid voltage. I don't know if that is significant. I can get 2N5950 JFETs and MPSH11 NPNs cheap at the moment so I will start off with them. On Wed, 02 Jan 2013 22:06:30 -0000, "Ed Breya" <edbreya@...> wrote: You may want to try swapping tubes from the other spots too, although it seems they're all biased for the same operating conditions, so should wear at the same rate - but if they are only on when needed for their specific function, then it all depends on the history of use. I think the only one that's always on then, would be the 10 MHz oscillator, so it should have the most wear.
I looked up the 7587 and see that it's rated for 20 mA max cathode current, but there were no charts showing what it should be near zero grid bias - unless the grid current gets pretty high, it doesn't seem that it will make much into 100k. It would be interesting to see what it is if you measure it and the actual plate currents.
I think your cascode version with a JFET and an NPN would be best, since you can then assure cutoff by pulling the base to ground. The double-JFET form won't positively cut off unless you can pull the gate negative, making it more complicated. You will have to watch out for the voltage ratings of the NPN to make sure it can handle the cutoff conditions.
I'd recommend around +15V or so for the bases (via small R so they won't oscillate on their own), and about +25V for the collector supply. Those should work with many types of VHF JFETs and small signal RF NPNs. For parts I'd recommend something like 2N5109 or MPSH10 for the NPN, and a J300 with around 15-25 mA Idss.
Ed
--- In TekScopes@..., David <davidwhess@...> wrote:
On Wed, 02 Jan 2013 17:55:38 -0000, "Ed Breya" <edbreya@...>
wrote:
It looks like the required RF power levels aren't out of reach for JFET substitution of the 7587 tubes. The tricky ones will be the 500 MHz multiplier and the 100 MHz driver for the 200 MHz diode doubler. Coincidently those are the stages that have been giving me trouble. I
have to break out one of my faster oscilloscopes to see what is going
on with them. It is possible that the 500 MHz multiplier just needs
tuning but so far the output from the 200 MHz diode doubler has been
low and not real symmetrical. Swapping V40 and V30 made no difference
but maybe both are weak.
The diodes in the doubler do not test as obviously bad in circuit but
I may try and find higher performance replacements.
None of the tubes use any cathode degeneration, and are only biased by grid current through the grounded 100k resistors, so dropping JFETs in their place means that they'll just run at Idss. For the 500 and 100/200 MHz circuits, the more, the better - in this case, high gain and nonlinearity are your friends. If you get a bunch of JFETs like U300, with Idss near the top of the range (>40 mA), they should be OK in those spots. If you come up short on power, you can try paralleling two devices with the same Idss for each spot. The other three tube spots have more modest power requirements, so lighter JFETs can be used there. For drain supplies, the +12V is the only other positive supply already available, which should work OK in place of the original +125V plate supplies. I guess I should go through and measure the operating points via the
grid voltages and maybe plate currents to get a better idea of what is
going on.
For the drain supplies, I was thinking of just adding a low quiescent
current series regulator to the 125 volt plate supply and dropping it
directly to a voltage compatible with JFET operation. Normally I
would eschew such a wasteful configuration but the original design
does essentially the same thing via the plate resistance and quiescent
current of each stage anyway which may be why the design only enables
each stage as needed.
One thing that will complicate matters is the disabling of certain sections when needed. The 100, 50, and 20 MHz amplifiers appear to be cut off when deselected, by grounding their screen grids. You would have to replicate this action in the new circuitry if you want to keep it operating the same as original. I noticed that every stage is disabled when not in use either via the
screen grid or in the case of the 500 MHz multiplier, both the screen
grid and anode supply. That is one of the reasons I wanted to try
using two JFETs or a JFET and bipolar per stage in a cascode
configuration which would allow the screen grid circuit to disable
each stage as designed.
Cascodes would also offer much higher performance at least as far as
gain at high frequencies. The higher needed voltage is obviously not
a problem in this case.
|
Wow - that means the grid currents must be over -100 uA to get that much on 100k grid leak resistors. I don't recall what the screens were hooked to, but I think you should calculate the screen currents too, to see the total cathode currents.
If they only used 5 mA or so plate current, you can easily replicate that with modest JFET and cascode NPN current.
Ed
toggle quoted message
Show quoted text
--- In TekScopes@..., David wrote:
I added a 10 ohm current shunt to the anode circuit of V20, the 20 MHz
amplifier, and used it to test all of the tubes. The anode currents
varied from 4.8 to 5.7 milliamps, the gate voltages varied from -10.3
to -12.6 volts, and the screen grid voltages varied from 72.2 to 76.0
volts except for one 7587 which had a screen voltage of 63.5 volts for
some reason.
That one also had the highest anode current but not the highest grid
voltage. I don't know if that is significant.
I can get 2N5950 JFETs and MPSH11 NPNs cheap at the moment so I will
start off with them.
On Wed, 02 Jan 2013 22:06:30 -0000, "Ed Breya"
wrote:
You may want to try swapping tubes from the other spots too, although it seems they're all biased for the same operating conditions, so should wear at the same rate - but if they are only on when needed for their specific function, then it all depends on the history of use. I think the only one that's always on then, would be the 10 MHz oscillator, so it should have the most wear.
I looked up the 7587 and see that it's rated for 20 mA max cathode current, but there were no charts showing what it should be near zero grid bias - unless the grid current gets pretty high, it doesn't seem that it will make much into 100k. It would be interesting to see what it is if you measure it and the actual plate currents.
I think your cascode version with a JFET and an NPN would be best, since you can then assure cutoff by pulling the base to ground. The double-JFET form won't positively cut off unless you can pull the gate negative, making it more complicated. You will have to watch out for the voltage ratings of the NPN to make sure it can handle the cutoff conditions.
I'd recommend around +15V or so for the bases (via small R so they won't oscillate on their own), and about +25V for the collector supply. Those should work with many types of VHF JFETs and small signal RF NPNs. For parts I'd recommend something like 2N5109 or MPSH10 for the NPN, and a J300 with around 15-25 mA Idss.
Ed
--- In TekScopes@..., David wrote:
On Wed, 02 Jan 2013 17:55:38 -0000, "Ed Breya"
wrote:
It looks like the required RF power levels aren't out of reach for JFET substitution of the 7587 tubes. The tricky ones will be the 500 MHz multiplier and the 100 MHz driver for the 200 MHz diode doubler. Coincidently those are the stages that have been giving me trouble. I
have to break out one of my faster oscilloscopes to see what is going
on with them. It is possible that the 500 MHz multiplier just needs
tuning but so far the output from the 200 MHz diode doubler has been
low and not real symmetrical. Swapping V40 and V30 made no difference
but maybe both are weak.
The diodes in the doubler do not test as obviously bad in circuit but
I may try and find higher performance replacements.
None of the tubes use any cathode degeneration, and are only biased by grid current through the grounded 100k resistors, so dropping JFETs in their place means that they'll just run at Idss. For the 500 and 100/200 MHz circuits, the more, the better - in this case, high gain and nonlinearity are your friends. If you get a bunch of JFETs like U300, with Idss near the top of the range (>40 mA), they should be OK in those spots. If you come up short on power, you can try paralleling two devices with the same Idss for each spot. The other three tube spots have more modest power requirements, so lighter JFETs can be used there. For drain supplies, the +12V is the only other positive supply already available, which should work OK in place of the original +125V plate supplies. I guess I should go through and measure the operating points via the
grid voltages and maybe plate currents to get a better idea of what is
going on.
For the drain supplies, I was thinking of just adding a low quiescent
current series regulator to the 125 volt plate supply and dropping it
directly to a voltage compatible with JFET operation. Normally I
would eschew such a wasteful configuration but the original design
does essentially the same thing via the plate resistance and quiescent
current of each stage anyway which may be why the design only enables
each stage as needed.
One thing that will complicate matters is the disabling of certain sections when needed. The 100, 50, and 20 MHz amplifiers appear to be cut off when deselected, by grounding their screen grids. You would have to replicate this action in the new circuitry if you want to keep it operating the same as original. I noticed that every stage is disabled when not in use either via the
screen grid or in the case of the 500 MHz multiplier, both the screen
grid and anode supply. That is one of the reasons I wanted to try
using two JFETs or a JFET and bipolar per stage in a cascode
configuration which would allow the screen grid circuit to disable
each stage as designed.
Cascodes would also offer much higher performance at least as far as
gain at high frequencies. The higher needed voltage is obviously not
a problem in this case.
|
What I think is happening is the control grid circuit is rectifying the positive input peaks to set the bias. A JFET replacement would operate in the same way although I may have to change the grid bias resistor or in the worst case, add a paralleled resistor and capacitor for source biasing. Vacuum tube design is before my time so I am just having fun reverse engineering. The time mark outputs which work fine do most of what I need. The screen grid series resistors are 27K and connect to the 125 volt anode supply so the currents are all about 1.8 milliamps which is well under the 3.6 milliamp maximum. That makes the total cathode current less than half of the 20 milliamp maximum at least on the 20 MHz amplifier. The 50 Mhz and 100 MHz amplifiers operate with self biased grid voltages closer to ground so their anode current should be higher. On later units like mine, the screen grid voltage of oscillator V10 is fixed by a zener diode at about 52 volts. None of the carbon composition resistors I checked were out of tolerance. Of course, everything up to the 100 MHz output *does* work. The 200 MHz doubler at least sort of works but the output looks low and distorted to me. I may try replacing the germanium diodes with 1N270s or RF schottky diodes. The only information I have on the 152-0075-00 diodes Tektronix used is from the Tektronix parts book. On Sat, 19 Jan 2013 18:44:49 -0000, "Ed Breya" <edbreya@...> wrote: Wow - that means the grid currents must be over -100 uA to get that much on 100k grid leak resistors. I don't recall what the screens were hooked to, but I think you should calculate the screen currents too, to see the total cathode currents.
If they only used 5 mA or so plate current, you can easily replicate that with modest JFET and cascode NPN current.
Ed
--- In TekScopes@..., David wrote:
I added a 10 ohm current shunt to the anode circuit of V20, the 20 MHz
amplifier, and used it to test all of the tubes. The anode currents
varied from 4.8 to 5.7 milliamps, the gate voltages varied from -10.3
to -12.6 volts, and the screen grid voltages varied from 72.2 to 76.0
volts except for one 7587 which had a screen voltage of 63.5 volts for
some reason.
That one also had the highest anode current but not the highest grid
voltage. I don't know if that is significant.
I can get 2N5950 JFETs and MPSH11 NPNs cheap at the moment so I will
start off with them.
On Wed, 02 Jan 2013 22:06:30 -0000, "Ed Breya"
wrote:
You may want to try swapping tubes from the other spots too, although it seems they're all biased for the same operating conditions, so should wear at the same rate - but if they are only on when needed for their specific function, then it all depends on the history of use. I think the only one that's always on then, would be the 10 MHz oscillator, so it should have the most wear.
I looked up the 7587 and see that it's rated for 20 mA max cathode current, but there were no charts showing what it should be near zero grid bias - unless the grid current gets pretty high, it doesn't seem that it will make much into 100k. It would be interesting to see what it is if you measure it and the actual plate currents.
I think your cascode version with a JFET and an NPN would be best, since you can then assure cutoff by pulling the base to ground. The double-JFET form won't positively cut off unless you can pull the gate negative, making it more complicated. You will have to watch out for the voltage ratings of the NPN to make sure it can handle the cutoff conditions.
I'd recommend around +15V or so for the bases (via small R so they won't oscillate on their own), and about +25V for the collector supply. Those should work with many types of VHF JFETs and small signal RF NPNs. For parts I'd recommend something like 2N5109 or MPSH10 for the NPN, and a J300 with around 15-25 mA Idss.
Ed
|
I just stumbled onto this blast from the past while looking for other info, and immediately remembered this discussion. It looks like Teledyne had the direct substitution problem solved in the 1970s. Hope this helps with cascode ideas - I don't know if the high voltage JFETs are available anymore.
)_Misc_Test_Equipment/FETRON_Solid_State_Vacuum_Tube_Replacement.pdf
Ed
toggle quoted message
Show quoted text
--- In TekScopes@..., David wrote:
What I think is happening is the control grid circuit is rectifying
the positive input peaks to set the bias. A JFET replacement would
operate in the same way although I may have to change the grid bias
resistor or in the worst case, add a paralleled resistor and capacitor
for source biasing.
Vacuum tube design is before my time so I am just having fun reverse
engineering. The time mark outputs which work fine do most of what I
need.
The screen grid series resistors are 27K and connect to the 125 volt
anode supply so the currents are all about 1.8 milliamps which is well
under the 3.6 milliamp maximum. That makes the total cathode current
less than half of the 20 milliamp maximum at least on the 20 MHz
amplifier. The 50 Mhz and 100 MHz amplifiers operate with self biased
grid voltages closer to ground so their anode current should be
higher. On later units like mine, the screen grid voltage of
oscillator V10 is fixed by a zener diode at about 52 volts.
None of the carbon composition resistors I checked were out of
tolerance. Of course, everything up to the 100 MHz output *does*
work.
The 200 MHz doubler at least sort of works but the output looks low
and distorted to me. I may try replacing the germanium diodes with
1N270s or RF schottky diodes. The only information I have on the
152-0075-00 diodes Tektronix used is from the Tektronix parts book.
On Sat, 19 Jan 2013 18:44:49 -0000, "Ed Breya"
wrote:
Wow - that means the grid currents must be over -100 uA to get that much on 100k grid leak resistors. I don't recall what the screens were hooked to, but I think you should calculate the screen currents too, to see the total cathode currents.
If they only used 5 mA or so plate current, you can easily replicate that with modest JFET and cascode NPN current.
Ed
--- In TekScopes@..., David wrote:
I added a 10 ohm current shunt to the anode circuit of V20, the 20 MHz
amplifier, and used it to test all of the tubes. The anode currents
varied from 4.8 to 5.7 milliamps, the gate voltages varied from -10.3
to -12.6 volts, and the screen grid voltages varied from 72.2 to 76.0
volts except for one 7587 which had a screen voltage of 63.5 volts for
some reason.
That one also had the highest anode current but not the highest grid
voltage. I don't know if that is significant.
I can get 2N5950 JFETs and MPSH11 NPNs cheap at the moment so I will
start off with them.
On Wed, 02 Jan 2013 22:06:30 -0000, "Ed Breya"
wrote:
You may want to try swapping tubes from the other spots too, although it seems they're all biased for the same operating conditions, so should wear at the same rate - but if they are only on when needed for their specific function, then it all depends on the history of use. I think the only one that's always on then, would be the 10 MHz oscillator, so it should have the most wear.
I looked up the 7587 and see that it's rated for 20 mA max cathode current, but there were no charts showing what it should be near zero grid bias - unless the grid current gets pretty high, it doesn't seem that it will make much into 100k. It would be interesting to see what it is if you measure it and the actual plate currents.
I think your cascode version with a JFET and an NPN would be best, since you can then assure cutoff by pulling the base to ground. The double-JFET form won't positively cut off unless you can pull the gate negative, making it more complicated. You will have to watch out for the voltage ratings of the NPN to make sure it can handle the cutoff conditions.
I'd recommend around +15V or so for the bases (via small R so they won't oscillate on their own), and about +25V for the collector supply. Those should work with many types of VHF JFETs and small signal RF NPNs. For parts I'd recommend something like 2N5109 or MPSH10 for the NPN, and a J300 with around 15-25 mA Idss.
Ed
|
Thanks for the link. It makes for interesting reading. At the very end, complete conversion kits are mentioned for specific instruments like the Tektronix 500 series CA amplifier and the Hewlett-Packard HP500 VTVM. Teledyne used the cascode configuration with internal bias since they were replacing triodes. They had their own high voltage JFETs which simplifies the design although high voltage enhancement or depletion mode MOSFETs could be used now. High voltage high frequency bipolars would be useful if they were available. In my case, I can just lower the anode supply voltage. On Sun, 27 Jan 2013 16:47:33 -0000, "Ed Breya" <edbreya@...> wrote: I just stumbled onto this blast from the past while looking for other info, and immediately remembered this discussion. It looks like Teledyne had the direct substitution problem solved in the 1970s. Hope this helps with cascode ideas - I don't know if the high voltage JFETs are available anymore.
)_Misc_Test_Equipment/FETRON_Solid_State_Vacuum_Tube_Replacement.pdf
Ed
|
David Hess