Re: Question about MMIC amp design for HP 8601a output replacement
Going back to school days the term transformer was defined as either a contraption using windings on a iron or ferrite core OR a transmission line transformer using microstrip or striplines usually in the form of a stepped impedance transformer
, usually neither is used in wideband MMIC especially not in the RF range for a variety of reasons including but not limited size and achievable bandwidth !
Wideband MMIC extending into the MHz range or lower are either utilizing feedback elements or are build as distributed amplifiers , the latter can achieve typical bandwidths of 100KHz to 50GHz and output power of 30dBm the only downside
is cost as this topology requires a lot of real estate
?
toggle quoted message
Show quoted text
From: [email protected] < [email protected]>
On Behalf Of Jim Ford via groups.io
Sent: Tuesday, March 22, 2022 5:26 PM
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] Question about MMIC amp design for HP 8601a output replacement
?
Well, Gary did say "without a transformer".? If by transformer we mean a circuit to transform the low output impedance of the amplifier to 50 ohms, then he is correct.? Yes, I have worked with MMIC amplifiers that run on 5 V that put out
greater than +30 damage (=1 watt).
Sent from my T-Mobile 4G LTE Device
-------- Original message --------
Date: 3/22/22 3:12 PM (GMT-08:00)
Subject: Re: [HP-Agilent-Keysight-equipment] Question about MMIC amp design for HP 8601a output replacement
Wrong !
There are several MMICs that can drive 20dBm and higher over the desired frequency range operating below 12V
?
?
My amplifier (as the amplifier by Dasaro Designs) replaces the output hybrid only. The part I used was the BFU590QX, from Digikey.? All three of these amplifiers require 2 stages to replace the output hybrid.
Unlike the Dasaro Designs amplifier, mine is not contained in a shielded enclosure, which allowed me a bit more area. To help remove the heat I have screwed a block of aluminum on the top cover that just contacts the top of the three transistors, with a
thermal pad between the aluminum and transistor case. Not a great solution, but then again we need to draw only about 0.6 watts from each of the two output transistors, about half that from the input transistor. Not a great solution, but it should be adequate
to hold the die temperature down.
+20 dBm is 2.25 Vrms, or 6.32 V peak to peak. You can't get +20 dBm from a +5 V supply without a transformer. A MMIC operating at perhaps less than 12 volts will not be able to provide +20 dBm.
Gary Appel
On 3/22/2022 11:24 AM, DB via groups.io wrote:
Hi Gary,
Thanks for posting your design of the output amp using BFU590Q's.? It looks like the BFU590Q are also becoming hard to obtain as I only saw one supplier had any in stock (BFU590QX only).
I was curious if your design is replacing both the preamp and final output amps inside the A6 module or just the final output amp.? On my 8601, It appears that both of the modules have an issue,so hoping that a cascade set of 2 or 3 MMIC's might work.? I have
a good signal level input into the A6 module.
Thanks,
DB
|
Re: Question about MMIC amp design for HP 8601a output replacement
Actually I found designing wideband circuits always to be fun , its actually not as quite as hard as you think if you got the right design tools
toggle quoted message
Show quoted text
-----Original Message----- From: [email protected] < [email protected]> On Behalf Of Gary Appel via groups.io Sent: Tuesday, March 22, 2022 5:49 PM To: [email protected]Subject: Re: [HP-Agilent-Keysight-equipment] Question about MMIC amp design for HP 8601a output replacement I guess I was making an assumption. This application requires a three decade bandwidth. The amplifier must drive a 50 ohm load, and must be capable of swinging at least 6.32 V peak to peak. This is met by each of these three amplifiers. Life is easier with a narrow bandwidth. Gary Appel On 3/22/2022 12:25 PM, Jim Ford wrote: Well, Gary did say "without a transformer".? If by transformer we mean
a circuit to transform the low output impedance of the amplifier to 50
ohms, then he is correct.? Yes, I have worked with MMIC amplifiers
that run on 5 V that put out greater than +30 damage (=1 watt).
Jim Ford
|
Re: Question about MMIC amp design for HP 8601a output replacement
Thanks Lothar Regarding available physical working space, I measured the A6 output module internals and attached some pics: The total available space within the inside walls of the A6 module is an absolute maximum of 1.125"? x 4.625" (28.58mm x 117.475mm) The maximum available space inside the output amp tin itself is only .75" x 1.125" (19.05mm x 28.58mm) I also included a pic of the meter scaling in case it might be helpful for the detector scaling or characterization.? The 8601 A9 board shows adjustments for the meter scaling? -10db (R181) and 0db (R131) and an ALC balance (R136), but no other ALC scaling adjustment. You may already have these, but thought I would post the links to the MiniCircuits S-Parameters for the MMIC's that I have on hand.? (not sure if these files have the low freq data you need) Thanks DB    
|
Re: HP8444A 1.55GHz oscillator
This is very useful - the oscillator adjusting instructions are rather different and much more detailed - especially the checks based on adjusting the power supply voltage.? Just need to see if I can locate a working thermistor head for a 431 for power measurements!
Alan
|
Re: Question about MMIC amp design for HP 8601a output replacement
I guess I was making an assumption. This application requires a three decade bandwidth. The amplifier must drive a 50 ohm load, and must be capable of swinging at least 6.32 V peak to peak. This is met by each of these three amplifiers.
Life is easier with a narrow bandwidth.
Gary Appel
toggle quoted message
Show quoted text
On 3/22/2022 12:25 PM, Jim Ford wrote: Well, Gary did say "without a transformer".? If by transformer we mean a circuit to transform the low output impedance of the amplifier to 50 ohms, then he is correct.? Yes, I have worked with MMIC amplifiers that run on 5 V that put out greater than +30 damage (=1 watt).
Jim Ford
|
Re: Question about MMIC amp design for HP 8601a output replacement
Well, Gary did say "without a transformer".? If by transformer we mean a circuit to transform the low output impedance of the amplifier to 50 ohms, then he is correct.? Yes, I have worked with MMIC amplifiers that run on 5 V that put out greater than +30 damage (=1 watt).
Jim Ford
Sent from my T-Mobile 4G LTE Device
toggle quoted message
Show quoted text
-------- Original message -------- From: Lothar baier <Lothar@...> Date: 3/22/22 3:12 PM (GMT-08:00) Subject: Re: [HP-Agilent-Keysight-equipment] Question about MMIC amp design for HP 8601a output replacement
Wrong !
There are several MMICs that can drive 20dBm and higher over the desired frequency range operating below 12V
?
?
My amplifier (as the amplifier by Dasaro Designs) replaces the output hybrid only. The part I used was the BFU590QX, from Digikey.? All three of these amplifiers require 2 stages to replace the output hybrid.
Unlike the Dasaro Designs amplifier, mine is not contained in a shielded enclosure, which allowed me a bit more area. To help remove the heat I have screwed a block of aluminum on the top cover that just contacts the top of the three transistors, with a
thermal pad between the aluminum and transistor case. Not a great solution, but then again we need to draw only about 0.6 watts from each of the two output transistors, about half that from the input transistor. Not a great solution, but it should be adequate
to hold the die temperature down.
+20 dBm is 2.25 Vrms, or 6.32 V peak to peak. You can't get +20 dBm from a +5 V supply without a transformer. A MMIC operating at perhaps less than 12 volts will not be able to provide +20 dBm.
Gary Appel
On 3/22/2022 11:24 AM, DB via groups.io wrote:
Hi Gary,
Thanks for posting your design of the output amp using BFU590Q's.? It looks like the BFU590Q are also becoming hard to obtain as I only saw one supplier had any in stock (BFU590QX only).
I was curious if your design is replacing both the preamp and final output amps inside the A6 module or just the final output amp.? On my 8601, It appears that both of the modules have an issue,so hoping that a cascade set of 2 or 3 MMIC's might work.? I have
a good signal level input into the A6 module.
Thanks,
DB
|
Re: Question about MMIC amp design for HP 8601a output replacement
Wrong !
There are several MMICs that can drive 20dBm and higher over the desired frequency range operating below 12V
?
toggle quoted message
Show quoted text
From: [email protected] < [email protected]>
On Behalf Of Gary Appel via groups.io
Sent: Tuesday, March 22, 2022 5:02 PM
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] Question about MMIC amp design for HP 8601a output replacement
?
My amplifier (as the amplifier by Dasaro Designs) replaces the output hybrid only. The part I used was the BFU590QX, from Digikey.? All three of these amplifiers require 2 stages to replace the output hybrid.
Unlike the Dasaro Designs amplifier, mine is not contained in a shielded enclosure, which allowed me a bit more area. To help remove the heat I have screwed a block of aluminum on the top cover that just contacts the top of the three transistors, with a
thermal pad between the aluminum and transistor case. Not a great solution, but then again we need to draw only about 0.6 watts from each of the two output transistors, about half that from the input transistor. Not a great solution, but it should be adequate
to hold the die temperature down.
+20 dBm is 2.25 Vrms, or 6.32 V peak to peak. You can't get +20 dBm from a +5 V supply without a transformer. A MMIC operating at perhaps less than 12 volts will not be able to provide +20 dBm.
Gary Appel
On 3/22/2022 11:24 AM, DB via groups.io wrote:
Hi Gary,
Thanks for posting your design of the output amp using BFU590Q's.? It looks like the BFU590Q are also becoming hard to obtain as I only saw one supplier had any in stock (BFU590QX only).
I was curious if your design is replacing both the preamp and final output amps inside the A6 module or just the final output amp.? On my 8601, It appears that both of the modules have an issue,so hoping that a cascade set of 2 or 3 MMIC's might work.? I have
a good signal level input into the A6 module.
Thanks,
DB
|
Re: Question about MMIC amp design for HP 8601a output replacement
My amplifier (as the amplifier by Dasaro Designs) replaces the
output hybrid only. The part I used was the BFU590QX, from
Digikey.? All three of these amplifiers require 2 stages to
replace the output hybrid.
Unlike the Dasaro Designs amplifier, mine is not contained in a
shielded enclosure, which allowed me a bit more area. To help
remove the heat I have screwed a block of aluminum on the top
cover that just contacts the top of the three transistors, with a
thermal pad between the aluminum and transistor case. Not a great
solution, but then again we need to draw only about 0.6 watts from
each of the two output transistors, about half that from the input
transistor. Not a great solution, but it should be adequate to
hold the die temperature down.
+20 dBm is 2.25 Vrms, or 6.32 V peak to peak. You can't get +20
dBm from a +5 V supply without a transformer. A MMIC operating at
perhaps less than 12 volts will not be able to provide +20 dBm.
Gary Appel
On 3/22/2022 11:24 AM, DB via groups.io
wrote:
toggle quoted message
Show quoted text
Hi Gary,
Thanks for posting your design of the output amp using BFU590Q's.?
It looks like the BFU590Q are also becoming hard to obtain as I
only saw one supplier had any in stock (BFU590QX only).
I was curious if your design is replacing both the preamp and
final output amps inside the A6 module or just the final output
amp.? On my 8601, It appears that both of the modules have an
issue,so hoping that a cascade set of 2 or 3 MMIC's might work.? I
have a good signal level input into the A6 module.
Thanks,
DB
|
Re: Question about MMIC amp design for HP 8601a output replacement
Hi Gary,
Thanks for posting your design of the output amp using BFU590Q's.? It looks like the BFU590Q are also becoming hard to obtain as I only saw one supplier had any in stock (BFU590QX only).
I was curious if your design is replacing both the preamp and final output amps inside the A6 module or just the final output amp.? On my 8601, It appears that both of the modules have an issue,so hoping that a cascade set of 2 or 3 MMIC's might work.? I have a good signal level input into the A6 module.
Thanks,
DB
|
Re: HP 54522A - risetime measurement vs. graticule readings
Off the top of my head, I would say that the algorithm the scope uses to measure rise-time is geared towards something that is not too far from a square wave. In particular, it needs to figure out the baseline (bottom) and top voltages that the 10% and 90% values are relative to. The saw-tooth waveform confuses this algorithm as it fails to provide any clues as to what the baseline and top voltage are.
Note that these baseline and top voltages are not the minimum and maximum voltages of the waveform since doing so would cause the algorithm to give wrong rise-time measurements for square waves that have any appreciable overshoot or undershoot, which is usually the case.
|
Re: HP 180 series scope CRTs
Perhaps you might be able to "shoehorn in" an LCD module to replace the CRT in the HP 182 ? There was a Blog showing this done for the HP 141 SA, likely different, but provides an idea: LCD module for the HP 141:
?
?
?
|
Re: HP YIG replacement 3-7GHz
Usually the older YIGs used a common base configuration, the coupling loop is placed in the emitter path with the other side either connected to ground or to a neg voltage supply , the base is connected to a small inductance ( bondwire
loop ) that is carefully selected to ensure oscillation over the desired frequency range , the collector is the RF output.
As you use a different device the inductance value originally used may not be sufficient to ensure reliable oscillation and might need to be adjusted , up to a certain extend you can fiddle with the bias if the toplogy allows it but its
not a given it may work
?
toggle quoted message
Show quoted text
From: [email protected] < [email protected]>
On Behalf Of Peter Hansen via groups.io
Sent: Tuesday, March 22, 2022 1:12 PM
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] HP YIG replacement 3-7GHz
?
Hello Lothar the device is AT41000 or AT42000 of course obsolete. I am sure if we could find a reliable source it maybe would get CT1DMK to use his bonder for repairs again. I know he is out of
parts but has repaired them in the past.
I love impossible challenges. People always told me the Rythm tower from HP8563E would be impossible to adjust. But none of the less I can do that on daily basis if needed. It was a long road but
now possible. In this case a Scalar netwoek analyzers 26.5Ghz comes in handy as the detektor is broadband so i look at the downconverted IF signal. I use that for adjustments at all frequencies
The first challenge is to figure out the transistor they used , HP had their own foundry and not all parts they made were sold as products on the open market.
Given the frequency range of the yig it was likely to be a HBT rather than a GaAs Fet but HP semiconductor became Avago and then broadcom and to my knowledge they don¡¯t sell die level parts!
The next challenge then is to remove the old transistor and mount a new one, usually devices are mounted using eutectic solder preforms , the stage is heated until the solder melts and the part is then ¡°scrubbed ¡° in as gold from the back
metalization diffuses into the solder the melting point increases and the solder solidifies , in order to remove the part you would have to heat up the solder past the new melting point without imposing thermal stress on the other parts used or dislocating
them !
To mount a new part you need a die attach as well as preforms, you could of course use conductive epoxy but the heat transfer generally is not as good as with eutectic die attach so you might run into issues with reliability!
As far as wirebonders concerns there are plenty on ebay , expect to pay between $5000 and $16000 for a refurbished one , for this type of work you need a wedge/wire bonder , if you plan on rebuilding modules plan on getting a deep access
bonder or a convertible machine that can be used for different processes .
The biggest challenge with bonding is to figure out the right settings though, usually in a production environment the supplier of the semiconductor devices used provides bond samples which in essence are parts that are not working ok those
samples are then bonded in a dummy circuit and a pull tester is used to check the strength of the bond?
Also make sure you have someone with a steady hand and young eyes to threat the bonder
? and also invest in a vibration isolation table or platform to set your bonder on?
?
On Mar 22, 2022, at 02:28, Peter Hansen via groups.io <oz1lpr@...> wrote:
?
I have succeded to integrate a YO84 YIG faily simple and you can use +/-15 for 30V for heater it is ok and shunting the Sense resistors in the Maincoil driver means it can span the 3-7Ghz. I have
made changes to the Loop so it is yield low Phase noise. Output from the YIG it self is good also. Mechanical implementation is done by using the original mechanic shell and make a hole for the YIG. It can be jammed in there when I mount the YIG in the unit.
A PCB "still prototype" is mounted with a 10pin pinheader so the orignal cable can be used. BUT BUT now the challenge. The original YIG does not get hot due to the sensitve coils meaning less drift. If the Loop is opened by placing the jumper in the TEST position
I observe a drift of the YIG of around 5Mhz from cold to warm. The Avantek YIG drift around 30-40Mhz. The Loop is build to compensate for 10Mhz drift. This means you can either have a unlock condition ERR301 during warm up or you will get it during use.
Playing around in the Loop is not a things for the average guy to do. The Frequency Lock? circuit is quite hard to work with. Changes to the loop will also challenge your phasenoise performance.
I am sure this is why HP deviated from the standard sensitivity on the maincoil to keep drift down. The avantek YIG gets hot the HP ones do not.
More fruitfull would be to find a replacement for the Bonded Transistor and get someone with a Wirebonder to service them.
I do currently not have a wirebonder but as I am repairing alot of advanced HP equipment I am considering it. But where to source components for tryouts.
best regards Peter OZ1LPR
Hi Peter, Et al,
Sorry for the delayed response to all the positive comments. I have been away camping, soaking up what is left of the good weather in autumn, trying my new HF radio set up.
No doubt, there will be many challenges ahead, but I am determined to find a solution. I am waiting on a mail from Micro Lambda to see if they have ever made a replacement, the chap is going to ask the question for me, they have been so helpful to date. There
are other Micro Lambda models to suit Tek and R&S, so they have made replacement models for branded SA and Sign Gens. No harm asking.?
Peter, I think the original HP data sheet is right, this does represent a challenge at? or around 40MHz/mA, the hardware and software is tightly integrated, thus a YIG with similar spec or the development of an interface driver is needed. I haven't thought
that far ahead yet.
Like yourself, and I am sure others on the forum have purchased a YIG in the past to adapt, I purchased a Avantek Y084-1215 about 10 years ago, I recently tried to adapt via a crude interface driver, but this did not work out to well. Although the Avantek YIG
works extremely well for it's vintage, it's main coil spec is way out at ~22.4 MHZ/mA. In addition, getting the juice to run the heater from an internal supply is a challenge.
I am prepared to lay out the cash up to a point to develop a solution.
I hope I do not have to eat my words in time.
Stay tuned.
Regards
Gerald
VK3GM
?
|
Re: HP YIG replacement 3-7GHz
Repairing things is one thing, repairing them so they work reliably over a extended period of time another one .
I am not aware of any company that makes SiGe HBTs offering die level products, one way to go is to find a foundry that offers shared wafer runs and put just a HBT cell with bond pads on a wafer, usually this will get you a bunch of devices with charges
around $1000 but its not guaranteed they will work as a replacement.
Either you would have to get their PDK and then run IV as well as RF simulation on the cell and then run the same tests on the original devices but that in turn requires models and i don¡¯t know if any information outside of S-parameter files for those
dinosaurs exist!
toggle quoted message
Show quoted text
On Mar 22, 2022, at 13:11, Peter Hansen via groups.io <oz1lpr@...> wrote:
?
Hello Lothar the device is AT41000 or AT42000 of course obsolete. I am sure if we could find a reliable source it maybe would get CT1DMK to use his bonder for repairs again. I know he is out of parts but has repaired them in the past.
I love impossible challenges. People always told me the Rythm tower from HP8563E would be impossible to adjust. But none of the less I can do that on daily basis if needed. It was a long road but now possible. In this case a Scalar netwoek analyzers 26.5Ghz
comes in handy as the detektor is broadband so i look at the downconverted IF signal. I use that for adjustments at all frequencies
73 Peter OZ1LPR
The first challenge is to figure out the transistor they used , HP had their own foundry and not all parts they made were sold as products on the open market.
Given the frequency range of the yig it was likely to be a HBT rather than a GaAs Fet but HP semiconductor became Avago and then broadcom and to my knowledge they don¡¯t sell die level parts!
The next challenge then is to remove the old transistor and mount a new one, usually devices are mounted using eutectic solder preforms , the stage is heated until the solder melts and the part is then ¡°scrubbed ¡° in as gold from the back metalization
diffuses into the solder the melting point increases and the solder solidifies , in order to remove the part you would have to heat up the solder past the new melting point without imposing thermal stress on the other parts used or dislocating them !
To mount a new part you need a die attach as well as preforms, you could of course use conductive epoxy but the heat transfer generally is not as good as with eutectic die attach so you might run into issues with reliability!
As far as wirebonders concerns there are plenty on ebay , expect to pay between $5000 and $16000 for a refurbished one , for this type of work you need a wedge/wire bonder , if you plan on rebuilding modules plan on getting a deep access bonder or a convertible
machine that can be used for different processes .
The biggest challenge with bonding is to figure out the right settings though, usually in a production environment the supplier of the semiconductor devices used provides bond samples which in essence are parts that are not working ok those samples are
then bonded in a dummy circuit and a pull tester is used to check the strength of the bond?
Also make sure you have someone with a steady hand and young eyes to threat the bonder ? and also invest in a vibration isolation table or platform to set your bonder on?
On Mar 22, 2022, at 02:28, Peter Hansen via groups.io <oz1lpr@...> wrote:
?
I have succeded to integrate a YO84 YIG faily simple and you can use +/-15 for 30V for heater it is ok and shunting the Sense resistors in the Maincoil driver means it can span the 3-7Ghz. I have made changes to the Loop so it is yield low Phase noise. Output
from the YIG it self is good also. Mechanical implementation is done by using the original mechanic shell and make a hole for the YIG. It can be jammed in there when I mount the YIG in the unit. A PCB "still prototype" is mounted with a 10pin pinheader so
the orignal cable can be used. BUT BUT now the challenge. The original YIG does not get hot due to the sensitve coils meaning less drift. If the Loop is opened by placing the jumper in the TEST position I observe a drift of the YIG of around 5Mhz from cold
to warm. The Avantek YIG drift around 30-40Mhz. The Loop is build to compensate for 10Mhz drift. This means you can either have a unlock condition ERR301 during warm up or you will get it during use.
Playing around in the Loop is not a things for the average guy to do. The Frequency Lock? circuit is quite hard to work with. Changes to the loop will also challenge your phasenoise performance. I am sure this is why HP deviated from the standard sensitivity
on the maincoil to keep drift down. The avantek YIG gets hot the HP ones do not.
More fruitfull would be to find a replacement for the Bonded Transistor and get someone with a Wirebonder to service them.
I do currently not have a wirebonder but as I am repairing alot of advanced HP equipment I am considering it. But where to source components for tryouts.
best regards Peter OZ1LPR
Hi Peter, Et al,
Sorry for the delayed response to all the positive comments. I have been away camping, soaking up what is left of the good weather in autumn, trying my new HF radio set up.
No doubt, there will be many challenges ahead, but I am determined to find a solution. I am waiting on a mail from Micro Lambda to see if they have ever made a replacement, the chap is going to ask the question for me, they have been so helpful to date. There
are other Micro Lambda models to suit Tek and R&S, so they have made replacement models for branded SA and Sign Gens. No harm asking.?
Peter, I think the original HP data sheet is right, this does represent a challenge at? or around 40MHz/mA, the hardware and software is tightly integrated, thus a YIG with similar spec or the development of an interface driver is needed. I haven't thought
that far ahead yet.
Like yourself, and I am sure others on the forum have purchased a YIG in the past to adapt, I purchased a Avantek Y084-1215 about 10 years ago, I recently tried to adapt via a crude interface driver, but this did not work out to well. Although the Avantek YIG
works extremely well for it's vintage, it's main coil spec is way out at ~22.4 MHZ/mA. In addition, getting the juice to run the heater from an internal supply is a challenge.
I am prepared to lay out the cash up to a point to develop a solution.
I hope I do not have to eat my words in time.
Stay tuned.
Regards
Gerald
VK3GM
?
|
Re: HP YIG replacement 3-7GHz
Hello Lothar the device is AT41000 or AT42000 of course obsolete. I am sure if we could find a reliable source it maybe would get CT1DMK to use his bonder for repairs again. I know he is out of parts but has repaired them in the past.
I love impossible challenges. People always told me the Rythm tower from HP8563E would be impossible to adjust. But none of the less I can do that on daily basis if needed. It was a long road but now possible. In this case a Scalar netwoek analyzers 26.5Ghz
comes in handy as the detektor is broadband so i look at the downconverted IF signal. I use that for adjustments at all frequencies
73 Peter OZ1LPR
The first challenge is to figure out the transistor they used , HP had their own foundry and not all parts they made were sold as products on the open market.
Given the frequency range of the yig it was likely to be a HBT rather than a GaAs Fet but HP semiconductor became Avago and then broadcom and to my knowledge they don¡¯t sell die level parts!
The next challenge then is to remove the old transistor and mount a new one, usually devices are mounted using eutectic solder preforms , the stage is heated until the solder melts and the part is then ¡°scrubbed ¡° in as gold from the back metalization
diffuses into the solder the melting point increases and the solder solidifies , in order to remove the part you would have to heat up the solder past the new melting point without imposing thermal stress on the other parts used or dislocating them !
To mount a new part you need a die attach as well as preforms, you could of course use conductive epoxy but the heat transfer generally is not as good as with eutectic die attach so you might run into issues with reliability!
As far as wirebonders concerns there are plenty on ebay , expect to pay between $5000 and $16000 for a refurbished one , for this type of work you need a wedge/wire bonder , if you plan on rebuilding modules plan on getting a deep access bonder or a convertible
machine that can be used for different processes .
The biggest challenge with bonding is to figure out the right settings though, usually in a production environment the supplier of the semiconductor devices used provides bond samples which in essence are parts that are not working ok those samples are
then bonded in a dummy circuit and a pull tester is used to check the strength of the bond?
Also make sure you have someone with a steady hand and young eyes to threat the bonder ? and also invest in a vibration isolation table or platform to set your bonder on?
toggle quoted message
Show quoted text
On Mar 22, 2022, at 02:28, Peter Hansen via groups.io <oz1lpr@...> wrote:
?
I have succeded to integrate a YO84 YIG faily simple and you can use +/-15 for 30V for heater it is ok and shunting the Sense resistors in the Maincoil driver means it can span the 3-7Ghz. I have made changes to the Loop so it is yield low Phase noise. Output
from the YIG it self is good also. Mechanical implementation is done by using the original mechanic shell and make a hole for the YIG. It can be jammed in there when I mount the YIG in the unit. A PCB "still prototype" is mounted with a 10pin pinheader so
the orignal cable can be used. BUT BUT now the challenge. The original YIG does not get hot due to the sensitve coils meaning less drift. If the Loop is opened by placing the jumper in the TEST position I observe a drift of the YIG of around 5Mhz from cold
to warm. The Avantek YIG drift around 30-40Mhz. The Loop is build to compensate for 10Mhz drift. This means you can either have a unlock condition ERR301 during warm up or you will get it during use.
Playing around in the Loop is not a things for the average guy to do. The Frequency Lock? circuit is quite hard to work with. Changes to the loop will also challenge your phasenoise performance. I am sure this is why HP deviated from the standard sensitivity
on the maincoil to keep drift down. The avantek YIG gets hot the HP ones do not.
More fruitfull would be to find a replacement for the Bonded Transistor and get someone with a Wirebonder to service them.
I do currently not have a wirebonder but as I am repairing alot of advanced HP equipment I am considering it. But where to source components for tryouts.
best regards Peter OZ1LPR
Hi Peter, Et al,
Sorry for the delayed response to all the positive comments. I have been away camping, soaking up what is left of the good weather in autumn, trying my new HF radio set up.
No doubt, there will be many challenges ahead, but I am determined to find a solution. I am waiting on a mail from Micro Lambda to see if they have ever made a replacement, the chap is going to ask the question for me, they have been so helpful to date. There
are other Micro Lambda models to suit Tek and R&S, so they have made replacement models for branded SA and Sign Gens. No harm asking.?
Peter, I think the original HP data sheet is right, this does represent a challenge at? or around 40MHz/mA, the hardware and software is tightly integrated, thus a YIG with similar spec or the development of an interface driver is needed. I haven't thought
that far ahead yet.
Like yourself, and I am sure others on the forum have purchased a YIG in the past to adapt, I purchased a Avantek Y084-1215 about 10 years ago, I recently tried to adapt via a crude interface driver, but this did not work out to well. Although the Avantek YIG
works extremely well for it's vintage, it's main coil spec is way out at ~22.4 MHZ/mA. In addition, getting the juice to run the heater from an internal supply is a challenge.
I am prepared to lay out the cash up to a point to develop a solution.
I hope I do not have to eat my words in time.
Stay tuned.
Regards
Gerald
VK3GM
?
|
Re: HP8444A 1.55GHz oscillator
Hi, I bought several years ago an 8444A opt59, this device was new, bought from HP directly. The service manual sold with this unit was the 1817A, with a "manual change" addition for the 2103A, 2121A, 2126A and 2325A. My 8444A is a 2325A. ? Maybe this information can help you. ? ? -Yves ? ? Hi! I've been looking at an HP8444A tracking generator originally for 140 series but used with an 858B. It seems to be a late serial unit 2126A.? Oscillator A7 is similar (but not identical - the coupling to the cavity is adjustable in this one) to the one in the 1323A manual and the 1817A prefix manual. I don't have a later manual - neither does Artek. Having replaced the transistor it is now oscillating at a reasonable (but not quite correct frequency).? When I came to tune I found two adjusters one coupling between transistor and cavity and the other a more normal capacitative tuning adjuster.? The earlier manuals only have one - has anyone seen an adjustment procedure for the two adjuster version??? My frequency reference is not without suspicion?(HP5245L with 0.3-3GHz plugin) which seems the only counter I have that goes over 1GHz. They are great things but it's rare to have no other option but to use it, but now with a new 61 tooth gear (thanks HP for choosing a large prime) it works!
|
Re: HP YIG replacement 3-7GHz
The first challenge is to figure out the transistor they used , HP had their own foundry and not all parts they made were sold as products on the open market.
Given the frequency range of the yig it was likely to be a HBT rather than a GaAs Fet but HP semiconductor became Avago and then broadcom and to my knowledge they don¡¯t sell die level parts!
The next challenge then is to remove the old transistor and mount a new one, usually devices are mounted using eutectic solder preforms , the stage is heated until the solder melts and the part is then ¡°scrubbed ¡° in as gold from the back metalization
diffuses into the solder the melting point increases and the solder solidifies , in order to remove the part you would have to heat up the solder past the new melting point without imposing thermal stress on the other parts used or dislocating them !
To mount a new part you need a die attach as well as preforms, you could of course use conductive epoxy but the heat transfer generally is not as good as with eutectic die attach so you might run into issues with reliability!
As far as wirebonders concerns there are plenty on ebay , expect to pay between $5000 and $16000 for a refurbished one , for this type of work you need a wedge/wire bonder , if you plan on rebuilding modules plan on getting a deep access bonder or a convertible
machine that can be used for different processes .
The biggest challenge with bonding is to figure out the right settings though, usually in a production environment the supplier of the semiconductor devices used provides bond samples which in essence are parts that are not working ok those samples are
then bonded in a dummy circuit and a pull tester is used to check the strength of the bond?
Also make sure you have someone with a steady hand and young eyes to threat the bonder ? and also invest in a vibration isolation table or platform to set your bonder on?
toggle quoted message
Show quoted text
On Mar 22, 2022, at 02:28, Peter Hansen via groups.io <oz1lpr@...> wrote:
?
I have succeded to integrate a YO84 YIG faily simple and you can use +/-15 for 30V for heater it is ok and shunting the Sense resistors in the Maincoil driver means it can span the 3-7Ghz. I have made changes to the Loop so it is yield low Phase noise. Output
from the YIG it self is good also. Mechanical implementation is done by using the original mechanic shell and make a hole for the YIG. It can be jammed in there when I mount the YIG in the unit. A PCB "still prototype" is mounted with a 10pin pinheader so
the orignal cable can be used. BUT BUT now the challenge. The original YIG does not get hot due to the sensitve coils meaning less drift. If the Loop is opened by placing the jumper in the TEST position I observe a drift of the YIG of around 5Mhz from cold
to warm. The Avantek YIG drift around 30-40Mhz. The Loop is build to compensate for 10Mhz drift. This means you can either have a unlock condition ERR301 during warm up or you will get it during use.
Playing around in the Loop is not a things for the average guy to do. The Frequency Lock? circuit is quite hard to work with. Changes to the loop will also challenge your phasenoise performance. I am sure this is why HP deviated from the standard sensitivity
on the maincoil to keep drift down. The avantek YIG gets hot the HP ones do not.
More fruitfull would be to find a replacement for the Bonded Transistor and get someone with a Wirebonder to service them.
I do currently not have a wirebonder but as I am repairing alot of advanced HP equipment I am considering it. But where to source components for tryouts.
best regards Peter OZ1LPR
Hi Peter, Et al,
Sorry for the delayed response to all the positive comments. I have been away camping, soaking up what is left of the good weather in autumn, trying my new HF radio set up.
No doubt, there will be many challenges ahead, but I am determined to find a solution. I am waiting on a mail from Micro Lambda to see if they have ever made a replacement, the chap is going to ask the question for me, they have been so helpful to date. There
are other Micro Lambda models to suit Tek and R&S, so they have made replacement models for branded SA and Sign Gens. No harm asking.?
Peter, I think the original HP data sheet is right, this does represent a challenge at? or around 40MHz/mA, the hardware and software is tightly integrated, thus a YIG with similar spec or the development of an interface driver is needed. I haven't thought
that far ahead yet.
Like yourself, and I am sure others on the forum have purchased a YIG in the past to adapt, I purchased a Avantek Y084-1215 about 10 years ago, I recently tried to adapt via a crude interface driver, but this did not work out to well. Although the Avantek YIG
works extremely well for it's vintage, it's main coil spec is way out at ~22.4 MHZ/mA. In addition, getting the juice to run the heater from an internal supply is a challenge.
I am prepared to lay out the cash up to a point to develop a solution.
I hope I do not have to eat my words in time.
Stay tuned.
Regards
Gerald
VK3GM
?
|
Re: Agilent 3458A Troubleshooting
... this is interesting re battery age, which battery dies first etc:
Alan
|
Re: Agilent 3458A Troubleshooting
p.s. clip of start-up sequence here:
After lighting up all the display segments it says 'TESTING RAM' and then 'TESTING HARDWARE'. About 4 seconds to first reading.
Alan
|
Re: Agilent 3458A Troubleshooting
Hi George,
>>>?6.????The ROMs are socketed. I will try pulling these out and reinserting them. Unfortunately I do not have a spare set.
These are easy enough to make from new. Just buy new chips, download the firmware (e.g. from xDevs), and program with a cheap TL866 programmer or similar. (Or just use the programmer to compare your current ROM contents with a known good download.)
>>>?7.????The Dallas NVRAMs should not affect the BIOS. They are soldered in. I will replace them once I get the meter going.
I'm pretty sure the firmware accesses these on start-up - others may comment. If it sees zeros or nonsense in the calram on boot, it may just stop with a dumbfounded grin :) For testing, you can always download any old machine's calram contents and program up a spare Dallas chip. But do see if you can read the old calram contents though - that may save you a $K or so. If it was me, I wouldn't progress further much without ensuring I had some valid ROM & RAM. Catch 22, but once you can get it to boot you can download the calram contents over gpib.
Yeah, Dave is in Sydney. Also last time I looked (several years ago) - you could still buy some parts from Keysight. I'm not sure if that's still the case.
Alan
|
Re: Agilent 3458A Troubleshooting
Hi Alan, ? Thank you for your post and the encouraging comment about getting the 3458A going. I have my fingers crossed. ? In answer to the points you have raised: ? 1.??? xDevs and eevBlog ¨C I am aware of these and have watched some of the Youtube videos. I thought I would start with groups.io and move on to these later. 2.??? Check power and for toasted components ¨C done, all OK 3.??? Double check for correctly plugged cable connectors etc. ¨C done, all OK 4.??? Boot sequence ¨C I think it is starting the boot sequence properly. The single BEEP at power ON I presume means it has past the first step in POST. I found a very informative site about POST for PCs here: ?
? ??????????? I am guessing the 3458A functions in a similar manner when booting up. ? 5.??? Display voltages ¨C checked OK 6.??? The ROMs are socketed. I will try pulling these out and reinserting them. Unfortunately I do not have a spare set. 7.??? The Dallas NVRAMs should not affect the BIOS. They are soldered in. I will replace them once I get the meter going. 8.??? Electrolytics ¨C I will replace them once I get the meter going. 9.??? Provenance ¨C I found a sticker with Intel written on it. 10. Boards ¨C I will do this if I get desperate. I would rather track down the fault logically if I can. Unfortunately I don¡¯t have any spare boards for swapping. 11. Manuals ¨C I have a lot of info, including the CLIP, but there is no mention of what the flashing single square on the display means. 12. Dave eevBlog ¨C I must get in contact with him. I have seen his Youtube videos. I know he is Australian. I hope he is in Sydney! ? Thank you for providing the checklist ¨C most helpful. ? Cheers, George Georgevits (BE Hons, PhD) Consulting Engineer ?
toggle quoted message
Show quoted text
From: [email protected] [mailto:[email protected]] On Behalf Of alan.ambrose@...
Sent: Tuesday, 22 March 2022 1:41 AM
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] Agilent 3458A Troubleshooting ? Hi George,
Well the good news is that it is something basic rather than something (as yet) complicated.... :)
Another vote here for xDevs site & EEVblog?forum threads. I assume you've done a complete check for power & toasted components. Double check that everything is plugged in where it should be (say, by comparison with say xDevs images) - and that all the cables and sockets seem good. Particularly those to the digital board and display (and check the display voltage too) - it sounds like it's not even starting its boot sequence properly.
If not socketed already, I agree with your intuition, and would desolder, read and then socket the Dallas rams and ROMs. Compare ROMs against known good and also see if you can get any data out of the Dallas rams - I would think you have more than a 50/50 chance. You may want to change the electrolytics while you're there - a list is on my site:?
You don't say much about the provenance - maybe there isn't any, but maybe you can get a few clues? Re: 'but someone has maintained over the years' - hmmm, I think I might take all the boards out and give both sides a good look under a magnifier to see if you can see whether any detail work has been done already. Great care with fingerprint contamination of the analogue board of course. A swap with a known good digital board would be a helpful and quick diagnosis aid if you can borrow one. All the manuals, including schematics, are out there as pdfs, of course.
Ah, you're in Sydney! - I think there's a good chance EEVblog Dave could help a lot in person :) From (hazy) memory I think he has a 3458a or two to compare with / test against.
Alan
|
