Forums :: NEW! Web Resources :: Features :: Photo Gallery :: Vintage Radio Shows :: Archives
Support This Site: Contributors :: Advertise


It is currently Nov Tue 24, 2020 2:32 am


All times are UTC [ DST ]





Post New Topic Post Reply  [ 9 posts ] 
Author Message
 Post subject: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 2:45 am 
New Member

Joined: Oct Sat 24, 2020 2:12 am
Posts: 4
Hi, a bit over a year ago I bought a HP 400D vtvm. The seller described it as working, but having prior experience buying vintage equipment I was expecting a fully working instrument at best or one that is completely fried at worst. The price was OK so I took the risk. One of the auction photos showed the unit powered on with the indicator lamp on, input shorted with a piece of wire, but the movement at full deflection,hmm. That's not good, but at least it is not fried... Also I read in a HP service note there was a known issue with early serial numbered units where the PSU would go into parasitic oscillation. This was supposed to be easily cured by adding a single resistor.

When the unit did eventually arrive, I found the resistor was already installed. Over the course of last year I would spend some time working on it, give up after few days/weeks returning to it later. During this time I recapped the whole unit testing extensively after every single capacitor replacement to make sure it worked the same. I added some additional capacitance to psu ripple filtering getting the actual ripple under 1.5mV at all times, I replaced tubes and reinstalled old ones back in at various combinations.

I used an oscilloscope and a function generator testing every amplifier stage by itself, everything seemed to work fine by itself, but the moment it is reconnected oscillation returns. The only reliable(but unusable) way to get rid of it I found is to replace a 10ohm resistor with 1ohm that makes the global negative gain signal smaller by a factor of 10. This completely ruins linearity of the device though.

The oscillation usually starts somewhere in the kHz range, but very quickly settles at slightly under 20-30hz. Interestingly the waveform is not a true sinusoid, but looks like a very short pulse superimposed onto a slightly skewed sinusoid. The pulse is only 1ms long and it is a
placed at around 30% into the raising edge of the sinusoid.

Most recently I disconnected the negative feedback circuit altogether and I noticed that if I removed the meter movement and its capacitor the oscillation became a lot less skewed and more like a nice sin signal, however with the pulse still in there. I suspect that pulse is key to resolving the issue. I suspect some frequency is shifted by the last stage slightly so when it arrives back at the cathode of the first stage it is 1ms out of phase. This causes that pulse, the rest of the signal being a consequence. Unfortunately fixing this I think requires understanding of the inner workings of amplifiers I don't have so I thought to ask here. Maybe this description "rings a bell" for someone. Maybe someone had a similar issue with another amplifier.

BTW, I also compared my unit with YouTube video showing another one to ensure no components were swapped before I got it, no wires were rerouted etc. As far as I can tell It was pristine. Also it is a model with so called resistive voltage divider.

I can attach the schematic if needed, screenshot of the signal itself etc. I'll be glad for any help.


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 2:09 pm 
Member

Joined: Nov Wed 30, 2016 7:35 pm
Posts: 8459
Location: Sunbury, Ohio 43074
Parasitic oscillation is usually described as higher frequency... not sub 100Hz range. Are you absolutely sure you don't have a power supply filtering issue? Put your scope on the power supply rails and see if your oscillation is there. Set the scope for AC at high gain for this. You don't care about the DC for this test... just any AC riding on it.

One other things none of the "books" mention is that you could easily have a high resistance ground. Any ground path that involves a bolt or rivet is suspect in this age equipment. I'd remove any bolts, clean the metal chassis and component, terminal strip, PC board ...whatever is being grounded there... and use star washers to assure good ground when you bolt it back. Resolder any suspect looking grounds to the chassis.

Also try the old trick of just tapping on things... a good looking solder joint can actually be bad and should show up by gentle tapping.

Another little isolation trick ... successively ground the input to each amplifier stage with a .47 capacitor while watching the last stage output with your scope. You may be able to tell which stage is actually causing the issue.

_________________
Preserving the hist. of electronics, one boat anchor at a time! :)
https://www.bbtvtestequipment.com


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 3:47 pm 
Member
User avatar

Joined: May Wed 18, 2011 2:40 am
Posts: 5065
Location: Littleton, MA
Rokr wrote:
The oscillation usually starts somewhere in the kHz range, but very quickly settles at slightly under 20-30hz.

This sounds like motor-boating. Which versions of the manual do you have? The military manual (http://www.mcmlv.org/Archive/TestEquipment/HP400.pdf, and also on Keysight's web site) is missing some troubleshooting info that's in some of the other versions. The 1956 manual on BAMA, 400DHL002-1/5-59 (http://bama.edebris.com/download/hp/400 ... 0D-H-L.pdf) has troubleshooting info for motorboating on page 13.

Also, check the power supply decoupling capacitors, the electrolytics after the decoupling resistors off the B+ bus. Perhaps one of your replacements is bad.

What's the serial number prefix for your unit?

Where did you add extra capacitance to the power supply? Right after the 6AX5? What value did you use?

_________________
Steve Byan https://www.byan-roper.org/steve/steve-at-play/


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 9:24 pm 
New Member

Joined: Oct Sat 24, 2020 2:12 am
Posts: 4
stevebyan wrote:
Rokr wrote:
The oscillation usually starts somewhere in the kHz range, but very quickly settles at slightly under 20-30hz.

This sounds like motor-boating. Which versions of the manual do you have? The military manual (http://www.mcmlv.org/Archive/TestEquipment/HP400.pdf, and also on Keysight's web site) is missing some troubleshooting info that's in some of the other versions. The 1956 manual on BAMA, 400DHL002-1/5-59 (http://bama.edebris.com/download/hp/400 ... 0D-H-L.pdf) has troubleshooting info for motorboating on page 13.

Also, check the power supply decoupling capacitors, the electrolytics after the decoupling resistors off the B+ bus. Perhaps one of your replacements is bad.

What's the serial number prefix for your unit?

Where did you add extra capacitance to the power supply? Right after the 6AX5? What value did you use?


Thanks I didn't know about that newer manual. I read the page you mentioned and it does have some extra info(One extra voltage measurement in the PSU, and more detail regarding bias voltages) I'll go through the extra bits, and recheck caps on the B+ again later tonight.

The meter serial is 26694.

I removed the extra filtering I added to B+ prior to starting the thread when I decided it made no difference(but I forgot about the removal so I mentioned it in the text). When it was there I had extra 68uF between pin 1 of V7 and ground. It lowered the 50 Hz Ripple to 1.2~1.3mV, but even without it the ripple doesn't exceed 3mV the manual states as max. Prior to this I did try adding extra capacitance right after 6AX5. Normally the ripple there is slightly over 2V, with extra capacitance it went down to 1.7V, but it didn't matter. The voltage stabilisation smooths out those 2V equally well as it did 1.7V. And prior to that I replaced 6U8 tube, when I measured the initial ripple on B+ at over 3mV as spec'ed. After the replacement the ripple is in spec as mentioned previously.

With no modifications to the original circuit, when the self oscillation was going on I could see the waveform previously mentioned in the amplifier and I could see that 1ms pulse making its way into B+(at 10mV short 1ms pulse repeating at around 20Hz). Then I added 68uF of extra filtering in parallel with C26 as I identified the pulse was coming to B+ from V4. Now, even while the oscillation is ongoing the B+ ripple is under 3mV and looks like a nice sinusoid with no pulses. So I'm doubtful this has anything to do with the PSU as the oscillation returns equally quickly after powering on even when no pulses make their way to B+ line.

Barry H Bennett wrote:
Parasitic oscillation is usually described as higher frequency... not sub 100Hz range. Are you absolutely sure you don't have a power supply filtering issue? Put your scope on the power supply rails and see if your oscillation is there. Set the scope for AC at high gain for this. You don't care about the DC for this test... just any AC riding on it.

One other things none of the "books" mention is that you could easily have a high resistance ground. Any ground path that involves a bolt or rivet is suspect in this age equipment. I'd remove any bolts, clean the metal chassis and component, terminal strip, PC board ...whatever is being grounded there... and use star washers to assure good ground when you bolt it back. Resolder any suspect looking grounds to the chassis.

Also try the old trick of just tapping on things... a good looking solder joint can actually be bad and should show up by gentle tapping.

Another little isolation trick ... successively ground the input to each amplifier stage with a .47 capacitor while watching the last stage output with your scope. You may be able to tell which stage is actually causing the issue.


I've been watching the power supply rails closely. Primarily the B+ voltage. Currently they appear in spec (ripple normally 1.5mV, nice sinusoid). I've been using a Rigol DS1054Z and a Tektronix 2465A scopes for this so I think I would see if there was anything else there. I even hooked it up to a spectrum analyzer via a DC block to check if there are no higher frequencies there, but power supply voltage ripples look unrelated to ongoing oscillations in the amplifier.

I heard somewhere about the high ground resistance issue you mention. I'm somewhat reluctant to drill out original rivets so I decided to connect ground points with a very heavy gage wire (2.5mm sq) so the ground connections of filter caps (and many more) are connected via low inductance paths to multiple grounding points. If I still fail to repair this in a bit I'll drill those rivets out just in case.

Tapping doesn't make any difference, but turning the range switch interrupts the oscillation sometimes. The newer manual mentions this as possibly pointing to V1 bias voltage being wrong - I have to check this next.

Regarding your suggested isolation trick, it is V2 - the stage the negative feedback is connected to. Grounding its output wit 22nF (it is the smallest cap that worked) kills the oscillation quickly. Unfortunately higher voltage signals supplied to the input restart it. When the signal is disconnected the oscillation dies down in half a second or so. I only tried this trick as a test and I removed the extra cap later.

For the sake of completeness I should mention that I also doubled the 1500uF filtering capacitors in the DC grid supply voltages lowering the ripple there from 1.4V to 0.7V.

I've been thinking about this in the following way:
- If not the PSU -then it must be negative feedback especially that the oscillation disappears when the feedback loop is disconnected.
- Then how, can negative feedback turn into positive feedback? Some signal must be out of phase between the feedback loop and input.
- From the above I assume the feedback signal is 1ms out of phase (causing the pulse). The pulse repeats at around 20Hz. So I'm looking for some RC network that shifts a 20Hz signal by 2% or 3.6 degrees that somehow appeared near the final stage of the amp.
- Next time I'll try injecting some signals into the final stage of the amp and looking for that phase shift.
Does this sound ok, or am I going on a wild goose chase?


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 9:43 pm 
Member
User avatar

Joined: May Wed 18, 2011 2:40 am
Posts: 5065
Location: Littleton, MA
Rokr wrote:
[I've been thinking about this in the following way:
- If not the PSU -then it must be negative feedback especially that the oscillation disappears when the feedback loop is disconnected.
- Then how, can negative feedback turn into positive feedback? Some signal must be out of phase between the feedback loop and input.
- From the above I assume the feedback signal is 1ms out of phase (causing the pulse). The pulse repeats at around 20Hz. So I'm looking for some RC network that shifts a 20Hz signal by 2% or 3.6 degrees that somehow appeared near the final stage of the amp.
- Next time I'll try injecting some signals into the final stage of the amp and looking for that phase shift.
Does this sound ok, or am I going on a wild goose chase?

I thought you had said it motorboated even when you broke the negative feedback loop? If so, then the motorboating may be from the classic cause of coupling between stages through the B+ line, due to inadequate decoupling.

You might also check the coupling capacitors and make sure the replacements are the correct value. If they're too big, they might also alter the low-end phase shift enough to cause motorboating, though this mechanism should only occur when the negative feedback is connected.

_________________
Steve Byan https://www.byan-roper.org/steve/steve-at-play/


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 10:36 pm 
Member

Joined: Nov Wed 30, 2016 7:35 pm
Posts: 8459
Location: Sunbury, Ohio 43074
I think you may be overanalyzing this. You don't need a spectrum analyzer to find the issue. A scope should be plenty. You said you grounded the output of the second stage with a cap and it went away. What happens when you ground the INPUT to that stage? Don't bypass an output... you want to see outputs. Bypass INPUTS and see where in the chain it is that the problem disappears.

KISS.... this is not a physics problem. Stay simple and basic. Hooking all the grounds up as you say probably won't hurt, but probably wont' help unless that is the issue. You could accomplish the same thing by using a short clip lead from each ground lug to the chassis nearby. One at a time. Nothing here indicates you have multiple problems.

Personally, I'd go back and restore every filter cap to what was designed to be in there. Ditto with any other changes you may have made. Then see where you are with the "kill one stage at a time" trick. I'd start with the first stage and work from there, not backwards from the last. In other words, look at the output of the last stage, and ground the input of stage 1. Then stage 2 . Take what you learn from that and if necessary then, work backwards still looking at the output wiht the scope, and grounding the inputs with a cap.

_________________
Preserving the hist. of electronics, one boat anchor at a time! :)
https://www.bbtvtestequipment.com


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 10:46 pm 
New Member

Joined: Oct Sat 24, 2020 2:12 am
Posts: 4
Barry H Bennett wrote:
I think you may be overanalyzing this. You don't need a spectrum analyzer to find the issue. A scope should be plenty. You said you grounded the output of the second stage with a cap and it went away. What happens when you ground the INPUT to that stage? Don't bypass an output... you want to see outputs. Bypass INPUTS and see where in the chain it is that the problem disappears.

KISS.... this is not a physics problem. Stay simple and basic. Hooking all the grounds up as you say probably won't hurt, but probably wont' help unless that is the issue. You could accomplish the same thing by using a short clip lead from each ground lug to the chassis nearby. One at a time. Nothing here indicates you have multiple problems.

Personally, I'd go back and restore every filter cap to what was designed to be in there. Ditto with any other changes you may have made. Then see where you are with the "kill one stage at a time" trick. I'd start with the first stage and work from there, not backwards from the last. In other words, look at the output of the last stage, and ground the input of stage 1. Then stage 2 . Take what you learn from that and if necessary then, work backwards still looking at the output wiht the scope, and grounding the inputs with a cap.



I can't believe it, but it seems I've fixed the problem and you are 100% correct. It was a very simple thing. Basically resistors R50 and R51 were swapped... I feel pretty stupid now, having spent so much time troubleshooting only for it to be such a simple issue.

I replaced caps connected to those resistors before and I reconnected them the same way as they were without checking if someone before me didn't swap them.

I actually came across it just now going through this newer manual it lists v5 plate voltage as 125 to 145. Imagine my surprise when I found it to be 16V.

Now I have to remove the extra parts I added and calibrate it, but I think I'm very close to making it usable. Even before calibration I hooked a function generator to the meter and the indications appear to be close to correct on all ranges I checked.

Thanks for your help :-)


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sat 24, 2020 11:07 pm 
Member
User avatar

Joined: May Wed 18, 2011 2:40 am
Posts: 5065
Location: Littleton, MA
Rokr wrote:
Basically resistors R50 and R51 were swapped.

That's a good catch, it's not easy to spot wiring errors like that.

_________________
Steve Byan https://www.byan-roper.org/steve/steve-at-play/


Top
 Profile  
 
 Post subject: Re: HP 400D issues with parasitic oscillation.
PostPosted: Oct Sun 25, 2020 9:40 pm 
New Member

Joined: Oct Sat 24, 2020 2:12 am
Posts: 4
I have one more question regarding HP 400D and other devices with very sensitive amplifiers.

Is it typical to see more noise when powered from a normal wall outlet vs powered via an isolation transformer?

After removing extra parts and doing a performance test I swapped jumpers in transformer input to convert the meter from 115V to 230V and I connected it to a normal wall outlet (that supplies 235V here). I'm pretty sure I'm seeing quite a bit more noise than before. (I checked all internal supply voltages etc)

One more thing. While being powered by the isolation transformer the meter chassis wasn't ground. After 230V conversion the unit is properly ground now.


Top
 Profile  
 
Post New Topic Post Reply  [ 9 posts ] 

All times are UTC [ DST ]


Who is online

Users browsing this forum: John Limbach and 10 guests



Search for:
Jump to:  




































Privacy Policy :: Powered by phpBB