[SOLVED] HP 6227B dual DC power supply - oscillation in master channel

[Sparky]

Hello folks,

I really like the HP 6227B dual DC power supply --- it's the one I use on my workbench.  Unfortunately mine developed a regulation problem in the master channel -- there is large ~400mV "sawtooth" oscillation @ power line frequency, ~120Hz.  I attached screen capture from oscilloscope showing this.

The average voltage is correct (5V in o'scope screenshot), just the oscillation @ 120Hz is the problem.  I would like to attempt to fix it     I really like this supply, and the problem is clear so it might be easy to diagnose and correct. 

My first thought is the problem could be with power diodes in the full-wave bridge rectifier circuit (CR15, CR16, CR17, CR18) in the master (A) channel...or perhaps the capacitors (C11, C12, C13, C14).

I am hoping some experts could provide a little guidance since I have not done a DIY-fix on a power supply before.  How should I go about diagnosing the problem?  Are there other parts of the circuit I should measure or inspect?  I wonder if there are capacitors or other parts that may typically need replacement in this HP 6227B?

Any advice will be greatly appreciated!  I hope to document the repair here so we can enjoy satisfaction of fixing the instrument.

For anyone who is following or can advise, the HP 6227B operating and service manual is here.  I extracted the schematics pages and merged them -- see attachment.

Many thanks!

[Tomorokoshi]

Check the diodes first. They are on the bottom of the A1 power board the transformer is soldered to. What kind of DMM do you have? In diode check mode you should see around 0.45 V in forward mode. In reverse it will slowly charge up the capacitor.

It looks like each of the bulk capacitors charges up to around 25V. There should be low ripple on them. Compare to the B channel. You can measure on the screw terminals of the capacitors, although the bleed resistor R77 may be easier to access. Be careful - those store a lot of energy!

[duak]

The waveform looks a little too sharp plus it's not quite 120 Hz so I think it's more likely oscillation.  This can be confirmed if changing the load current changes the frequency.

I found that on power supplies of this design that the output bypass capacitor is needed for stability.  Try placing a 1000 uF > 40 V cap on the output terminals and then checking to see if it makes a difference.  If so I'd verify that C1 is still good.

Cheers,

[Sparky]

Thank you, Tomorokoshi for replying with some guidance!     I am using Keysight 34465A DMM

With DMM in diode test mode I measured CR15(A)-CR18(A) and found consistently 0.438 to 0.455V across each diode when forward biased.  I also checked the SLAVE side (CR15(B) - CR18(B)) with same result.  I accessed these diodes from removing the top cover.  Sometimes it takes a few seconds for the forward voltage to stabilize.

I also measured the other diodes on the A1 motherboard: CR19-CR22(A,B) and readings were consistent (~0.50V).


I then removed the side panels and powered up the unit.  I measured the bulk capacitors: C6, C7 and the bleed resistor R77.  Both MASTER and SLAVE boards were very similar -- about 24.5VDC at the capacitors, and ~0.5V ripple at ~120Hz (power line).  Across R77 I have ~49VDC and 0.8 to 1.0V of ripple.

So far it seems the input stages of MASTER and SLAVE are behaving similarly.  I checked the voltage at output terminals and MASTER channel still showing 400mV oscillation whereas SLAVE does not.  So...the problem is a little further on.


Last night I read more of the operating manual and in Table 5-7 on page 5-17 I noticed it describes the symptom "Oscillates" and probable causes involve C4, C15, C1, C5, C13.  I will test the reference and bias voltages in Table 5-2 and then these capacitors.

Will update with results

[Sparky]

The waveform looks a little too sharp plus it's not quite 120 Hz so I think it's more likely oscillation.  This can be confirmed if changing the load current changes the frequency.

I found that on power supplies of this design that the output bypass capacitor is needed for stability.  Try placing a 1000 uF > 40 V cap on the output terminals and then checking to see if it makes a difference.  If so I'd verify that C1 is still good.

Thanks duak and great observation from the waveform!     I have done a quick test with applying different load and measuring the waveform -- it does seem to change frequency a bit.

The comment you make about the output capacitor C1 is similar to what I have read in the operators manual.  I will look at this capacitor and update soon!

[Sparky]

@Tomorokoshi, duak: I have completed tests of the reference and bias voltages, and also checking the the output bypass capacitor and a few others mentioned in the manual.  Results are as follows:

1) Capacitors: C4, C15, C1, C5, C13

Without removing components from the boards I made several measurements (capacitance, continuity, resistance) across the terminals of the capacitors and compared the component in MASTER (faulty) channel to that of SLAVE (working) channel.  The measurements themselves are meaningless because the readings with the parts in circuit; I just tried to identify any major differences.  I did not find any.  In all instances capacitors in the MASTER channel appear to behave similarly to that in the SLAVE channel.  I did not see any capacitors bulge or leak...all appears okay from the outside.

@duak: Unfortunately I do not have 1000uF > 40V electrolytic cap to put across the output terminals...I will try to find one to perform the simple test you recommended.


2) I performed the reference and bias voltage measurements as indicated in the operators manual (see attachments). 

All measurements for SLAVE channel were within spec -- only the ripple is a little bit out of spec in some cases, but still close.  Several measurements for MASTER channel --- particularly related to REFERENCE REGULATOR part of circuit --- had huge oscillation and well out of spec.  I summarized the measurements in a table and used digitizing mode on Keysight 34465A to capture the voltage waveform at the test points.  The screenshots from DMM show differences between MASTER and SLAVE for each "STEP" of the testing procedure.

From the screenshots in the PDF attached, Step 1 for MASTER shows similar ~0.5V peak-to-peak saw-tooth profile like the SLAVE channel, but after every couple of oscillations the voltage drops rapidly by ~3V, and then quickly rises back up for another couple of ~0.5V oscillations, and drops again, etc.  I wonder what cause this voltage to "give way" suddenly?

As I did not find a problem with capacitors yet -- and not to say that is definitively not a problem -- but is it possible the problem is with REFERENCE REGULATOR section of circuit?  VR1, VR2, Q12-Q15?

Hope the information is clear that I am posting.  Will appreciate your feedback and suggestion for next steps

EDIT: Edits for clarity.

[Sparky]

I continued reading and made some additional measurements...

From the operators manual page 5-12 paragraph 5-63 it says:

"In many cases the reference voltages may be incorrect even though all the components in the reference regulator are functional; this may occur because a defective component or stage in the regulator feedback loop is drawing excessive current and loading down the reference voltages.  In this situation the defective stage can usually be located by following the detailed troubleshooting procedure appropriate for the main symptom (high or low output voltage, etc.)."

Then in paragraph 5-66 on the same page:

"Table 5-7 contains a list of less common troubles and their probable causes.  The troubles in this table are less catastrophic than those listed in Table 5-3; for the most part these symptoms describe degraded performance rather than complete failure.  The possible causes listed should be checked in the order they appear in the table."

So now I am looking at the issues listed in Table 5-7, page 5-16.  It mentions symptoms like "High ripple in output" and "Oscillates" with probable cause and remedial actions.

I wonder, how can I tell if a capacitor is "leaky" ?  I suppose I will require to remove the components from the board to test them?

[duak]

Indeed, if the reference voltage is noisy, the main regulator  will pass the noise on to the output and perhaps distort it making it hard to figure out.

I suggested a value of 1000 uF because it is more common than the existing 1500 uF C1.  Try anything just to see what happens.  It may give you a clue.

Has this supply been worked on before?  I have an hp 6002A that had a number of problems.  There was one shorted pass transistor, but worse, the previous fixer substituted opamps without considering open loop gain, bandwidth and frequency compensation.  The result was that the supply oscillated at various voltages and currents.

Is there any chance that the programming jumpers and screws on the back panel are loose or missing?

Cheers,

[Sparky]

Indeed, if the reference voltage is noisy, the main regulator  will pass the noise on to the output and perhaps distort it making it hard to figure out.

I suggested a value of 1000 uF because it is more common than the existing 1500 uF C1.  Try anything just to see what happens.  It may give you a clue.

Has this supply been worked on before?  I have an hp 6002A that had a number of problems.  There was one shorted pass transistor, but worse, the previous fixer substituted op-amps without considering open loop gain, bandwidth and frequency compensation.  The result was that the supply oscillated at various voltages and currents.

Is there any chance that the programming jumpers and screws on the back panel are loose or missing?

Hi duak, thanks for the follow-up!  I bought the supply on eBay from a liquidator and given I don't see any evidence like flux residue or replaced components I think it's unlikely to have been worked on in the past.  When I got it there were no missing screws or evidence of having been taken apart...

Straps on the back are all in default positions; no screws loose...

I will keep in mind to do a test with a capacitor at the output terminals to check if C1 might be the cause... 

In the meantime, I will work through Table 5.7 and make comparison measurements between MASTER and SLAVE to see if I can isolate anything.

[Sparky]

Update: I looked at a few items in Table 5-7, including reference regulator and mixer amplifier and made various measurements.  For the SLAVE side, measurements are stable and close to nominal values listed on the schematics.  For the MASTER, the measurements are oscillating -- similar to what I captured pictures of for the reference voltages.  In this case, the average voltage doesn't describe what is going on in the circuit -- it is not just different from nominal, but it is oscillating.

And, since the output of one stage of the circuit is fed into the next there are many points of measurement that are exhibiting oscillation, but not really the cause of the problem (merely responding to their changing inputs).

I will get a suitable bypass capacitor to test if C1 is the problem. 

Otherwise, I'm wondering what is a good approach to isolate the circuits from each other and test one stage at a time.  Any ideas, or perhaps something else I should try before doing this?

[Sparky]

Update:  I got a couple of capacitors to try as replacement of the bypass capacitor C1 (1500uF 40V, a strange package with 3 -ve and 1 +ve terminals).  I tried both 1000uF 50V and 3300uF 50V electrolytic.  These had a minor effect on the output, but it still retains the oscillation that I have showed in the original post. 

It seems that C1 is not the problem, and I am still looking for the root cause.

I wonder if there is a way to decouple various parts of the circuit and test each independently? 

Any help much appreciated!

Best,
Sparky

[duak]

I looked at the schematic again and would like to check if the reference regulator output is clean.  The reference circuit is in the upper left of the schematic.  It's floating relative to -Vout but is in fact tied to +Vout through +S (+sense).  Making sure that none of the 6227B outputs are jumpered to earth or safety, connect the ground lead of the 'scope to the master +Voutput.  Measure the DC and AC voltages on testpoints 30 to 33.  I would expect all of these to be clean.  However, when I look at this circuit I see that it uses a common emitter pass transistor which gives it a lot of gain and there could be some sort of instability.

BTW, the unusual waveform shown above reminds me of a square wave after passing through a capacitor or high pass filter.  Is there any chance a probe is defective or a probe clip or ground is not on right?  Is there any other interfence that could be generating a 117 Hz square wave?  Does it make any difference if the unit is covered with something opaque or the lights are off?  Semiconductors are sensitive to light and if they are in transparent cases they could pick up noise.

Cheers

[Sparky]

I looked at the schematic again and would like to check if the reference regulator output is clean.  The reference circuit is in the upper left of the schematic.  It's floating relative to -Vout but is in fact tied to +Vout through +S (+sense).  Making sure that none of the 6227B outputs are jumpered to earth or safety, connect the ground lead of the 'scope to the master +Voutput.  Measure the DC and AC voltages on testpoints 30 to 33.  I would expect all of these to be clean.  However, when I look at this circuit I see that it uses a common emitter pass transistor which gives it a lot of gain and there could be some sort of instability.

Hello duak!  Thanks for keeping in touch with this repair and checking the schematic again.  I do see the REFERENCE REGULATOR section, and the '+S' identified as a "reference" from which the other voltages at test points are measured relative to.

You are correct -- there is instability from somewhere.  I took measurements with the 'scope of MASTER side of circuit at testpoints 30 to 33 (all relative to +S = scope GND connection)...and 34 as a bonus   Please see the attachments

My comments:
TP30 - waveform steady at 12.4V, but something triggers the voltage to drop suddenly...and then charge back up....not sure why it cannot maintain the 12.4V reference.
TP31,32 - perhaps consequential from the changes at TP30.

BTW, the unusual waveform shown above reminds me of a square wave after passing through a capacitor or high pass filter.  Is there any chance a probe is defective or a probe clip or ground is not on right?  Is there any other interfence that could be generating a 117 Hz square wave?  Does it make any difference if the unit is covered with something opaque or the lights are off?  Semiconductors are sensitive to light and if they are in transparent cases they could pick up noise.

The original measurement (in original post) was done with 'scope, but I repeated with DMM on digitizing mode and see the same oscillations.  I also did a "probe check" and made sure the test equipment is working correctly.  Initially I had done all tests with the HP 6227B side panels attached...now the panels are off, but I have not noticed any differences with placing panels back on.

By the way, excellent comment to take precaution that the power supply -VE terminal is not jumpered to Earth, otherwise it would form a ground loop with the 'scope GND lead and the probe or 'scope might blow up!  It is always something to be very careful of when probing with the 'scope!


Could there be a component failure in the REFERENCE REGULATOR circuit that is causing the oscillation?  Why the 12.4V reference cannot be maintained and oscillates?

[Sparky]

One more image: easier to see relationship of wave forms at TP 30, 31, 32 using a few probes at once

TP 30 = Ch 1 = Yellow
TP 31 = Ch 2 = Green
TP 32 = Ch 3 = Blue

Can see that Ch2 and Ch3 are "mirror" each other about the reference signal (0V = +S in schematics), and likely the deviations away from the +/- 6.2V references is due to the change in TP 30 = 12.4V.  So I think need to find what is causing the 12.4V to not remain stable.

[duak]

Sparky, dollars to donuts C10 is bad.  It is not storing enough energy between line half cycles to keep its voltage from dropping below the reference regulator output voltage (plus its operating voltage drop). Its value has either decreased or it's developed a large series resistance.  The regulator has gain plus HF rolloff  giving the waveform an unusual shape

About C10: 80 uF at 65 V is an odd value. I'd go with 100 uF at 63 V (nearest standard).  In a pinch a 50 V cap should be fine as it has 27.5 V nominally across it.  You could try one of the other caps you've got but one consideration is that in a power supply with more than one interlinked power source is that during start up and shut off the regulator could develop unusually high voltages.  I think this circuit is OK, but some supplies are infamous for damaging their loads.

Cheers,

[Sparky]

Sparky, dollars to donuts C10 is bad.  It is not storing enough energy between line half cycles to keep its voltage from dropping below the reference regulator output voltage (plus its operating voltage drop). Its value has either decreased or it's developed a large series resistance.  The regulator has a lot of gain thus giving the waveform it's unusual shape

About C10: 80 uF at 65 V is an odd value. I'd go with 100 uF at 63 V (nearest standard).  In a pinch a 50 V cap should be fine as it has 27.5 V nominally across it.  You could try one of the other caps you've got but one consideration is that in a power supply with more than one interlinked power source is that during start up and shut off the regulator could develop unusually high voltages.  I think this circuit is OK, but some supplies are infamous for damaging their loads.

Cheers,

Thanks duak!    This makes sense and looks like we might have a simple fix!  I don't have suitable replacement capacitor lying around...  I will try to scavenge one from something...otherwise might be best that I place a Digikey order for several of the capacitors and replace them together.

Will update with how replacement goes!

Best!

[Sparky]

SOLVED! 

duak is right on the money -- dollars and not donuts in this case!  I replaced C10 (80uF, 65VDC) with a 100uF 160VDC (closest I could get off the shelf locally) and the oscillation problem is gone! 

See photo attached the the part replaced, and also highlighted with yellow box where replaced.  The cap that went in was measured (using Keysight 34465A) to be ~96uF, whereas the one the came out was measured at 0.01uF!  Possibly cannot trust that figure since there could be more than one problem with the faulty capacitor...but it's clearly not right.

I think it will be a good idea to replace the other electrolytic capacitors -- they would be on the way out after all these years, and difficult to assess their condition without removing them.  So I'm considering to replace them.  In that respect I have a few general questions:

1. Should I just do the electrolytic caps?  What about the ceramic and mylar caps?  Do they also deteriorate over time?  Is it a good idea to simply replace without much thought?

2. Many new electrolytic caps seem are much smaller (at same rated capacitance/voltage) as the ones that have been used.  Is this just because construction/materials are better, or are the original parts really custom and it's not recommended to replace with "like" parts of the same (or close) value.

3. C1, C6/7 electrolytics are unique mechanically with "snap in" (for C1) or "screw terminals" for the larger bulk capacitors.  These are the largest caps in blue in the second picture attached.

 - C1 looks to be of type 4-lead "snap in".  Not sure I can find a replacement.  Would it be essential to have 3 GND leads, or not affect the performance if they were simply connected?

 - This looks like potential replacement for C6/7.  I would have to check they will fit the space and the screw terminals are located correctly.

Perhaps it's wiser to leave these unique capacitors "as is" than attempt to replace?



Thanks very much to duak for the guidance leading to success of fixed HP 6227B!   It was a valuable experience for me. 

[bitseeker]

This thread is a little old now, but since there were open questions, I figured I'd fill them in for the next 6227B owner who might wonder about the same things.

1. Should I just do the electrolytic caps?  What about the ceramic and mylar caps?  Do they also deteriorate over time?  Is it a good idea to simply replace without much thought?

There's a lot of debate over this topic, but my rule of thumb is if it's an alumin(i)um electrolytic cap from before the capacitor plague era and it's not leaking or causing ripple issues (or have a known history of containing leaking caps), I'd leave them alone.

2. Many new electrolytic caps seem are much smaller (at same rated capacitance/voltage) as the ones that have been used.  Is this just because construction/materials are better, or are the original parts really custom and it's not recommended to replace with "like" parts of the same (or close) value.

The smaller size of modern capacitors is due to technological improvements. It's OK to use them as replacements as long as the capacitance is the same or a little higher (e.g., next value up from the original) and the voltage rating is the same or higher. Depending on the in-circuit purpose, other capacitor characteristics such as ESR (too high or too low can affect behavior) may also be important.

3. C1, C6/7 electrolytics are unique mechanically with "snap in" (for C1) or "screw terminals" for the larger bulk capacitors.  These are the largest caps in blue in the second picture attached.

 - C1 looks to be of type 4-lead "snap in".  Not sure I can find a replacement.  Would it be essential to have 3 GND leads, or not affect the performance if they were simply connected?

 - This looks like potential replacement for C6/7.  I would have to check they will fit the space and the screw terminals are located correctly.

Perhaps it's wiser to leave these unique capacitors "as is" than attempt to replace?

If they're not leaking and are filtering properly, I'd leave them alone.

[Sparky]

This thread is a little old now, but since there were open questions, I figured I'd fill them in for the next 6227B owner who might wonder about the same things.

Thanks bitseeker for taking time to answer those stray questions.    Its great contribution for the reasons you mentioned!

My HP 6227B is working flawlessly    

[Chris56000]

Hi!

I'll just mention in passing that if you've access to Digi–Key or Mouser, you've a good chance of being able to  buy an 82uF (or even 80uF!)/electrolytic of suitable size and voltage rating – I have found, over many years experience, that many U.S. designed power–supplies are very fussy about the quality of replacement capacitors, op–amps, transistors, etc., etc., and if you do go to the extra cost of rooting thro' Mouser, even if it means paying more & waiting longer, you'll be rewarded with a repair that works first time!

Have a look through them, you'll be surprised how many odd American capacitors Mouser still list!

Chris Williams

[kj7e]

In my recent HP 6227B acquisition I found most of the electrolytics were bad or at best degraded.  The 3000uF 40v C6 & C7 on both the main and slave board were bad.  The 5uF 50v C2 on both the main and slave boards, bad.  The 325uF 35v C9 & C9A on the motherboard were slightly degraded as well as the 1500uF 40v C1's.  I'm just going to replace all of the electrolytics.  The Mylar and ceramics are fine.

Even with all the degraded and bad caps the power supply still worked very well until the current draw or voltage was raised beyond a certain point.  The 120Hz ripple got a bit ugly near max output.

C1, I found NOS 1500uF 40v Sprague on eBay, they tested perfectly on my LCR meter when they arrived, both at 1560uF with very little leakage.  Staying original as these are 5 pin snap-ins for mechanical stability.  A radial cap could be used here, only two pins are being used on the board.

C2, replacing with 4.7uF 63v https://www.mouser.com/ProductDetail/75-516D475M063JL6AE3

C6 & C7, replacing with 12000uF 50v of the exact same physical size https://www.mouser.com/ProductDetail/661-E36D500N123TC54M
These are a bit larger but not by so much it will cause inrush concerns, the larger caps here will only do good things.

C9 & C9A replacing with 470uF 63v https://www.mouser.com/ProductDetail/75-516D477M063PS6AE3
This is a DC filter cap for the crowbar supply, a bit larger will not hurt anything here.

C10 & C10A, replacing with 180uF 100v with the correct 5mm lead width https://www.mouser.com/ProductDetail/661-EKYA101E181MK20S
DC filter cap for the reference regulator supply, a bit larger here is good.

C11, replacing with 4.7uF 63v https://www.mouser.com/ProductDetail/594-2222-021-28478

C13, replacing with 22uF 63v https://www.mouser.com/ProductDetail/75-516D226M063LL6AE3

Mine starts with serial # 2451A, all the caps have date codes from 1984, but I suspect it was made a bit later than that due to the rocker style power switch, however mine still has the course and fine pots, not the 10 turn pots the OP's has.  Sparky, what is your serial number?



Bad main filter caps;


The main and slave boards slide out the back with ease;


Then split in two;


Marginal C1 caps, replaced with NOS;


Slightly degraded crowbar supply C9;


The dead C2;


Getting to the caps on the Motherboard;




The parts will be here later this week.  Ugh!... the wait.

[kj7e]

Some of the caps came in, unfortunately these will be needed last 





From HP, the 3000uF caps were held off the boards with zinc plated steel spacers.  I'm ditching the flaking zinc spacers and steel 10-32 cap screws.  Going to use shorter stainless 10-32 screws/washers.

You can see the flaking zinc spacers here;

[Tomorokoshi]

I hope the fuse and bridge diodes are okay when initially charging those big capacitors.

[kj7e]

I hope the fuse and bridge diodes are okay when initially charging those big capacitors.

It is a concern of mine as well.  The fuse is 4 amp and the 3amp rectifiers have a 300amp peak forward surge capacity, they should be fine.

However, these would have been my first choice but for the minimum order and 18 week lead time;
https://www.mouser.com/ProductDetail/United-Chemi-Con/E32D800LPC472MC48M?qs=sGAEpiMZZMtZ1n0r9vR22b0zSdKjonvpl1EYJRvMDy0%3D


This got me searching again, I found these Cornell Dubilier 4000uF 100v caps, correct size and 10-32 screw lugs spacing (22.2mm), a bit more pricey but they would be a better fit for the application.  They appear in stock at both Digikey, Mouser and Newark.   CDE Part Number: 500C402T100BA2B
https://www.digikey.com/product-detail/en/cornell-dubilier-electronics-cde/500C402T100BA2B/338-1264-ND/953312
https://www.mouser.com/ProductDetail/Cornell-Dubilier-CDE/500C402T100BA2B?qs=%2Fha2pyFadujg2rIDzyLu8NDFTr4qtIMRCLsXUHHCCsovyJRsNFGW8g%3D%3D
https://www.newark.com/cornell-dubilier/500c402t100ba2b/aluminum-electrolytic-capacitor/dp/31P6804?ost=500C402T100BA2B&ddkey=https%3Aen-US%2FElement14_US%2Fsearch

[kj7e]

I decided to play it safe and went for the Cornell Dubilier 4,000uF 100V caps.  The really nice thing about these are they have the High Post's so no need to use the spacer between the posts and the board;




The blue caps are the replacements;


Blue and the two 180uF 100v;


Re-assembly;




Testing and calibrating the meters;




Not much ripple at max load;




All of the caps I replaced;