Author Topic: Strange Osiloscope Anomaly  (Read 2194 times)

0 Members and 1 Guest are viewing this topic.

Offline JwillisTopic starter

  • Super Contributor
  • ***
  • Posts: 1725
  • Country: ca
Strange Osiloscope Anomaly
« on: February 13, 2020, 10:57:44 am »
I'll try to give as much information as I can.
I'm building a powers supply and things are going well but I discovered a weird anomaly on the Output that measures around 40-60 uVpp and I can't pin it down. At first I thought it was an oscillation from one of the op amps but  it's not a consistent oscillation one would expect. It pulses at around 1 second on load and doesn't seem to increase with an increased load.

At the output at  0.1V/Div I get a nice clean DC line  at loads up to 10 Amps at 10 V (haven't gone higher than that ) and about 100mV of ripple at the rectifier which I expected. I need more caps I don't have right now.
But at 20uV/Div this weird pulse shows up. It only appears under load. I changed some caps around the op amps but it doesn't make any difference.Could it be external?

I added a schematic and the best pictures I could of the scope .It's hard to see because it's faint on the scope unless I increase the intensity. And it's hard to take an photo of both channels because of the scan rate between both channels of the scope. 
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 15159
  • Country: de
Re: Strange Osiloscope Anomaly
« Reply #1 on: February 13, 2020, 01:24:00 pm »
The second scope picture looks like something is oscillating at some times. So the circuit may be just at the edge.
It is odd to see a different picture with the other range.

The load at the output of the voltage regulating OP could be a problem: the OP sees quite some  capacitive load (68 nF in FB) with 100 nF to the output. Not all the capacitance is directly visible, but at the higher frequencies some of it is. So there is a chance the OP could oscillate at least under some load conditions.

The circuit looks like a high current version of the common simple kit. However this circuit is not really that useful at high power, as it does not use tap switching or similar to reduce the power loss. Chances are the transistors would not survive much more than 10 A total at low output voltage due to SOA limitations.  The left 2 OP in the circuit do not have to be expensive high voltage OP. One could power them from a lower supply.
There are few other odd choices (e.g. transistor types).
 
The following users thanked this post: Jwillis

Offline rdl

  • Super Contributor
  • ***
  • Posts: 3667
  • Country: us
Re: Strange Osiloscope Anomaly
« Reply #2 on: February 13, 2020, 01:41:57 pm »
It could be external. Disconnect the scope probes and see if the noise changes or goes away.

Scopes can be good radio receivers at times.   ;D

 
The following users thanked this post: Alex Eisenhut, Jwillis

Offline Andy Watson

  • Super Contributor
  • ***
  • Posts: 2138
Re: Strange Osiloscope Anomaly
« Reply #3 on: February 13, 2020, 02:44:57 pm »
Are the busrts of oscillation synchronised with the mains ? You could try slowing-down the main bridge rectifier by adding 10nF across each diode.
 
The following users thanked this post: Jwillis

Offline JwillisTopic starter

  • Super Contributor
  • ***
  • Posts: 1725
  • Country: ca
Re: Strange Osiloscope Anomaly
« Reply #4 on: February 13, 2020, 11:07:08 pm »
Its a one second burst/pulse. And it seems to go back to the rectifier after further investigation. Harder to see because it's seems to be imposed on the AC wave .I'm going to set up another transformer later and check the Mains for noise .
The second picture I turned the time/Div higher so I could get a decent picture. That's why the Chan 2 is different from the first picture. I should have mentioned that.

I chose the transistors because at the time the 2SC2922 wasn't obsolete . But the 6  will indeed  handle 20 Amps. These are really tough transistors.  I have 4 more That I want to add when I can get a larger heat sink.
The other ones were more standard when I started but I found that these worked  best and It's what I have on hand without ordering different ones.
Most long duration loading  is around 10 to 15 amps with forced air cooling. The 2SC2922 stay at a reasonable temperature at 15 amps . I even had a malfunction with the load and they handled the a full 20 amp short easily for a couple minutes without damage . So I'm not to worried about the power transistors. I plan to make the fans temp controlled but will do that later. As for the expensive op amps I have 50 of them . I'm not worried about that and not intending on putting this thing on any market . Personal use and learning experience only.
Any thing less than the 68nF at the output reg causes severe oscillations to the point of being audible. Surprised the Op amp didn't burn out when that happened.
Getting off the shelf multiple tapped transformers for specific applications is hard unless I get custom made which is expensive. I understand that multi tapped  is a more efficient way to go , but not so practical for a simple hobbyist. Besides winding high current toroids  by hand is not something I would look forward to doing .But I did have thoughts about  using a second or third smaller transformers  to switch between. Maybe later.

The design is indeed a variation of a popular kit to resolve some issues that kit had. And it kind of evolved into what I have now. I could go with lower voltages or another proven circuit for sure . That's to easy.  I wanted to try something a little different  so I can learn from my errors as well as success .

https://www.semicon.sanken-ele.co.jp/sk_content/2sc2922_ds_en.pdf
Thank you for your insight and suggestions.
 

Offline JwillisTopic starter

  • Super Contributor
  • ***
  • Posts: 1725
  • Country: ca
Re: Strange Osiloscope Anomaly
« Reply #5 on: February 14, 2020, 07:42:22 am »
The second scope picture looks like something is oscillating at some times. So the circuit may be just at the edge.
It is odd to see a different picture with the other range.

The load at the output of the voltage regulating OP could be a problem: the OP sees quite some  capacitive load (68 nF in FB) with 100 nF to the output. Not all the capacitance is directly visible, but at the higher frequencies some of it is. So there is a chance the OP could oscillate at least under some load conditions.

The circuit looks like a high current version of the common simple kit. However this circuit is not really that useful at high power, as it does not use tap switching or similar to reduce the power loss. Chances are the transistors would not survive much more than 10 A total at low output voltage due to SOA limitations.  The left 2 OP in the circuit do not have to be expensive high voltage OP. One could power them from a lower supply.
There are few other odd choices (e.g. transistor types).


I figured it out. Theirs a timer on the control board I'm using for my electronic load.  That's where the 1 second burst/pulse noise is coming from. I'll have to reverse engineer it and see if I can filter that out . But at least I know to ignore it for now.
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Strange Osiloscope Anomaly
« Reply #6 on: February 14, 2020, 01:10:47 pm »
Try removing the 0.01uF if not already.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 
The following users thanked this post: Jwillis

Offline orbanp

  • Regular Contributor
  • *
  • Posts: 169
Re: Strange Osiloscope Anomaly
« Reply #7 on: February 14, 2020, 04:11:34 pm »
Hi Jwillis,
I did simulate the voltage control loop of the PS.The 920 Ohm resistors in series with the pass transistor bases will limit the max output current of the PS. Leave them out, those are not needed.The 100nF capacitor parallel with the 56k resistor in the negative feedback path will cause instability at light loads. Leave that out too.
Regards, Peter
 
The following users thanked this post: Jwillis

Offline JwillisTopic starter

  • Super Contributor
  • ***
  • Posts: 1725
  • Country: ca
Re: Strange Osiloscope Anomaly
« Reply #8 on: February 14, 2020, 11:12:21 pm »
Try removing the 0.01uF if not already.

Not sure which 0.01 capacitor your referring to.


Hi Jwillis,
I did simulate the voltage control loop of the PS.The 920 Ohm resistors in series with the pass transistor bases will limit the max output current of the PS. Leave them out, those are not needed.The 100nF capacitor parallel with the 56k resistor in the negative feedback path will cause instability at light loads. Leave that out too.
Regards, Peter


In theory your simulation would be correct but theirs a problem with simulations of parallel transistors. They give false positives unless you account for the fact that like type of BJT transistors do not have exactly the same Vbe . The transistor with the lowest Vbe will turn on first preventing the others from turning on. This leads to load hogging. So what happens is the lowest transistor turns on and quickly burns out. Then the next lowest Vbe turns on and burns out and so on . This problem effects  all BJTs including  darlinton pairs because of the PN junction.   By adding a current  limiting resistors at the base , this keeps the transistors more balanced to an extent .But mostly I want set  the parallel transistors from exceeding a maximum current of 20 Amps  at 40 volts . At 40 volts the maximum current  of each transistor cannot exceed 4 amps or it falls outside the SOA . So I set each transistor to to carry a maximum of 3A to 3.5A . It's an added safety measure in case something goes south. At 40V with such a high amperage things start to get very close to the edge of the SOA. I have more transistors but not enough room on the heat sink.
This of course doesn't prevent all load hogging because the transistor with lowest VbE  is more forward biased then the rest. So it  takes  more load then the others. This causes it to heat up more thereby  conducting more current which heats it up more and so on. It goes into thermal runaway.  By adding a emitter resistor of low value it keeps the thermal run away under control. Right now the resistors installed are not very good and I'm in the process of changing them to better ones that just arrived.This should further improve stability.
By removing the 100nF by the 56K resistor cap this causes enormous oscillations in the final regulator.  I tried that. The 68nF would have to be increased to compensate.
The pulse issue is not in the PS but the electronic load. Theirs a really noisy timer on the control board. So the problem needs to be fixed there.
By using a 1 ohm resistor as the load instead of the electronic load theirs no noise or oscillation on the PS output or at any of the op amps .
The only issue I have now is the amount of capacitance on the reserve side of the PI filter is to low. Right now it's sitting at 100mV under full load so this causes a voltage drop at the output. But I expected that. At the time the pictures were taken there was only 20Kuf at the reserve side of the PI filter. Since then I've added 40KuF. That's why it looks like more ripple than 100mV.

Thank you I really appreciate the input.
     
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Re: Strange Osiloscope Anomaly
« Reply #9 on: February 15, 2020, 05:46:38 am »
My mistake. Try removing the 0.1uf across the 56K.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline orbanp

  • Regular Contributor
  • *
  • Posts: 169
Re: Strange Osiloscope Anomaly
« Reply #10 on: February 15, 2020, 12:16:47 pm »
Hi Jwillis,
The pass transistor balancing is done by the 0.1 Ohm emitter resistors, and not by the base resistors!Learn from the schematics of other linear regulators!
Your simulation is as good as your models! I did not find a 2SD2061 model, I used a TIP31 model. The 2SD2062 has an Ft of 8MHz, the TIP31 3MHz.
Still, experience has shown that at such low frequencies BJT circuit simulations are quite reliable.Light load stability is unacceptable at 150mA load and with the 100nF cap attached.Do read up about linear regulator compensation here: https://www.ti.com/lit/an/snva020b/snva020b.pdfI enclose the picture of the loop gain!
Regards, Peter
PS: Sorry for the jumbled format, my browser settings, or something else, cuts spaces and new lines arbitrarily.
« Last Edit: February 15, 2020, 12:20:55 pm by orbanp »
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 15159
  • Country: de
Re: Strange Osiloscope Anomaly
« Reply #11 on: February 15, 2020, 01:53:20 pm »
The 100 nF parallel to the 56 K may be part of the loop compensation, though they look rather large.  To reduce the capacitive load on the OP, one could add some 100 Ohms to 1K in series to that cap.  This should not effect the regulation loop that much, but keep the OP more happy.

The usual SOA is for a perfect heat sink. So one should keep the actual power lower than the limit quite a bit. The 2SC2922 is in deed a rather large transistor, but still boarder line.
Current sharing is more efficient with emitter resistors than base resistors. So the base resistors may be OK, though the cost a little extra voltage. This circuit may like an additional fast acting current limit to avoid excessive spikes on transients (e.g. a short from a high voltage). The base- resistors alone are not very effective here. 

Still to get 40 V out the raw DC voltage may need to be in the 45-50 V range. This is one weakness of the circuit. Especially at high power, one should have separate filtering for the critical OP, so that the OPs supply does not see as much 120 Hz ripple. This way one could save quite a bit on the large filter caps and this way improve the power factor a little and save also on the transformer. An excessive amount of capacitance also needs a larger transformer.

Transformers with a symmetric split supply (e.g. 2x18 V) are relatively common. Those low power auxiliary supplies may have to use separate transformers.

 
The following users thanked this post: Jwillis

Offline JwillisTopic starter

  • Super Contributor
  • ***
  • Posts: 1725
  • Country: ca
Re: Strange Osiloscope Anomaly
« Reply #12 on: February 16, 2020, 10:00:14 am »
I'm scared of not having Base resistors coming from past experience. The  2SC2922 is an expensive transistor and hard to find without the complementary paired 2SA1216.Occasionally I find them for around $8 but most times they can be as high as $20 . So I'm being cautious by putting the base resistors in to limit their maximum to 3.5A each,
I need more Heat sink but really don't want to spend hundreds of dollars for a new one. I'm looking for another one that's more in my price range. I have 4 more transistors to bring things into a safer operating region.
I have a 40V 1000VA transformer for this project but I'm concerned  about the  56Vdc after rectification.
I would have to recalculate everything I've done .I guess I could set it for 30V instead of 40V output. That's easy to do.
I have the project working really well but will look into the small tweaks people have suggested.

Thank you everyone
 

Offline orbanp

  • Regular Contributor
  • *
  • Posts: 169
Re: Strange Osiloscope Anomaly
« Reply #13 on: February 16, 2020, 01:07:08 pm »
Hi Jwillis,
Building a linear power supply for the above specs (40V, 20A) is the brute force approach. You need the "big" hardware for that. Maybe you need to reevaluate your specs or the regulator architecture. Commercial linear PSs with such specs usually have an SCR based pre-regulator so the linear part needs to drop only 5v to 10V, the power dissipation is limited.For inspiration look for the HP-625x or HP-626x series power supplies.Also, you could look at this thread: https://www.eevblog.com/forum/projects/thrysistor-voltage-regulator/ The Linear application note by Jim Williams (AN-32, also referenced in the above thread) is also a worthwhile read. As for expensive transistors, you could use the KSE44H series transistors, have a bit lesser specs, 80V, 10A, but a lot cheaper (C$12.53 for 10): https://www.digikey.ca/en/products/detail/on-semiconductor/kse44h11/1053454. The complementer pair is the KSE45H.
Regards, Peter
« Last Edit: February 16, 2020, 01:11:25 pm by orbanp »
 
The following users thanked this post: Jwillis

Offline JwillisTopic starter

  • Super Contributor
  • ***
  • Posts: 1725
  • Country: ca
Re: Strange Osiloscope Anomaly
« Reply #14 on: February 16, 2020, 11:30:51 pm »
Hi Jwillis,
Building a linear power supply for the above specs (40V, 20A) is the brute force approach. You need the "big" hardware for that. Maybe you need to reevaluate your specs or the regulator architecture. Commercial linear PSs with such specs usually have an SCR based pre-regulator so the linear part needs to drop only 5v to 10V, the power dissipation is limited.For inspiration look for the HP-625x or HP-626x series power supplies.Also, you could look at this thread: https://www.eevblog.com/forum/projects/thrysistor-voltage-regulator/ The Linear application note by Jim Williams (AN-32, also referenced in the above thread) is also a worthwhile read. As for expensive transistors, you could use the KSE44H series transistors, have a bit lesser specs, 80V, 10A, but a lot cheaper (C$12.53 for 10): https://www.digikey.ca/en/products/detail/on-semiconductor/kse44h11/1053454. The complementer pair is the KSE45H.
Regards, Peter
 

I completely forgot about SCR regulators. Now you got the gears turning again. I haven't worked with many SCRs.
I have some of those D45H11 and D44H11 but I thought they were obsolete so I didn't bother looking for more .They're a pretty robust transistor . 

Thanks.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf