Author Topic: Audio power amp has large DC offset in output - SOLVED!  (Read 1434 times)

0 Members and 1 Guest are viewing this topic.

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #25 on: June 05, 2021, 11:39:36 pm »
Quote
Also, if you can, measure R1, R2, R3 and R4 in circuit and check their related capacitors, especially C4.

The resistors all check out okay. I checked the caps with my ohm meter and there were no shorts. They all had a slowly increasing resistance which I interpret as the meter charging the caps and the meter getting confused by the way the caps are absorbing the current.

While testing the pins of the LM3886s for DC voltage some even more strange behaviour is manifesting. I haven't touched the amp since yesterday when I was seeing the negative DC offset at the output. Today when picked it up again to take voltage measurements on the pins of the LM3886s I kept the scope on the output for good measure. The first time I powered on the amp I heard a high pitched whistle-like noise that lasted about 3 seconds. When that was happening I saw a positive offset on one channel, and both channels had some oscillations. After about 3 seconds the behaviour stopped and the signals were relatively flat. I don't recall if there was still some offset...the episode came and went rather quickly and I didn't get all the details.  I was unable to recreate the symptom. I figured it was an artifact of some cap charging so I waited about 5 minutes and tried again but the symptom did not reappear. But later as I was taking voltage measurements on the pins a similar but shorter lived thing happened a couple of times. But there was less than a minute between powering on for a reading and powering on again for the next reading.

So I continued on with taking voltage measurements on the pins. I turned on the amp for only a few seconds to get a reading and  then I turned it off again. For the first 4 pins, the left channel continued to show about +20VDC offset. When I went to take a reading on the 5th pin the offset on the left channel had changed to -16VDC. The offset has remained at -16VDC with 2.8 vpk ripple at around 140 Hz. The chip on the left channel does not appear to be heating up despite the large negative current at the output. The 8 ohm resistive load is definitely heating up (2 16 ohm, 25 watt resistors in parallel).

The right channel also has some very erratic behaviour. On power up there is a lot oscillation with about -2VDC offset at the output (with 8 ohm resistive load) that changes rapidly while the offset slowly decreases to 0V. The chip on the right channel gets quite warm within a few seconds although unlike the left channel, the resistive load on the output is not heating up at all.

I'm attaching the data for the pin measurements as well.

It sure looks like the LM3886s are toast. The only thing I can think of as a possible source of the damage could be the power to the preamp board that I'm working on. I don't know exactly when the problem started...I was focused on the preamp when I powered it all up together. There was no signal from the preamp to the power amp. But the preamp is powered by the same power supply so it's plausible that something cause a spike to the LM3886s. It uses some linear voltage regulators to step down the voltage from +-27 VDC from the power supply. I'm attaching a schematic of the preamp for reference in case it helps.
 

Online xavier60

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: au
Re: Audio power amp has large DC offset in output
« Reply #26 on: June 06, 2021, 12:12:48 am »
It looks like DC is being externally applied to the inputs.
Hioki AS100D vom, HP 54645A dso, Fluke 87V dmm, AN8008 dmm, Agilent U8002A psu,  FY6600 function gen, New! Brymen BM857S-(With Battery)
 

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #27 on: June 06, 2021, 12:27:25 am »
It looks like DC is being externally applied to the inputs.

I agree it looks that way but I'm at a loss to see how it could be coming from outside the chip.
 

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #28 on: June 06, 2021, 12:30:00 am »
Quote
LM3886 amplifiers kits do fall into audiophool territory, some ripoff parts out there like the filter capacitors being hollow and tiny crap parts inside, as fakes.

Parts were sourced from Mouser so I assume they are genuine.
 

Online xavier60

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: au
Re: Audio power amp has large DC offset in output
« Reply #29 on: June 06, 2021, 12:32:04 am »
Have you checked for DC at the left side of the input caps?
Is signal ground tied to power ground?
Hioki AS100D vom, HP 54645A dso, Fluke 87V dmm, AN8008 dmm, Agilent U8002A psu,  FY6600 function gen, New! Brymen BM857S-(With Battery)
 

Online bdunham7

  • Super Contributor
  • ***
  • Posts: 2632
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #30 on: June 06, 2021, 12:59:06 am »
So look at that circuit and tell me how it is even remotely plausible that you have +20 volts on the output (pin 3) and -19.5 volts on both inputs  (9 & 10).  Something is not connected correctly.  You need to put your negative lead of your DMM on the first power ground point and then test every other point in the circuit, including the ground points.   Can you post a good photo?
A 3.5 digit 4.5 digit 5 digit 5.5 digit 6.5 digit 7.5 digit DMM is good enough for most people.
 

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #31 on: June 06, 2021, 01:11:06 am »
Have you checked for DC at the left side of the input caps?
Is signal ground tied to power ground?

I haven't. For the right channel I see - 22.5 VDC on the chip side of C1 that is slowly increasing toward 0 VDC and - 14 VDC on the input side. For the left channel, I get - 20.5 VDC steady on the chip side and -16ish VDC on the left that is slowing increasing toward 0 VDC.  And just for fun the offset on the left channel is now + 20 VDC.

Neither signal ground nor input are connected to anything for both channels.

 

Online bdunham7

  • Super Contributor
  • ***
  • Posts: 2632
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #32 on: June 06, 2021, 01:18:10 am »
Neither signal ground nor input are connected to anything for both channels.

 :o :o :o

The signal ground should be connected to the power ground at all times!  Didn't you respond to madires and show a diagram of them connected that way?

And both of your C4 capacitors may have issues, they really need to be bipolar.  I suspect they leak at working voltages.
« Last Edit: June 06, 2021, 01:20:29 am by bdunham7 »
A 3.5 digit 4.5 digit 5 digit 5.5 digit 6.5 digit 7.5 digit DMM is good enough for most people.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 4470
  • Country: ca
Re: Audio power amp has large DC offset in output - SOLVED!
« Reply #33 on: June 06, 2021, 01:18:49 am »
In your schematic, I see missing resistors for the pre-amp section. U4B, U4C (+) inputs need them for input bias current of the op-amp and capacitor leakage current. C10, C11 33uF is kinda huge there.
This could make mystery DC offset going into the tone controls as well, and getting out to the power amp.
 

Online xavier60

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: au
Re: Audio power amp has large DC offset in output - SOLVED!
« Reply #34 on: June 06, 2021, 01:24:01 am »
Solved?
Hioki AS100D vom, HP 54645A dso, Fluke 87V dmm, AN8008 dmm, Agilent U8002A psu,  FY6600 function gen, New! Brymen BM857S-(With Battery)
 

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #35 on: June 06, 2021, 01:24:37 am »
Have you checked for DC at the left side of the input caps?
Is signal ground tied to power ground?

The problem was that I didn't have the signal ground tied to the star ground point. I had the input disconnected because I was testing a new preamp board design and I didn't want to involve the power amp in the signal chain. But when I disconnected the signal ground it isolated he signal side of the power amp board and so it was left floating.

I'm pretty new to this but it seems like it makes sense that the op amps were acting like comparators and as someone else commented there didn't seem to be any functioning feedback. Also makes sense that sometimes the offset was positive and sometimes it was negative since the feedback circuit is referenced to ground. Since it was floating that would seem to explain the very erratic behaviour. With the signal grounds actually grounded everything is quiet, cool, and sitting stable at O V at the output.

Apologies to everyone for the wild goose chase! But I have to say not sure how long it would have taken to figure this one out without all the great comments and feedback. With the pandemic shutting so many things down, forums are one of my only resources for help with this project so much gratitude for everyone's input!!
 

Online xavier60

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: au
Re: Audio power amp has large DC offset in output - SOLVED!
« Reply #36 on: June 06, 2021, 01:26:30 am »
 :)
Hioki AS100D vom, HP 54645A dso, Fluke 87V dmm, AN8008 dmm, Agilent U8002A psu,  FY6600 function gen, New! Brymen BM857S-(With Battery)
 

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #37 on: June 06, 2021, 01:32:37 am »
Neither signal ground nor input are connected to anything for both channels.

 :o :o :o

The signal ground should be connected to the power ground at all times!  Didn't you respond to madires and show a diagram of them connected that way?

And both of your C4 capacitors may have issues, they really need to be bipolar.  I suspect they leak at working voltages.

yes, I did. But the design uses a star ground at the power supply board and separate ground planes for the power and signal sections of the amp boards. The power section was connected but I made a classic noob mistake and failed to register that the signal ground was isolated from the power ground on the amp board.

As for the caps, I've seen bipolar input caps in different designs as well. I built this one from a design I found here: https://www.circuitbasics.com/design-hi-fi-audio-amplifier-lm3886/. I'm just recently graduated with an EET degree but I'm really new to this and I know enough to know how much I don't know. The power amp has worked really well as is and is very quiet so I'm inclined to let it be without a compelling reason to change the input caps. I hope that makes sense. I appreciate your input!
 

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output - SOLVED!
« Reply #38 on: June 06, 2021, 01:43:21 am »
In your schematic, I see missing resistors for the pre-amp section. U4B, U4C (+) inputs need them for input bias current of the op-amp and capacitor leakage current. C10, C11 33uF is kinda huge there.
This could make mystery DC offset going into the tone controls as well, and getting out to the power amp.

Wow...thanks for the observation. In fact I do have an issue with the preamp that is next on my list of problems to tackle. This is v2 and I was just getting started on troubleshooting when the rather severe power amp issue arose. So now that that is resolved I'm getting back to the preamp and will make a new post about that. But the first major issue is that the 2.5 VDC input offset is not being filtered by C10 and C11. I breadboarded the input offset circuit ( I found this configuration in Maxim app note) and it worked perfectly with a unity gain buffer so not sure what is missing.  It sounds like you have an idea about what's not right with my design. I should probably continue the conversation in a new posting but if you wouldn't mind could explain further what you mean by "missing resistors"?
 

Online bdunham7

  • Super Contributor
  • ***
  • Posts: 2632
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #39 on: June 06, 2021, 02:11:55 am »
As for the caps, I've seen bipolar input caps in different designs as well. I built this one from a design I found here: https://www.circuitbasics.com/design-hi-fi-audio-amplifier-lm3886/. I'm just recently graduated with an EET degree but I'm really new to this and I know enough to know how much I don't know. The power amp has worked really well as is and is very quiet so I'm inclined to let it be without a compelling reason to change the input caps. I hope that makes sense. I appreciate your input!

OK, so let me explain what the problem is in detail and maybe you'll change your mind.  This design is one where you can 'get away with' the error in the cap selection under most circumstances.  However, it looks like the differing responses to the lack of signal ground were likely at least in part due to leakage in C4.  In particular, there appears to be no other way to get that -19.5 volts on the + input without C4 leaking.

Audio blocking capacitors usually are biased if they are polar--that is there is a constant DC voltage on them that exceeds the p-p audio signal.  In this case, it is not biased and is subjected to whatever the amplifier output is, less what is attenuated by the feedback voltage divider resistors (R3 + R4).  Polar electrolytics can withstand a volt or so in reverse bias without any immediate damage or gross nonlinearity, so at normal audio frequencies where the capacitor is relatively low impedance, the voltage across it is low even at higher output levels.  At low frequencies--very low in this case with such a huge capacitor, perhaps at the level of rumble from a phonograph--the capacitor's impedance starts to go up and the voltage across it increases.  At DC, it is exposed to the full output voltage, which if it happens to be the wrong way is damaging.  The only reason it won't pop is because of R3+R4 limiting the current.  As I said, you might get away with it for a while, but any leakage will increase the DC gain, which isn't a good thing.

The size of the capacitor seems audiophoolish to me.  A 47uF capacitor should give you a -3dB point of about 3 Hz.
A 3.5 digit 4.5 digit 5 digit 5.5 digit 6.5 digit 7.5 digit DMM is good enough for most people.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 4470
  • Country: ca
Re: Audio power amp has large DC offset in output - SOLVED!
« Reply #40 on: June 06, 2021, 02:18:20 am »
if you wouldn't mind could explain further what you mean by "missing resistors"?

OK that would look like a resistor to GND at each of those pre-amp op-amp's (+) input. Just check textbooks and such about op-amp non-inverting configuration, they will show those resistors because otherwise the JFET op-amp's input is floating (for DC) there is no return path for the input JFET's gate current there, and it will cause offset trouble.
So U4C pin 3, 10k-100k resistor to GND, and U4B pin 5, 10k-100k resistor to GND.
C10, C11 do not need to be so big and are best a film cap for fidelity instead of electrolytic. I would use 1uF with 47k there. For now, just try add the two resistors and that should fix up the 2.5V issue.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 4470
  • Country: ca
Re: Audio power amp has large DC offset in output
« Reply #41 on: June 06, 2021, 03:06:15 am »
[...] The size of the capacitor seems audiophoolish to me.  A 47uF capacitor should give you a -3dB point of about 3 Hz.

It's been proven that a much larger capacitor there (NFD divider) gives slightly lower distortion. It was in Silicon Chip magazine, I can dig for it but they went huge I think 1,000uF.
I thought because it's part of the NFB divider network, non-linearities and dielectric absorption get magnified.
« Last Edit: June 06, 2021, 03:07:51 am by floobydust »
 

Online bdunham7

  • Super Contributor
  • ***
  • Posts: 2632
  • Country: us
Re: Audio power amp has large DC offset in output
« Reply #42 on: June 06, 2021, 03:14:21 am »
It's been proven that a much larger capacitor there (NFD divider) gives slightly lower distortion. It was in Silicon Chip magazine, I can dig for it but they went huge I think 1,000uF.
I thought because it's part of the NFB divider network, non-linearities and dielectric absorption get magnified.

I don't doubt that non-linearities in polar elcaps would be reduced by lowering the peak reverse voltage across them, but did they compare to something like a 22uF polypropylene film cap?
A 3.5 digit 4.5 digit 5 digit 5.5 digit 6.5 digit 7.5 digit DMM is good enough for most people.
 

Offline hummusdude

  • Regular Contributor
  • *
  • Posts: 113
  • Country: us
Re: Audio power amp has large DC offset in output - SOLVED!
« Reply #43 on: June 06, 2021, 05:35:21 am »
if you wouldn't mind could explain further what you mean by "missing resistors"?

OK that would look like a resistor to GND at each of those pre-amp op-amp's (+) input. Just check textbooks and such about op-amp non-inverting configuration, they will show those resistors because otherwise the JFET op-amp's input is floating (for DC) there is no return path for the input JFET's gate current there, and it will cause offset trouble.
So U4C pin 3, 10k-100k resistor to GND, and U4B pin 5, 10k-100k resistor to GND.
C10, C11 do not need to be so big and are best a film cap for fidelity instead of electrolytic. I would use 1uF with 47k there. For now, just try add the two resistors and that should fix up the 2.5V issue.

I added a 10k resistor as suggested and it fixed the offset as expected. One channel is working very well at the gain stage (U4B) but the other has a 481 kHz oscillation at the output (TP3). I've disconnected the tone stage pots to try to isolate the issue. Whatever is causing the oscillation (maybe some interference from a power rail?) is only on one channel. I will start a new thread for the preamp problem and add details and scope shots but wanted to confirm your solution and say thanks for all your help!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf