(Fixed?!) source of dc offset or why it's not being filtered (Apple HomePod)

[Drnict63]

Hello! I'm trying to crack a tough nut that's been plauging many other Apple HomePod owners. For some background, the homepod is a wireless speaker with only a power cable. No inputs, and everything from the power supply, to amp, to logic is all contained inside the unit. After ~8 minutes of idle, the HomePod will go into a low power standby mode and turn off some components. Any action wakes it up out of standby mode with everything fully powered again. The transition is almost instant.

The common problem is a dc offset voltage being introduced somewhere along the signal for the subwoofer. When measuring the subwoofer speaker output terminals, I typically see somewhere around ~200mV of dc give or take a few 100, and it steadily rises a few mV every second over time. The dc offset is only present at the subwoofer when the HomePod is not in standby...

I found some test pads on the amp board (first phot0) inbetween the amplifier IC for the subwoofer and the DAC. These appear to feed directly into the channel inputs on the amp IC. When the homepod is not idling, the same dc offset is measured on these pads, however, when the homepod idles, unlike the subwoofer terminals, I can still see the dc offset voltage on these pads and the dc is slowly falling in mV until the homepod is awaken again.

I was chatting with a redditor who was successful in resolving the dc offset by replacing the two capacitors (circled in red in second photo). I measured these as 10uF capacitors, they said they replaced them with 0.1uF capacitors?? I don't have any 0.1uF capacitors small enough to try the same yet, but I tried other 10uF capacitors and it didn't ahve an effect, so I think there's a different cause (at least for mine and many others)

I have a few of the amplifier boards for testing. I've depopulated one in an attempt to try and trace out some of how the board is layed out and been googling common causes for dc offset in audio amplifiers. I am still very much a beginner at all this and not sure what could be causing it or where to start, please go easy on me ;-;

Third and fourth photo is a high quality image of the amp board from the homepod from ifixit.

Datasheet for essentially the same amp IC chip: https://www.infineon.com/dgdl/ir4302.pdf?fileId=5546d462533600a4015355d602a9181d

Datasheet / schematics for an evaluation board featuring the same amp IC with a lot of similarities to the homepod amp board, if it helps?  https://www.infineon.com/dgdl/iraudamp18.pdf?fileId=5546d462533600a40153569a96eb2be9

Datasheet for DAC: https://d3uzseaevmutz1.cloudfront.net/pubs/proDatasheet/CS4350_F4.pdf

iFixit's identified components on the homepod amp board: https://www.ifixit.com/Teardown/HomePod+Teardown/103133#s192623

[fzabkar]

The DC offset spec in the datasheet is +/- 18mV between the amplifier inputs. As far as I can tell, the gain is between 15 and 16, so this would result in an offset at the output of +/- 270mV. Your circuit may have a different gain.

Can you find some way to measure the voltage differential between IN+2 and IN-2, or IN+1 and IN-1 (pins 3,4,8,9)?

[Drnict63]

Thank you for replying! Seems like I have some more reading to do.

The pads I circled in the second photo seem to have direct continuity to pins 3,4,8, and 9 on the amp IC pads on my depopulated board, so I would guess that means I can use these to take measurements?

I set my multimeter to read "200m"V scale and see 3.5mV between pins 3 and 4. I see 4.9mV across pins 8 and 9. In my original post when I said "I see the same dc offset on these pads" I must have been mistaken and switched the channels up, between IN+1 and IN-2 for example, I see around ~200mV and rising when the amp is on and falling when off.

Is that all expected?

[fzabkar]

I would wait for the audio experts to give their opinions, but it seems to me that the offset is inherent in the amplifier and is within spec.

[EHT]

The common problem is a dc offset voltage being introduced somewhere along the signal for the subwoofer.

Do you mean this is a known common fault in this unit? You could Google to see what level of DC voltage a protection circuit would typically trip at. You could measure the DC resistance of the speaker and work out how much power this voltage translates to. This may be tricky to find unless you can spot damage or find a component that measures out-of-spec (i.e. diode, transistor, resistor, capacitor with DMM).

[Drnict63]

It's an issue that affects quite a large number of units, not all of them but a lot! https://youtu.be/lptLswqx-Sc
Luck may very well be on my side!

I was reading this: https://classicalcandor.blogspot.com/2019/10/on-power-amplifier-dc-offset.html

"...Since DC offset will increase as operating temperature rises"

Eureka! I connected my multimeter to the subwoofer terminals, and started heating up small areas of the amplifier board until I saw a very rapid spike in dc offset from ~300mV to well over 2v!! Blowing on the same area rapidly decreased it again...  https://youtu.be/tfJc6xMt7ZE

It's coming from the DAC or somewhere very, very near it. This has drastically reduced the amount of guess work and hopefully I'll be back soon again to post the solution

[Drnict63]

Placed an order for various capacitors and waiting for them to arrive in a few days...some 0.1uF, 2.2uF, 10uF and 11uF.

With the 0.1uF I wanted to use to try and reproduce the success one redditor said they had (in my OP) replacing the two 10uF capacitors directly under the amp IC. Not sure that will work, I already tried this with some other 0.1uF caps I harvested from a dead GPU but I'll still try with new ones "just in case"

The 2.2uF ones I'm getting are from an idea I got from chatting with another redditor (the same person who made the youtube video I linked who originally diagnosed where the popping sound was coming from, dc offset). They said 2.2uF or 4.7uF capacitors are usually used in series with the input signal into the amp, and reviewing the schematics for the evaluation amp board featuring the same amp IC I spotted four 2.2uF capacitors in series with the audio input signal. These capacitors still measure 2.33uF with my multimeter when removed from the board, can they still be bad?

The 10/11uF ones are what appear to be the capacitors that cause the dc offset to greatly increase / decrease with temperature right after the DAC. Hopefully it's these. Hopefully it's not one of the dozens of super tiny caps / resistors around the DAC, or the DAC itself because I cannot find a replacement CS4350 QFN DAC : (

So I'm unsure if I should be chasing the cause of the dc offset or the failure of filtering it? In any case if no one knows I will be testing more when parts arrive!

I've been slowly but surely trying to map out how everything is connected on this homepod amplifier board by using my multimeter and measuring resistance / capacitance of components as I pull them off, comparing it to the evaluation board schematics, and using continuity tracing to figure out what connects to what. I'm starting to get an understanding of how the audio signal flows through the board, from digital to analog to amplification to the woofer. Kinda confused why when I measure capacitors that all should be the same value I see variations even when I go back and re-measure the same capacitor (for example I'll measure 10.01uF one minute then a few minutes later double check and it's 11.3uF and they're not on the board). Maybe that's normal with my cheap meter up to a 10% tolerance could explain seeing between 10uF and 11uF, derp!

[Drnict63]

Update: We fixed it!

These four capacitors pulled off popping boards measured around ~13uF and were very sensitive to dropping capacitance with heat. Pulling the same caps off good boards shows ~10uF

When I replace these four capacitors with new 10uF ones, the dc offset almost entirely vanishes and the subwoofer no longer pops when I unplug it and plug it back in while powered on, confirmed on a couple affected boards with a couple hundred mV of offset that it's down to 0-2mV!!

What seems strange still is even with no more dc offset, when the homepod is first powered on and woken up on some units you can still hear the faintest "pop". I can at least confirm this is not coming from any dc offset because I don't hear any popping when disconnecting and reconnecting the woofer, so I guess this is just normal operation and maybe a result of these capacitors actually doing their job. Not sure we'll ever know without schematics or more hours grinding at this.

I made a new YT video on my channel for those interested in watching the repair (mods I apologize if this is not allowed, totally fine with removing);

I went with X7R 1206 10uF caps as replacements.