How much DC offset is coming out of your amp?

[DocG]

I'm a little late to the party, but thanks for pointing me to this issue, Vlad. I use an active set-up, and my power amps are DIY: two good reasons to check this DC thing ASAP.

And it's not as good as I had hoped...

I've built two stereo units. One unit gives 7,2 and 9,4 mV respectively, but the other one reads 19,6 and 20,0 mV.

When building the units, I used one case, one power supply and two NC-400 amplifier modules for each; I split the wiring from PS to amps (artisanal cutting and soldering together, connection protected with a heat shrink).

I never noticed any sound quality issues when listening these past few months -- all crystal clear and tight...

Any suggestions for solving this issue? What to look for when opening the case?

If the amplifier has DC offset setting on the PCB (small turn pot), you simply hook up your multimeter as before and adjust the pots to get close to 0mV DC reading. I see three blue trim pots on pics of the Hypex NC-400. Can you find out which one is for biasing and which for DC offset?

I’ll have to dig up the manual to check that. But I’ll start digging right away! Thanks!

(btw, the speaker wire is not properly twisted on that picture!)

 

[Vladimir]

I’d invalidate all warranties if I started doing that, haven’t had any issues I don’t think, so sensibly I’d have to leave it.

Have you measured the DC offset on the speaker terminals? Simple measurement doesn't void warranty and it's good to know if you have a healthy amp. If the readings are no good, you send it to the manufacturer and they'll adjust it for you.

 

[DocG]

But I recall having read something about it on the diyaudio forum, and a quick google search taught me that I'm looking for "R136" (amp offset) and "R95" (input bias compensation). And this picture shows them nicely (at the top)!

So I know how to try and make it better... Having said that, on the diyaudio forum, people complain about modules with 500 mV offset and they mention that the pots are very sensitive around zero. So I hope I don't make it worse when trying...

If this turns out well, I owe you, Vlad! Especially since I just got me a pair of Maggie .7s to replace the MMGs. Mustn't think of frying the .7s! *shok*

 

[Gray]

I used to measure this as part of the final checks on amps when I briefly worked at Rega.

On my Cyrus 8 (purchased around 2002):

July 2012: L 9.8 R 6.8

Today: L 16.4 R 10.6mV

 

[Vladimir]

I used to measure this as part of the final checks on amps when I briefly worked at Rega.

On my Cyrus 8 (purchased around 2002):

July 2012: L 9.8 R 6.8

Today: L 16.4 R 10.6mV

Looks good. It will soldier on.

 

[Vladimir]

But I recall having read something about it on the diyaudio forum, and a quick google search taught me that I'm looking for "R136" (amp offset) and "R95" (input bias compensation). And this picture shows them nicely (at the top)!

So I know how to try and make it better... Having said that, on the diyaudio forum, people complain about modules with 500 mV offset and they mention that the pots are very sensitive around zero. So I hope I don't make it worse when trying...

If this turns out well, I owe you, Vlad! Especially since I just got me a pair of Maggie .7s to replace the MMGs. Mustn't think of frying the .7s! *shok*

Trimp pots of that type are all touchy, use minute turns. Wait for the amp to stabilize sitting idle for 15min first.

You can't ruin anything with the DC offset pot, but you can ruin your amp with the Bias pot.

I'm thinking class D shouldn't have biasing, but I'm speaking general caution. So make sure you got the right one. Mark your current positions with a felt pen.

 

[DocG]

But I recall having read something about it on the diyaudio forum, and a quick google search taught me that I'm looking for "R136" (amp offset) and "R95" (input bias compensation). And this picture shows them nicely (at the top)!

So I know how to try and make it better... Having said that, on the diyaudio forum, people complain about modules with 500 mV offset and they mention that the pots are very sensitive around zero. So I hope I don't make it worse when trying...

If this turns out well, I owe you, Vlad! Especially since I just got me a pair of Maggie .7s to replace the MMGs. Mustn't think of frying the .7s! *shok*

Trimp pots of that type are all touchy, use minute turns. Wait for the amp to stabilize sitting idle for 15min first.

You can't ruin anything with the DC offset pot, but you can ruin your amp with the Bias pot.

I'm thinking class D shouldn't have biasing, but I'm speaking general caution. So make sure you got the right one. Mark your current positions with a felt pen.

After having the amps on for over half an hour, I started tinkering with "R136". I finally managed to get a DC offset of < 1 mV on all four channels. Left it on for a while: all stayed stable at < 1mV. Bolted the top plates on again, and remeasured and the readings went from 3 to 19 mV! Opened the lid again, and all was <1 mV again! And then I noticed that tightening/loosening bolts changed the offset!

I ended up with offsets of 1, 4, 6 and 10 mV, which is quite acceptable I guess. But I really don't get what's happening here...

 

[Vladimir]

Maybe the cable loom is tugging too hard.

Probe stuff with the plastic handle of your scredriver. Tap on things, see how the measurement changes on the multimeter. If for example it shoots up when you tap on the cable loom, then that's the trouble spot.

Also try tapping or lightly flexing the case while measuring.

 

[Vladimir]

But I recall having read something about it on the diyaudio forum, and a quick google search taught me that I'm looking for "R136" (amp offset) and "R95" (input bias compensation). And this picture shows them nicely (at the top)!

So I know how to try and make it better... Having said that, on the diyaudio forum, people complain about modules with 500 mV offset and they mention that the pots are very sensitive around zero. So I hope I don't make it worse when trying...

If this turns out well, I owe you, Vlad! Especially since I just got me a pair of Maggie .7s to replace the MMGs. Mustn't think of frying the .7s! *shok*

Trimp pots of that type are all touchy, use minute turns. Wait for the amp to stabilize sitting idle for 15min first.

You can't ruin anything with the DC offset pot, but you can ruin your amp with the Bias pot.

I'm thinking class D shouldn't have biasing, but I'm speaking general caution. So make sure you got the right one. Mark your current positions with a felt pen.

After having the amps on for over half an hour, I started tinkering with "R136". I finally managed to get a DC offset of < 1 mV on all four channels. Left it on for a while: all stayed stable at < 1mV. Bolted the top plates on again, and remeasured and the readings went from 3 to 19 mV! Opened the lid again, and all was <1 mV again! And then I noticed that tightening/loosening bolts changed the offset!

I ended up with offsets of 1, 4, 6 and 10 mV, which is quite acceptable I guess. But I really don't get what's happening here...

That is some wierd voodoo going on. First time I hear its happening on an amp. Maybe the whole case is flexing including the boards when you reassemble it. They are probably also double sided PCBs.

Look for some solution that makes any flexing of the case not affect the board. IDK, thick silicone washers? Put your innovator hat on.

 

[DocG]

Maybe the whole case is flexing including the board when you reassemble it.

No, the board is on a thick aluminum heatsink. That cannot be flexed together with the case...

Look for some solution that makes any flexing of the case not affect the board. IDK, thick silicone washers?

Even if the board were affected by flexing of the case, the aluminum case itself works as part of the heat sink. That's why the modules are so tightly bolted to the case. Isolating them with silicone washers will shorten their life span...

Maybe the cable loom is tugging too hard.

No stress on the cable loom whatsoever.

Probe stuff with the plastic handle of your screwdriver. Tap on things, see how the measurement changes on the multimeter. If for example it shoots up when you tap on the cable loom, then that's the trouble spot.

Also try tapping or lightly flexing the case while measuring.

Yeah, I might try that. Later.

As there's no more offset than 10 mV, I just put everything in place again this morning. It's definitely better than it was anyway.

I think I'll measure again in a month or so. If it gets worse, I'll open it all up again. I'll keep you posted. Thanks again for educating me, Vlad!

 

[Vladimir]

You are right, leave it as is. I'm just perplexed by the wierd behavior.