advertisement


Problem with DC offset

ToTo Man

the band not the dog
I've just discovered my DAC is producing DC at its single-ended outputs, 1.61V on the Lch and 1.01V on the Rch. The BAL outputs are fine, Pins 1 & 2, 1 & 3, and 2 & 3 all measure 0.00V.

I only discovered this by chance as I'm currently demo'ing a headphone amp that has an LCD display that warns when there is DC present at an input and shuts off that input as a precaution. Testing for DC voltage from a source isn't something that's been on my radar; amplifiers yes but not sources....

I'm still investigating the source of the DC, i.e. whether it's only from the DAC's USB input or whether it's present across all the DAC's inputs**. All I know so far is that the DC voltage is only present when the DAC is locked onto a source.

I've now tested the coax and optical inputs and the DC is present on these also.

What's of particular concern is the amp I've been using to perform all my loudspeaker bench-testing has been driven from this DAC's single-ended outputs. I always check the amp's speaker terminals before every test session to ensure there's no DC, and I always get readings below 2mV. However, I do not perform this amp test when the DAC is locked onto a source, hence my unawareness of the issue!

The DC offset at the amp's speaker terminals when the DAC is engaged is 140mV on the Lch and 90mV on the Rch, so fortunately the amp appears to be blocking just over 90% of the DAC's DC. I am concerned however that 90mV and 140mV is still quite high to connect to a loudspeaker driver without a crossover, particularly a low-power handling tweeter, if it's subjected to it for a prolonged period of time. Am I right to be worried?

Coincidentally, for the tweeter tests I've performed recently I've connected a 22uf Mundorf bipolar Ecap in series with the amp's output as a safety measure. However I've just measured the DC voltage with this cap in the circuit and it only reduces it from 140mV to 50mV, so it's not really acting as a "DC blocker" but more of a "DC attenuator", and a pretty mild one at that. Is this the behaviour you'd expect from a 22uf bipolar cap?
 
Last edited:
How did you measure the dc. If you just put a DMM on the output, leakage can give non zero readings.
What is the voltage into a 47k load, a fairly typical preamp input?
 
Not just leakage! Into an infinite resistance a cap will show the same DC voltage on both ends.... it needs to charge into a load.
 
How did you measure the dc. If you just put a DMM on the output, leakage can give non zero readings.
What is the voltage into a 47k load, a fairly typical preamp input?
I put a multimeter across the tip and ring of each RCA cable coming out of the DAC. How do I measure into a 47k load, do I need to buy a 47k resistor for this?
BTW - I've asked a couple of other owners of this DAC to measure the DC at the RCA outs and they are reporting readings of 0.00V, - I'm assuming they'll be using the same measuring technique.
 
Not just leakage! Into an infinite resistance a cap will show the same DC voltage on both ends.... it needs to charge into a load.
I don't understand. Are you saying I can't measure DC output from the DAC by simply putting a DMM across the RCA out?

EDIT - Apologies, I see you're referring to my question about the cap I put inline with the tweeter.
 
I don't understand. Are you saying I can't measure DC output from the DAC by simply putting a DMM across the RCA out?

Correct. If it's DC coupled then you can and the result will be accurate. Many cap coupled items have a resistor to ground after the output cap for just this reason, plus that if you switched on with nothing connected and then did a silly thing like plugging in the RCA plugs from a switched on power amp (asking for an accident anyway!) you would get one hell of a thump!
The input resistance of the DMM should slowly remove the offset over a few minutes... bar the leakage...

My latest power amp is cap output and I've put 2k2 to ground from the outputs to gradually remove the offset so that if it is switched on without speakers connected it will go down to zero over a few minutes. Otherwise about 8V will be just sitting there waiting for the speakers to be connected... that's not going to damage them but will cause one hell of a crack accompanied by a briefly bottomed out bass unit. It would be just for a second ish as the output caps charge.
 
I'll need to order one as I don't have any resistors bar some 4R, 6R and 8R I bought to use as dummy loads. Could you point me to a suitable 47k resistor?

Literally anything will do and of any value from about 10K to 100K. As I said above just the DMM should gradually do it due to its input resistance. If you see the offset slowly falling then probably all's well.
 
Ok I've connected my DMM to my DAC's RCA output cable with croc leads and have been watching the reading, over a period of 10 minutes it has increased very slightly from 1.597V to 1.601V.
 
Ok I've connected my DMM to my DAC's RCA output cable with croc leads and have been watching the reading, over a period of 10 minutes it has increased very slightly from 1.597V to 1.601V.

That's not looking good then. If output cap is electrolytic then the leakage could hold it there though.
 
Reading after 30 minutes is now 1.606V.

I've emailed the UK service agent and he thinks the servo has gone so wants me to send it in for repair.
 
Servo in a dac with balanced outputs? Likely not using the dac chip/s in actual balanced configuration then.
 
Not just leakage! Into an infinite resistance a cap will show the same DC voltage on both ends.... it needs to charge into a load.

I connected a Celestion HF2000 tweeter to the in-line 22uf cap and remeasured the DC offset. DC on the 'amplifier end' of the cap is 140mV, DC on the 'speaker end' of the cap is 0.0mV, so the cap is working properly!

Would 140mV DC be enough to damage a tweeter with tiny Xmax like an HF2000? I posted recently about an HF2000 I bought which, when installed in a Ditton 66 loudspeaker, would suddenly 'come to life' and jump +5dB in output once I reached a certain voltage level. Yet when I measured it out of circuit and connected directly to the amplifier's outputs its output level increased linearly as I turned the voltage up and did not suddenly jump in level. I now wonder whether the amplifier's 140mV DC offset was assisting the tweeters motion by forcing the diaphragm and voice coil assembly into a more forward position?
 
I connected a Celestion HF2000 tweeter to the in-line 22uf cap and remeasured the DC offset. DC on the 'amplifier end' of the cap is 140mV, DC on the 'speaker end' of the cap is 0.0mV, so the cap is working properly!

Would 140mV DC be enough to damage a tweeter with tiny Xmax like an HF2000? I posted recently about an HF2000 I bought which, when installed in a Ditton 66 loudspeaker, would suddenly 'come to life' and jump +5dB in output once I reached a certain voltage level. Yet when I measured it out of circuit and connected directly to the amplifier's outputs its output level increased linearly as I turned the voltage up and did not suddenly jump in level. I now wonder whether the amplifier's 140mV DC offset was assisting the tweeters motion by forcing the diaphragm and voice coil assembly into a more forward position?

Yep it's doing what a capacitor does.

Whilst 140mV is bugger all a tweeter never see's DC in normal operation as a high pass filter has a capacitor in line at some point.
 
If the DAC is working with a DC offset, then a cap will be needed at the output to block it. Assuming that the DAC used some other means to get rid of that DC (no DC should be present on an output, or input beyond a few mV) then something is gone awry.
 


advertisement


Back
Top