dc offset question

[gh0st]

I've not managed to find a consensus on this, perhaps someone can chime in: is it more relevant (as in: will the amp damage my headphones) to measure the dc offset with or without a headphone load connected?

Example: my crack (+speedball, film output caps) measures around a meandering few +/- mV unloaded, sometimes hitting 10-20mV and, with my DMM set to record min/max/avg over ten minutes or so, a peak of -340mV, although this was definitely a spike, it didn't slowly climb to, or fall from this value.

With a 600ohm load connected, it's typically <1mV, peak of 5mV.

All measurements taken after the amp has warmed up for over a minute or two.

Which is the figure to be more confident with??

[Laudanum]

I think there is surely more confidence in the measurement taken with phones connected, definitely over the unloaded amp but also over a 600 ohm "purely" resistive load as well I would think.  
The turn on voltage is also much lower with phones connected compared to the 9 Vdc as stated in the manual or the higher voltage that is measured with film caps (and speedball?).  Im sure you will find that to be the case as well with phones connected and checking DC offset (which I think is leakage) compared to no load, although maybe a bit different compared to the 600 ohm measurement.
I dont think you would risk damaging your phones doing your measurements with them plugged in.  But more informed replies should confirm or correct me on any or all of the above and we will both learn something with more certainty.

[gh0st]

I should have been slightly clearer - when I said 600ohm load, I meant 600ohm headphones rather than a resistor

[Paul Birkeland]

Hello gh0st,

I don't believe the real questions is whether the actual act of measuring will potentially damage your headphones, but rather what a slip of the hand could end up doing.  If you accidentally shorted the output capacitor while your headphones were plugged in, that would be an unhappy event, so I would recommend buying a set of 600 ohm 1w resistors and clip leading them into the circuit to perform such measurements.

-PB

[gh0st]

thanks Paul!

I really worded my original post poorly though, I've already measured the offset with and without my 600ohm headphones without catastrophe, what I really meant to ask was:

are a few mV with the headphones plugged in a cause of concern about longterm damage of the headphone?

I'm thinking not, but because without the 'phones plugged in the results are a bit higher, and hit transient peaks in the 150-300mV range, I felt like sanity checking before I pull the trigger on some Beyer T1s to do justice to the Crack

[Grainger49]

   .  .  .   what I really meant to ask was:

are a few mV with the headphones plugged in a cause of concern about long term damage of the headphone?

The answer is, NO!  A few mV of leakage will dissipate across the headphone impedance and will be gone after a few seconds.  I don't expect it will continue.  And maybe I'm wrong.  But it won't affect your headsets.

[gh0st]

it doesn't go away as such (the couple of mV with the 'phones plugged in remains constant over time) But I get the general message

[Grainger49]

That is leakage through the electrolytic caps.  DC will dissipate across the ~2500 ohm resistor and the parallel combination of the headsets.  Again, it won't affect your headsets.

[Paul Birkeland]

It is also important to remember that a given DVM is going to have some tolerance in terms of what it is measuring that may account for some of what you are seeing.

There is some good reading on this available here.

Also, if we assume that there is indeed 5mv of DC voltage at the output, we can use ohms law with a 600 ohm load to approximate the heat this is dissipated by the headphones: P=(V^2)/R or (0.005^2)/600=0.0000000416 watts.  Something like an HD-800 will handle 0.500 watt maximum, so this additional heat is insignificant.

Also, the presence of a negative DC offset makes me wonder if there is signal playing while doing this measurement?

[gh0st]

thanks for the comprehensive answer - nothing was connected to the inputs at the time of measurement.