What to use to measure amps THD?

[Mike82]

I was looking to test my amps THD and was wondering what type of hardware would I need to do this. Any ideas?

[Bear]

This would not be an inconsiderable undertaking if you want to do it well. You want a stable resistive load and an oscilloscope.

[---k---]

This thread shows how Chuck did his testing:

http://www.avsforum.com/avs-vb/showthread.php?t=855865

[5th element]

A high performance sound card + ARTA and a resistive load are all you really need.

You can typically measure to 0.0003-5% if you use the full input bandwidth of good modern sound cards.

Something like this

http://uk.farnell.com/tyco-electronics-cgs/hsc1004r7j/resistor-100w-5-4r7/dp/1174286

is all that's required for the load. I've got 4 of them mounted on a heatsink, this way I get 9.4, 4.7, 2.35 and 1.18 ohm loads at enough power rating for some pretty serious powers.

[mazurek]

You have to be careful about the voltage levels, but you can even use a resistive divider into a sound card input.

[5th element]

You have to be careful about the voltage levels, but you can even use a resistive divider into a sound card input.

Indeed, this is what I do, luckily if I did manage to fry the input of the sound card I'd be able to replace whatever had blown up.

[Dr.EM]

This may be misguided but I would assume that by using a PC based measurement set-up it'd be possible to attain the distortion profile of the soundcard by taking a measurement of itself linked back to its own inputs. Then by using this as a calibration file one could null any inherent distortion, thus giving even more low-level measurement resolution.

Possible? Necessary? I don't know, it seems possible, at least for taking measurements at just 1kHz for example.

[5th element]

This may be misguided but I would assume that by using a PC based measurement set-up it'd be possible to attain the distortion profile of the soundcard by taking a measurement of itself linked back to its own inputs. Then by using this as a calibration file one could null any inherent distortion, thus giving even more low-level measurement resolution.

Possible? Necessary? I don't know, it seems possible, at least for taking measurements at just 1kHz for example.

Indeed I have seen this done in one of the Hifi News magazines I've got. They use a Lynx sound card for that and modify the test signal to cancel out the effect of the sound card itself. The only trouble I can see with this is at the distortion levels we're talking about, you only need an extremely small deviation of error within the modified test signal to completely botch things up.

Of course I am talking about turning 0.0003% into something considerably smaller, here there might not be much real benefit from such a thing, but if your system was working to something like 0.005%, improving this to 0.0003% would certainly be beneficial.