Measuring THD with Keithley 2015

[belgianbrain]

So I have a Keithley 2015 THD analyzer, which I am trying to use to set the bias on a Yamaha B2.

The Keithley, with internal sine generator connected straight from output to input, gives about 0.01% THD. Even the manual suggests this number will be about 0.02%. Is it me or does this suck? For some frequencies and magnitudes, the Keithley actually does approach 0.02% THD when simply reading itself.

Now, the Yamaha AMP is supposed to achieve 0.005% at most sane volume levels. So I hook up the Keithley output through my Adcom GFP-750 (set in passive mode, of course, so I can adjust volume and hit 30W continuous output according to the Yamaha Service Manual), to the Yamaha, and back to the Keithley, and get about 0.013% THD. I can send it 1Khz, 10Khz, 20Khz, anywhere from 0.5V to 2.0V. Problem is, the Yamaha never seems to output THD number that is any different from what the Keithley reads when looped back on itself. In other words, the Yamaha is adding no significant measurable distortion, according to the Keithley. I can sweep the bias from 40mV to 100mV, and nothing really changes. It makes it hard to set the bias for lowest distortion when the Keithley doesn't seem to be able to measure the change.

Am I doing something wrong here, is the Yamaha amp really that good, or is the Keithley simply that crappy?

For comparison purposes, if I flip the GFP-750 into active mode, THD jumps to about 0.6% to 0.9% - huge distortion from the Adcom, but perhaps that's a problem for another day.

 

[MattiaM]

I believe the problem is the internal signal generator, which does not have particularly low distortion. The specs indicate that the distortion of the signal generator (1V rms output) is -64dB, or about 0.09% THD.
In fact, to check that the distortion meter works correctly, the manual requires thje use of an external low distortion signal generator:
1) Connect a low-distortion function generator to the Model 2015 INPUT jacks.
2) Set the function generator to output a 1kHz, 0.95V RMS sine wave with an unbalanced, high-impedance output.
3) Using the MEAS key, set the following operating modes: TYPE: THD; FREQ: AUTO; UPR HARM: 10; UNITS: PERC; SFIL: NONE
4) Select the Model 2015 THD function by pressing SHIFT then THD
5) Use the down RANGE key to select the 1V range
6) Verify that the Model 2015 THD reading is <0.004%

It's not the accuracy of the THD measurement of the Keithley that's crappy, it's the signal it generates that isn't good enough for measuring the THD of your amplifier.

 

[MattiaM]

You will struggle to find a good function generator with distortion lower than 0.04% without spending a fortune though

 

[ConradH]

AFAIK, the Keithley isn't "crappy", it's just not up to what low thd hifi amps do these days. As said above, a better signal source should improve things greatly. If you can't find a reasonable generator, possibly a CD player would do the job. Burn yourself a test CD with the desired signals using Audacity. A decent CD player should be better than your amp. Bias is usually set down at 1W output or less, or you won't see the result of your adjustments. It also gets worse at low impedances, so I often set it with a 4 ohm load. Open circuit you may not see anything at all. IMO, what you really want is a thd analyzer that has an output connector so you can see the residual on a scope, but the meter should do OK.

edit- FET amps are a bit different, so yours gets adjusted at 30 watts output and 20 kHz apparently.

 

[belgianbrain]

Tried downloading a 24bit tone generator app and hooked up my Schiit Modi to the Keithley. This thing can generate any audio tone and the Keithley measures 0.001% THD!

The Yamaha B2 actually measures 0.005% THD at 20Khz, 30W continuous output (test conditions as instructed by service manual). At lower volume levels and lower frequencies, it measures closer to 0.001% THD!

 

[KentTeffeteller]

AFAIK, the Keithley isn't "crappy", it's just not up to what low thd hifi amps do these days. As said above, a better signal source should improve things greatly. If you can't find a reasonable generator, possibly a CD player would do the job. Burn yourself a test CD with the desired signals using Audacity. A decent CD player should be better than your amp. Bias is usually set down at 1W output or less, or you won't see the result of your adjustments. It also gets worse at low impedances, so I often set it with a 4 ohm load. Open circuit you may not see anything at all. IMO, what you really want is a thd analyzer that has an output connector so you can see the residual on a scope, but the meter should do OK.

edit- FET amps are a bit different, so yours gets adjusted at 30 watts output and 20 kHz apparently.

A major great point in this post. The higher quality equipment requires top of the line test equipment in top calibration to really measure it's performance on the bench. And always has. A big reason why some general TV/Radio shops might not be the best to work on higher end gear. Be mindful of this. ConradH is very right.