4
\$\begingroup\$

Im a bit confused about the concept of residual distortion in audio THD analyzers, for example I have a Keithley 2015 analyzer, the data sheet shows that it has a minimum resolution to measure distortion of 0.0001%, but the residual distortion is 0.004%, What exactly is the residual distortion figure?

From what I understand its like the distortion of the instrument itself, so even if the instrument has a minimum resolution of 0.0001% doesnt that mean that the minimum distortion that can be measured is 0.004%? however I tried connecting the output of my SG505 low distortion oscillator to the meter and it reads 0.0008% which is on spec for the oscillator, so what exactly is going on? I've tried looking for this on Google with no luck.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ Residuals are what is left over after your ability to predict and/or explain observation has been accounted and all you have left are (hopefully) stochastic residuals. Suppose you have a voltmeter that can resolve down to a \$1\:\mu\text{V}\$ but where the instrumentation noise is \$\sigma=100\:\mu\text{V}\$. This just means you can "see" the noise taking place with some resolution into it and could, if desired, integrate multiple readings to improve it. Suppose instead the resolution is \$500\:\mu\text{V}\$ and noise is \$\sigma=100\:\mu\text{V}\$. Then you can't see the noise. \$\endgroup\$
    – jonk
    Commented Jan 24, 2018 at 5:57
  • 1
    \$\begingroup\$ @jonk I know you usually write three page answers, but that comment really is complete, and you should not be ashamed posting it as an answer and expand on it later if you have time! \$\endgroup\$
    – pipe
    Commented Jan 24, 2018 at 8:23
  • 1
    \$\begingroup\$ "residual distortion" may be a worst case spec across a specified bandwidth, say 10Hz-20kHz. Even if it is that bad at 20kHz it could well be much lower at, say, 1kHz. \$\endgroup\$
    – user16324
    Commented Jan 24, 2018 at 11:03

3 Answers 3

Reset to default
5
\$\begingroup\$

The residual distortion is basically the analyzer's internal noise floor - you can't measure anything less than that, because all you'd be measuring is the analyzer's internally-generated noise. (Your measurement of .0008% says that their specs are fairly conservative, meaning that .004% is likely a worst-case number).

Think of resolution as being like the number of digits on the display. In the simplest case, .0001% resolution means that you could (theoretically) discern the difference between a .1000% reading and a .1001% reading. It has nothing to do with how low you can measure, but only how many digits will be in the answer you get. The expression "minimum" resolution probably means that this is the best-case resolution, probably available only on one of the lower ranges.

So, they'll guarantee you can read down to .004%, but with .0001% resolution, so the display should actually read .0040%. And you'll be able to see the difference between that and .0041%.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ It isn't only about the internal noise; the residual distortion figure also accounts for any imperfections in the analyzer's ability to cancel out the fundamental signal, as well as any distortion that the input stages of the analyzer itself add. \$\endgroup\$
    – Dave Tweed
    Commented Jan 24, 2018 at 12:32
0
\$\begingroup\$
  • Measurement Input: Residual Distortion THD –65 dB relative to full scale
    • ¹ Input at full scale.
  • Display Resolution 100 dB

It is normal for THD to increase with signal level.
In this case the THD is defined at full scale.

It means -65dB is a worst case value, not the best case.
The resolution is -100dB of full scale.

Also slightly lower THD is possible with input signals less than than full scale, which also reduces the dynamic display range.

\$\endgroup\$
-3
\$\begingroup\$

The Keithley 2015 is not an ordinary DMM, rather it has an ultralinear front end with a DSP that can measure SINAD and Harmonic distortion like an Audio Spectrum Analyzer.

So it has high performance specs of;
THD –65 dB (residual, relative to full scale)
Resolution -100 dB ( relative to 0dB being full scale)
Noise 100 µV rms

\$\endgroup\$
3
  • 5
    \$\begingroup\$ It's already established that it is not a DMM, both in the question title and the first two sentences of the question. OP asks what those figures actually mean. \$\endgroup\$
    – pipe
    Commented Jan 24, 2018 at 9:17
  • \$\begingroup\$ I believe it should be apparent that the spec I posted for resolution -100 dB is why you can measure lower than the internal THD . No insult but my answer shows obviously more clarity in dB. If you need more input, than this read the manual. The confusion comes from using percentages and oversight on Resolution \$\endgroup\$
    – D.A.S.
    Commented Jan 24, 2018 at 16:05
  • \$\begingroup\$ His main question I answered by duplicating specs he overlooked on Resolution not what causes residual error So I need to explain Resolution? \$\endgroup\$
    – D.A.S.
    Commented Jan 24, 2018 at 16:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.