OP
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Review and Measurements of RME ADI-2 DAC
- Thread starter amirm
- Start date
-
- Tags
- rme adi-2 dac
OP
I hope you can figure that out. Would be nice to see a comparison of one headphone at a few different SPL. There might be some changes as headphone distortion decreases. The Benchmark whitepaper does not go very in depth here.
Quite so, I have never heard of a plausible way of separation noise from distortion.
Trivial with spectrum analysis. Note that Amir qualified his statement with "using the THD meter."
OK, how?
It is 40 years since I was regularly using a spectrum analyser but, even with a spectrum showing peaks which one may wish to describe as harmonic distortion how does one evaluate how much of an individual peak is a distortion product and how much a contribution from noise? Is assuming the noise level either side of the peak as an acceptable value to subtract?
Having got these separated using the spectrum analyser how do you derive a single percentage value to quote as the distortion contribution for comparison with a typical THD meter result?
Unless things are really pathological, the noise won't be significant- remember, you're only looking at the noise in two adjacent bins (unlike a THD meter which sums ALL the noise across the spectrum).
Here's an example, a distortion spectrum from an ADI-2 Pro. The second harmonic is down -115 dB, 3rd is down -118 dB, 4th is down -130 dB, and the noise floor is at -145dB on a bin for bin basis. So the contributions of harmonics above 2nd and 3rd will not dominate distortion energy, but more importantly, the noise contribution is more than an order of magnitude below them. So to a good approximation, sum 2nd and 3rd, maybe add the 4th (though that won't change things much), and the noise is like a third decimal place correction, i.e., negligible.
Remember, logs and RMS summing.
Remember, logs and RMS summing.
Attachments
Last edited:
OP
- Thread Starter
- #172
Correct. To do it using spectrum analysis, you would have to add up the harmonics yourself which would be hugely tedious in a sweep like this.
Looking at the labels below their graph, it matches the (template) test I am using from AP which relies on THD meter. So I am dubious how they excluded noise.
That said, there are analyzers like Prism Sound which can automate this. So there may be a way to do that here too. I just have not found an easy way to do it yet with the version of Audio Precision software/hardware that I (and Benchmark) have.
Fortunately, real world stuff that isn't horribly designed can almost always be closely approximated with 2nd and 3rd. Horrible designs might take a minute with a calculator. Literally a minute. There's probably a way to automate it with the AP that I have, but I'm insufficiently motivated.
OP
- Thread Starter
- #174
For a single number, that would be the case. But this is a frequency sweep vs THD. So if it takes you to compute one sample point one minute, then 30 sample points in the sweep would take 30 minutes.
And then you have to multiply that by the number of headphones. With four headphones you would be looking at 120 minutes or two hours of tedious work! And then there are multiple devices to test.
I love you guys but not that much!
Let me see if I can figure out a way to do this automatically.
At least with my software, you can display (e.g.) 2nd or 3rd or whatever harmonic vs frequency. That makes the math a lot quicker! I'm sure there's a way to add them within the AP software, but I'm lazy about learning, so I'd probably export to Excel and take 3 or 4 seconds to add the columns.
Jimster480
Major Contributor
Awesome measurements, even unseats the D50!
Jimster480
Major Contributor
APx500, running 4.4 on my lab computer and 4.5 on my laptop.
