Newbie here.
I’m a HiFi hobbyist, want measure DAC, preamp, integrated amps and amps performance.
Particularly noise, THD, THD+N for DAC and preamps. And noise, THD, THD+N over frequency band and power output for integrated amps and amps.
Can I use the QA403 to do this?
Is it precise enough to measure ultra high performance HiFi equipment such as Topping and Benchmark products, since these products get into the .0000x THD+N.
Hi @gallantus, generally to accurately measured THD+N, you need need a measuring device with ~10 dB better THD+N than the device you want to measure. The best THD+N from a DAC these days is around -121 dB or so, which suggests you need a measuring device with about -130 dB or so. But, those don’t exist. So, the best you can really do is confirm it’s getting close. Next, you can use a notch to improve things further.
A good thread on attempting to measure the E70 Velvet is below. This should help you understand the limits.
Just for reference purposes, I have a new Q403. I consistently get noise = -117.4dBV, from the RMS measurement. This is for a 10-80kHz bandwidth. Every displayed peak is below -150 dBV in the FFT.
The best THD at 1kHz using the built in DAC and ADC is around .0001x where x is between 0 and 4. So you will not be able to directly measure sub part per million distortion. A high-quality external oscillator and matching notch filter should get you there. You might also be able to get there with Bob Cordell’s distortion magnifier, at least if the distortion is in the lower harmonics. There is a guy over in DIYaudio forums that has used a QA with external accessories to measure an amp at will under the part per million range.
The biggest problem you will have at those levels is managing the noise. Power line hum easily runs -120 or more unless extreme care is taken, and EMI also becomes a serious issue at those levels.
So essentially, I can use the QA out of the box on DUT up to 107db, since Matt is saying, you need at 10db headroom?
I would think that should work fine if you can keep the noise out. At about -112dB, things will start to get questionable, in my limited experience. You still get a measurement, but the amplitude range over which you can get this will be limited. There is a sweet spot for both the ADC and DAC. When you go above it, the THD picks up, and when you go below it, the SNR gets worse. Fortunately, the 403 has a much better attenuator than the 401 had, and each step in the attenuator seems to have the same ADC characteristic with regard to optimum, just as you might expect.
Hi @Gallantus, @mhuth1776 is correct here.
Here’s a bit more on wny the 10 dB margin is important. First, let’s do an experiment on the QA403.
Set up single-ended loopback (L+ OUT to L+ IN), pick a 32K FFT, dial in some averaging, set the full scale to +42 dBV. What this setup does is measure the self-noise of the QA403 in +42 dBV mode. as the plot below shows, the noise is about -78.6 dBV (20 kHz BW). With the full scale being _42, this means the noise is about -120 (42+78.6) below the full scale input. So far so good.
Now, enable white noise and set the white noise level to -78.6 dBV. Now we have two sources of noise at the same level and uncorrelated. We see the RMS noise is -76, which is 2.61 dB higher.
When you have two uncorrelated noise sources and you wish to add them, you must do as follows:
N=SQRT(X^2+Y^2+Z^2…)
So, if we have noise source one which is -100 dBV = 10uV rms, and noise source 2 which is the same, then the total noise is SQRT(10uV10uV + 10uV10uV) = 14.16uV = 96.98 dBV, which is 3.01 dBV. This is pretty close to what we saw above.
This is a very important property to understand: Uncorrelated sources (noise) added together increase the signal by 3 dB = sqrt(2) = 1.41. And correlated sources (think of two identical sine waves) added together increase the signal by 6 dB = 2X.
This is how you can parallel two chip DACs together and improve the SNR. The sine waves add by 6 dB, but the noise only adds by 3 dB. So, every time you double the numbers of DACs, you improve the SNR by 3 dB
OK, so back to the 10 dB rule I’d first mentioned. Much of what we want to measure is defined by the noise level. So, if your instrument has -120 dBV of noise, and your DUT has -120 dBV of noise, you’d measure -117 dBV of noise. That is a 3 dB error.
You can run through the math and you get a table like that below. And that shows that if your analyzer noise is 10 dB better than your DUT noise, your measurement error is 0.41 dB. That is, the DUT noise measures about a half dB worse than it actually is. This is because of the analyzer noise.
Now, the good news is that if you are certain of the analyzer noise, you can back that out. That is, if you know your analyzer noise is -120, and you are measuring DAC noise at -117.5, it’s probably a safe bet that your DAC noise is actually -121.2 dBV.
Hi @Matt. Silly question on my part. How do you set the RMS value of the “White Noise” generator? I when I activate it I always see it at -12 dBV and I don’t know how to change its value. Thank you
Hi @Claudio, you can’t directly set it (by typing numbers) but you can adjust via the up/down buttons and control/alt modifiers (to jump 10 dB per click or 0.1 dB per click).
The noise generator needs an overhaul which is coming. The ability to generate pink noise is in the code, just not exposed yet. And the shortcut ctrl-1 to enter amplitude only works on sines, but needs to be extended to deal with all modes. That is, if you are generating multitone, and you hit ctrl-1, the amplitude you enter should apply to the multi-tone/noise, chirp, etc.
Thank you @Matt. So, if I understand correctly, at the moment “White Noise” is fixed at -12dBV but soon it will be possible to change it.
Hi @Claudio, you can change it now. When in white noise mode, click the GEN1 AMP buttons to increase/decrease. And if you hold CTRL while clicking up, it will increase by 10 dB. And if you hold ALT while clicking up, it will increase by 0.1 dB.
So, you should be able to hit just about any level of white noise you need quickly using the mouse.
Hi @Matt. I tried your suggestion and it works. Now I can change the dBV of “White Noise” as I need. Thanks again Matt!
