QA403 Generator Spec

@matt I set up differential output from the QA403 into the balanced input of an e1da cosmos adc. I have used this adc to measure thd+n @1kHz of dacs to almost -120db.
The QA403 dac output had its best overall measurement at thd -119.5 and thd+n -112.3(48kHz, 32k fft). Is the spec of the dac in the QA403 being limited by other output buffers, relays, etc. in the QA403 or is there a problem with my measurement?

On another question the snr on the right input channel does not work properly

because the F indicator does not find the 1kHz fundamental.

Hi @Moto, the QA403 is a few dB away from the ES9038 spec due to the reference voltage noise inside the QA403. The ES9038 requires exceptional performance there to meet the spec–so exceptional that ESS Tech made a special part to provide the reference voltage (which isn’t used in the QA403). The higher output levels (8Vrms) have also had a slight contribution in THD.

The spec on the QA403 DAC is -115 dB for THD and -105 dB for THD+N. These are loopback numbers. So, you are performing very well relative to spec.

For the SNR, take a look at the green “F” on the display. It looks like it has locked onto 2 kHz as the fundamental, which is why your SNR probably doesn’t make sense. Remember, the left channel is the dominant channel, and if you are going to make a mono measurement, you should use the left channel OR if you must use the right channel, provide a signal to the left channel to ensure the fundamental is correctly picked.

Alternately, you could right click on the THD or THDN button and specify a frequency for the fundamental.

@matt thx for the reply. I did try right clicking thd button and setting to 1k then match sig gen 1, but still have a similar problem. It looks like the F is at the right frequency but incorrect snr.

Hi @Moto, I can replicate. I think this has been fixed for release 1.152 located HERE.

There has been a context menu added for SNR. It’s mostly the same as the THD (shared settings) but conceptually it should improve things. Also, notice the fundamental selection has changed a bit. Instead of LEFT channel peak, you now specify CHANNEL peak. So, in this selection mode, each channel will use its own peak to do the calc.

What you encountered was confusing at first because with the left channel unplugged (no shorting) there was enough leakage for a 1 kHz to show up in the left channel and that let the algorithm correctly determine the frequency BUT it was the wrong amplitude. And with the left channel shorted, it worked as expected–meaning it didn’t work at all.

Please give this new version a try when time permits and see if it works the way you expect, and thanks for reporting.

@matt thx. 1.152 works perfectly.
Is there any plan to include an harmonic cancelling feature such as the one in REW?

Hi @Moto, the option for hardware harmonic cancellation will be there, but not the ability to tweak a harmonic by specifying an equal but opposite phase.