QA402 - Phono preamp measurements

Hello
I’m using the QA402 for the development of a preamplifier with phono input. Measuring the phono inputs with the QA402 has some challenges which need to be considered.

  1. Output noise
    A phono preamp has a gain of typcally +40dB for a MM cartridge or +60dB for a MC cartridge (@ 1kHz). Therefore the input to the preamp needs to be low level (e.g. 5mV @ 1kHz = -46dBV). The QA402 output stage has an output attenuator followed by an output driver stage. For such small output signals, the output driver stage adds noise (or the internal -30dB attenuator adds the noise). Therefore, the THD+N and SNR measurements are not accurate. Using a -40dB (9.9k Ohm + 100 Ohm) or -60dB (10k Ohm + 10 Ohm) passive attenuator to feed the preamp input gives better results.

  2. Frequency response
    The frequency response of the phono preamp is non linear according to RIAA (or IEC). The weighting feature of the the QA402 allows easy evaluation of the frequency response deviation of the preamp. The standard RIAA playback curve file of the QA402 has relatively low frequency resolution. Therefore the accuracy is not very good if measurements with high resolution (high number of FFT bins) are made. I used a higher resolution RIAA playback file (about 80 frequency/gain pairs) for my measurements. I can share this file if someone is interested.

  3. Frequency response of the QA402
    A second little issue is the frequency response of the QA402. At 20kHz, the frequency response is down -0.25dB. While this can be compensated using the right channel as a reference for linear gain amplifiers, I have not tried this yet for the phono preamp. As an alternative, I consider adapting the weighting file accordingly which would allow to make a two channel frequency response measurement of the preamp.

Overall the QA402 is an excellent device for measuring audio equipment such as my preamp. Well done Matt and team!

Regards
Andy

1 Like

Hi Andy

I’d very much appreciate you sharing the high accuracy RIAA weighting file as I’d noticed there are some odd bumps in the phono response when testing various vintage amplifiers previously yet hadn’t made time to create a better one myself.

Also I’d noticed the noise and had originally tried external reverse compensation instead but that was worse. In the end I’d simply adjusted the signal level for MM to -30 to -35dBV to improve it.

Thanks,

Simon.

The HF response error seems unusual to say the least. Most ADC’s show infinitesimal errors in the response band. Do you see the .25 dB error in a straight loopback? Could there be some capacitive loading introducing the error? Maybe a loopback with the attenuator in place from the input to the attenuator?

Accurate measurement of RIAA to .1dB is not easy, You need both frequency and amplitude precision. A digital system should provide that. Effective source impedance is also important. Moving coils are very low but a moving magnet cartridge is pretty inductive so that should be emulated to quantify the effects on the input capacitance.

Hi 1audio
the 0.25dB error at 20kHz seems normal for the QA402. It can be easily measured with a loopback frequency response measurement - even without my passive attenuator. While the ADC and DAC have very little ripples at high frequencies the QA402 has analog anti aliasing filters which introduce this little error.
Best regards
Andy

Hi @Avo, yes indeed you are correct. On the QA402 release 0.997 the User button was moved over one slot to the left and renamed from User to User1. To the right of that, a User2 will be added shortly, and this will allow you to have two curves active simultaneously. So, you could have your high-resolution RIAA in User1 and a gentle correction for the ~0.25 dB around 20 kHz.

This will also be useful if you want to notch out powerline freqs that can creep into up in high-gain measurements Although sometimes I wonder if a generic 50 or 60 Hz IIR notch in SW might be useful.

image

You raise a good question on the DAC noise floor that should probably be part of the spec. Below I’m running the DAC output at 0 dBV into the QA480 notch. That is whacking the 1 kHz about -55 dB and ensuring the ADC isn’t contributing anything meaningful to the measurement. The Noise Minus Distortion (versus N+D) is -104.4 dBV (20 kHz, no weighting). Note the shape–the noise here is quite a bit higher than the QA480 noise and thus you can see the notch filter impacting the noise across the band.

At -10dBV output the N-D is -113 dBV

At -20 dBV output the N-D is -117 dBV

At -30 dBV output the N-D is -118 dBV

At -40 dBV output the N-D is -118 dBV

At -50 dBV output the N-D is -118 dBV. This plot is shown below:

Note the shape here is flat, which suggests it is dominated by the QA480.

Thanks for a very informative post!

1 Like

Many thanks @AVO for sending me the high resolution RIAA filters.

Here is a comparison measuring the phono input of a 1978 Luxman R-1030 that’s currently on my bench.

Using the default RIAA playback filter:

And then with your high res RIAA filter:

Both of these have the same settings including -40dBV SG output, averaging over 5 plots and maximum smoothing.

The right channel is within -1dB 10-30kHz through the amplifier’s line input, so the 20Hz -3dB on the right channel is due to the phono stage. The left channel has poor tone control calibration, so the bottom end prematurely rolls off. The OEM specification is 10-40kHz -1dB.

There’s still a bit of work for me to do to straighten this amp out. Can’t say I’d be worried about 0.25dB at 20KHz though :grin:

I had just finished re-measuring the MM phono input on my mid 80’s NAD intg amp when I read this topic chain. I am measuring out of one of the tape outputs, with no extra capacitive loading applied to the input.


There is a little 60Hz spur that comes an goes, I just made a capture with it not there. I am very pleased with the FR of both channels and don’t care about the little bit of noise >10kHz. While we all like measurement perfection, were are talking about playing vinyl after all… :slightly_smiling_face: I will look at the MC of the intg amp as well since it is switchable (-66dBv input will be used).
Update- I had made a spreadsheet to calculate the RIAA curves prior to the user weighting function being enabled, and decided to see if I could make my own Hi-res file, which I was once I read how to combine elements in Excel with a “,” . Below is what I ended up with-

It looks much nice nicer, though the 60Hz spur poked up. The file uses 9312 frequency points from 21Hz to 20001Hz.

1 Like