Measurements of RME ADI-2 DAC and Headphone Amp

[PPP]

Speaking of measurements, here are RME's own (from the manual).

Also bear in mind that the graph above is measured by an ADI-2 Pro FS, whose ADC has -112 dB THD+N. I just don't understand how can it produce results better than the AD converter.

 

[boXem]

Also bear in mind that the graph above is measured by an ADI-2 Pro FS, whose ADC has -112 dB THD+N. I just don't understand how can it produce results better than the AD converter.

Chips specs are given full scale. Their best THD is not always at full scale.
For the ADC, it's quite easy: just don't use it full scale by attenuating the input. A notch will improve things further. I believe that even the AP555 uses a standard ADC.
For the DAC, it's certainly a bit more complicated, but the principle may be the same: use the IC at it's best working point.

 

[MC_RME]

And the old one to compare:
Huge improvement in THD and SINAD improves from 111db to 116db.

In fact my self-built notch steals about 0.5 dB due to its own noise. The APx555B measured -117.1 dB for fs/2, and -117.5 dB using the widely spread AES17 filter.

 

[PPP]

Chips specs are given full scale. Their best THD is not always at full scale.
For the ADC, it's quite easy: just don't use it full scale by attenuating the input. A notch will improve things further. I believe that even the AP555 uses a standard ADC.
For the DAC, it's certainly a bit more complicated, but the principle may be the same: use the IC at it's best working point.

I am not talking numbers for mere distortion. All the thd numbers I give above all include noise. You can lower the volume but the noise will come up.


For instance, amirm’s measurements here don’t make sense. In the original post he was astonished by the low harmonics it produced. But he is using a much lower volume and higher output level to get to 2v Se or 4v balanced, Far from getting close to the full volume the chip produces. However, You will see the thd+n number doesn’t get better.


also RME products avoid full volume already. Their standard +7dbu output level is slightly less than 2vrms.

 

[Rja4000]

ADI-2 Pro FS, whose ADC has -112 dB THD+N

Mine measures better than that, in loopback.

 

[boXem]

I am not talking numbers for mere distortion. All the thd numbers I give above all include noise. You can lower the volume but the noise will come up.


For instance, amirm’s measurements here don’t make sense. In the original post he was astonished by the low harmonics it produced. But he is using a much lower volume and higher output level to get to 2v Se or 4v balanced, Far from getting close to the full volume the chip produces. However, You will see the thd+n number doesn’t get better.


also RME products avoid full volume already. Their standard +7dbu output level is slightly less than 2vrms.

XLR/RCA output voltage is not directly produced by the chip.
Just look at the SMPTE graph, behavior is completely different from the DX3 pro which uses a similar chip, even if not the exact same, but in a simpler implementation.

 

[PPP]

Mine measures better than that, in loopback.

Mine just doesn't beat the spec in loop back (using either RMAA [default setting] or Electroacoustics Toolbox).
I guess this depends on the way the measurement is performed.

AKM perhaps measures their chips in the same condition (their test conditions are published in spec sheet) and the thd+n only differs by 1db apart.
So the question is how RME perform the two tests.

 

[JIW]

The new version has 4 dB of digital attenuation before the DA. Thus for a 0 dB FS input, the THD of the DA chip should correspond to that of the DA chip at -4 dB FS which should be a good bit lower.

Regarding noise, AKM specs the AK4493's SNR at 123 dB for 2 V RMS output but also as 125 dB for large amplitude mode. The 4 dB attenuation should result in at most a 4 dB reduction of the SNR so RME's spec of 120 dB (123 dBA) SNR seems plausible.

 

[Rja4000]

Mine just doesn't beat the spec in loop back (using either RMAA [default setting] or Electroacoustics Toolbox).
I guess this depends on the way the measurement is performed.

AKM perhaps measures their chips in the same condition (their test conditions are published in spec sheet) and the thd+n only differs by 1db apart.
So the question is how RME perform the two tests.

 

[PPP]

View attachment 38597

That's awesome. What AD and DA level and volume do you perform the test with?

 

[Rja4000]

That's awesome. What AD and DA level and volume do you perform the test with?

There is a trick.
I use balanced phone output (low gain, so +13dBu or 3.5Vrms full scale), balanced inputs (+19dBu range here).
And finally, with new firmware, you may sum both ADC channels, for some additional dBs

 

[Willem]

To be absolutely honest I am not that interested in or concerned with such even better specifications, even though I will always admire the technology. The ADI-2fs was already much better than human hearing acuity. What I would be happy to upgrade to would be an RME DAC with more dsp processing power for at the very least rather more parametric filters, but preferably integrated Dirac or similar.

 

[VintageFlanker]

Oh my... I'm so weak and stupid...

Starting to think about "upgrade" mine, while there is no f**king reason to do so (at least, in the audible realm)...

Good news for folks here is I could put mine on sale on ASR...

OK, stop...saying...BS...resist...resist...

 

[Bliman]

Oh my... I'm so weak and stupid...

Starting to think about upgrade mine, while there is no f**king reason to do so (at least, in the audible realm)...

Good news for folks here is I could put mine on sale on ASR...

OK, stop...saying...BS...resist...resist...

Stop. This nonsense will not be tolerated here . Be stronger than this.

 

[JIW]

Oh my... I'm so weak and stupid...

Starting to think about "upgrade" mine, while there is no f**king reason to do so (at least, in the audible realm)...

Good news for folks here is I could put mine on sale on ASR...

OK, stop...saying...BS...resist...resist...

You could ask for a board replacement... Should fit right in.

 

[VintageFlanker]

You could ask for a board replacement... Should fit right in.

Yes it should!
@MC_RME, is that a possible scenario?

 

[PPP]

There is a trick.
I use balanced phone output (low gain, so +13dBu or 3.5Vrms full scale), balanced inputs (+19dBu range here).
And finally, with new firmware, you may sum both ADC channels, for some additional dBs

balanced phone output/summing adc channels is like using the mono mode of the chip. of course it has much better spec. that's not a fair measurement.

 

[Rja4000]

balanced phone output/summing adc channels is like using the mono mode of the chip. of course it has much better spec. that's not a fair measurement.

It's not about fair or not.
It's about being able to use the device to measure deeper.

 

[Willem]

Relax and avoid all this audiophilia nervosa.

 

[VintageFlanker]

Relax and avoid all this audiophilia nervosa.

Easy to say. I have to confess: Even after 2 years reading ASR, I'm still quite sick about upgrading.