AVR or processors from China with Topping performance?

[mdsimon2]

I really doubt this is the case, considering that Home Cinema pushes for at least above 90dB/2.83V sensitivity speakers to get adequate levels across long distances.

if we plug in the same calculation but using something more realistic, like the Denon's own amplifiers, the residual noise of the DAC will be 200uV and the residual noise of the amplifier will be 200uV (instead of 35uV) resulting in a total S/N ratio of ~94dB.

This is honestly audible in most situations, specially during the night or in a dedicated room.

I think your numbers are a little pessimistic as the Denon amps have lower gain than your previous example, although in general I agree with the conclusion that noise contribution will be approximately equal between DAC and amplifier. Here is how I see the residual noise calculations, I might be missing something here so definitely welcome any feedback.

DAC - DR @ 2V = 108 dB
noise at DAC output = V * 10^(-DR/20) = 2 * 10^(-108/20) = 8 uV
noise from DAC at amplifier output after 26 dB gain, 8 * 10^(26/20) = 159 uV

Amp - DR @ 5 watts in to 4 ohms = 89 dB
voltage for 5 watts @ 4 ohms = sqrt(P * R) = sqrt(5 * 4) = 4.5 V
noise from amplifier = 4.5 * 10^(-89/20) = 159 uV

Amp - DR @ 168 watts in to 4 ohms = 104 dB (sanity check for amp noise calculation above)
voltage for 168 watts @ 4 ohms = sqrt(168 *4) = 25.9 V
noise from amplifier = 25.9 * 10^(-104/20) = 164 uV

Summing DAC and amplifier noise and calculating DR at 5 watts
DR = 20 * log(4.5 / ((164 + 159) * 10^-6)) = 83 dB

Summing DAC and amplifier noise and calculating DR at full power
DR = 20 * log(25.9 / ((164 + 159) * 10^-6)) = 98 dB

All of this is quite theoretical, I have a variety of DACs with DR ranging from 101 dB to 118 dB (at 2 V) and a 91.5 dB sensitivity tweeter in the Seas 27TBFCG, I think I will do some listening tests this weekend with a Hypex NC252MP amplifier (noise definitely not as low as Purifi or Benchmark but probably more in line with what most people will use) and see how audible the DAC noise is near field. My recollection from past testing is that a DAC with ~100 dB DR does have audible noise at very close listening distances but at a listening position of more than 2 m the noise is not audible. I will also try to make some electrical measurements of the amplifier outputs to add some quantification. As I've mentioned I've done listening in the past but it will be fun to do some more structured testing.

Michael

 

[abdo123]

I think your numbers are a little pessimistic as the Denon amps have lower gain than your previous example, although in general I agree with the conclusion that noise contribution will be approximately equal between DAC and amplifier. Here is how I see the residual noise calculations, I might be missing something here so definitely welcome any feedback.

DAC - DR @ 2V = 108 dB
noise at DAC output = V * 10^(-DR/20) = 2 * 10^(-108/20) = 8 uV
noise from DAC at amplifier output after 26 dB gain, 8 * 10^(26/20) = 159 uV

Amp - DR @ 5 watts in to 4 ohms = 89 dB
voltage for 5 watts @ 4 ohms = sqrt(P * R) = sqrt(5 * 4) = 4.5 V
noise from amplifier = 4.5 * 10^(-89/20) = 159 uV

Amp - DR @ 168 watts in to 4 ohms = 104 dB (sanity check for amp noise calculation above)
voltage for 168 watts @ 4 ohms = sqrt(168 *4) = 25.9 V
noise from amplifier = 25.9 * 10^(-104/20) = 164 uV

Summing DAC and amplifier noise and calculating DR at 5 watts
DR = 20 * log(4.5 / ((164 + 159) * 10^-6)) = 83 dB

Summing DAC and amplifier noise and calculating DR at full power
DR = 20 * log(25.9 / ((164 + 159) * 10^-6)) = 98 dB

All of this is quite theoretical, I have a variety of DACs with DR ranging from 101 dB to 118 dB (at 2 V) and a 91.5 dB sensitivity tweeter in the Seas 27TBFCG, I think I will do some listening tests this weekend with a Hypex NC252MP amplifier (noise definitely not as low as Purifi or Benchmark but probably more in line with what most people will use) and see how audible the DAC noise is near field. My recollection from past testing is that a DAC with ~100 dB DR does have audible noise at very close listening distances but at a listening position of more than 2 m the noise is not audible. I will also try to make some electrical measurements of the amplifier outputs to add some quantification. As I've mentioned I've done listening in the past but it will be fun to do some more structured testing.

Michael

I did adjust for the lower gain of the Denon's Amp, please do your listening tests late at night, it makes a gigantic difference.

 

[mdsimon2]

I did adjust for the lower gain of the Denon's Amp, please do your listening tests late at night, it makes a gigantic difference.

Then how do you get 200 uV?

Michael

 

[abdo123]

Then how do you get 200 uV?

Michael

well the gain is set at ~26 dB so that's 25 times increase the residual noise of the DAC, 8*25 = 200uV

am I missing something?

Edit: I should stop doing complex math in my head

 

[mdsimon2]

well the gain is set at ~26 dB so that's 25 times increase the residual noise of the DAC, 8*25 = 200uV

am I missing something?

Edit: I should stop doing complex math in my head

10^(26/20) = 20

Looks like you had the 29 dB calculation correct at 10^(29/20) = 28. Rule of thumb is 3 dB is 1.4x so 28 / 1.4 = 20 also matches the above calculation for 26 dB.

Michael

 

[mdsimon2]

OK, measurements and listening tests done. All the DACs used were reviewed by Amir in some fashion.

Measurements were done using a Motu M4 audio interface, an Audiophonics NC252MP amplifier (25-26 dB gain) connected to a 4 ohm dummy load with a voltage divider to protect the M4 from any spurious pops or accidental over voltage. The MOTU M4 was powered by a MacBook pro which was battery powered. In order to make somewhat meaningful measurements I calibrated the M4 by sending a 9 V from the amplifier and used REW's calibrate level functionality. Probably good to mention that I do not trust these measurements in an absolute sense but as relative comparisons they seem reasonable. Measurements were done with no output from the DAC, I did not find any evidence of obvious low level muting. Also important to note that these measurements would not reveal any noise floor modulation with increased output.

For listening tests I used a Seas 27TBFCG tweeter which as mentioned previously has a spec'd sensitive of 91.5 dB. The tweeter was connected directly to the amplifier with no padding or crossover. I listened at 1 m and 2 m distances, done late at night.

First up is the M4 ADC with NC252MP inputs shorted, 112 uV noise. All noise measurements are unweighted, 20 - 22000 Hz but you can also see noise levels above 22 kHz in the measurements.

Starting with the lowest performing DAC, a miniDSP OpenDRC-DA8. This uses the same base DAC as the U-DAC8 reviewed by Amir. Amir measured the DR at 100-101 dB. The graph for this one is not pretty. I had made full scale output measurements in the past and got similar SINAD values to Amir at 79 dB but spikes in this measurement were a surprise. I measured the noise level at 403 uV. As you can imagine noise was easily audible from the Seas 27TBFCG at 1 m. At a distance of 2 m you can definitely tell when the DAC is on although it is much less noticeable. Personally I can live with this level of noise but I can see how the noise could be an issue, especially if listening near field.

Next up is a miniDSP 2X4HD. I do not think Amir measured DR but the company spec is 103 dB so just slightly better than the OpenDRC-DA8. I measured the noise level at 346 uV. If the OpenDRC-DA8 is used as a baseline the measured noise performance is 1.9 dB better which is exactly as expected. The graph does not show the spikes seen in the OpenDRC-DA8. Similar listening impression to the OpenDRC-DA8, maybe a bit better but that may be my own bias knowing that it measures slightly better.

The next 3 DACs have very similar noise performance and listening impressions were very similar to each other and were a significant improvement over the first two DACs. Very difficult to hear anything at 1 m, if you get within 0.5 m you can definitely hear whether the DAC is on/off. At 2 m+ noise is inaudible. I would be very happy with the noise performance of all DACs from here on out even if listening near field.

Amir measured the MOTU M4 DR at 118 dB but this was at an output voltage of 5.2 V, correcting this to 2 V output for an apples-to-apples comparison gives ~110 dB. I measured the noise level at 164 uV. Comparing to the OpenDRC-DA8 baseline this is 15 dB better which is quite a bit better than expected. It makes me think that the unbalanced DACs (OpenDRC-DA8, 2X4HD) have more noise than anticipated as a result of ground loops.

Amir measured the MOTU Ultralite Mk5 DR at 115-116 dB but this was at an output voltage of 4.2 V, corrected to 2 V this is ~109 dB. I measured the noise level at 160 uV which is slightly better than the MOTU M4, 15.7 dB better than OpenDRC-DA8.

Amir measured the Schiit Modius DR at 115-116 dB but this was at an output voltage of 3.95 V, corrected to 2 V this is ~110 dB. I measured the noise level at 153 uV which is slightly better than the MOTU Ultralite Mk5, 17.0 dB better than OpenDRC-DA8.

And finally the Okto dac8 pro. Amir measured the 8 channel Okto prototype DR at ~122 dB but this was at an output voltage of 3.3 V, corrected to 2V this is ~118 dB. I measured the noise level at 130 uV, a 24.2 dB improvement over the OpenDRC-DA8. This DAC is absolutely silent from any listening distance, if you stick your ear directly next to the tweeter you can maybe convince yourself there is some noise but I have to think that the majority of that noise is from the amplifier.

A few takeaways, a DAC with a mid 100 dB DR at 2V is more than good enough for me. It is also interesting that the delta between the unbalanced DACs and balanced DACs is more than predicted but if you compare the unbalanced DACs to each other the delta is exactly as expected and if you compare the balanced DACs to each other the delta is also exactly as expected. To me this points to the fact that the noise performance of unbalanced DACs in the real world may be worse than indicated by the ASR measurement. Another reason to stick with balanced inputs/outputs if at all possible.

To me the Denon AVR-X3700H with a DR of 108 dB at 2 V discussed earlier in the thread is likely inaudible at normal listening distances although the lack of balanced outputs would certainly give me pause.

Michael

 

[ad_fletch]

Thanks for writing up the results of that test Michael.

So I have been searching, but it seems there's no option for anywhere near a reasonable price for an AVR with balanced outs to use a better amp (such as the Hypex) for the main channels.

Not really keen to buy a second poweramp for other speakers, but I suppose it's either that or keep waiting for a product that may never arrive....

EDIT: may have found an option - Yamaha RX-V3085. Same as RX-A3080 but without the "Aventage" name which adds $500. Now just need some measurements....

 

[mdsimon2]

I had been thinking about this thread recently and how it relates to residual noise thresholds that result in hiss. I was also a bit bothered that the expected noise results didn't seem to quite match reality in terms of comparison to the OpenDRC baseline. As it turns out my relative comparisons were wrong. I was straight subtracting OpenDRC baseline noise from measured noise but that does not make any sense as noise adds as the square root of the sum of squares. Taking this in to consideration I was able to calculate DAC DRs that roughly match Amir's results!

I'll go through an example with the MOTU M4 first and then show a table of all DACs.

Take measured noise (164 uV) and subtract residual noise from ADC + amplifier (112 uV) to get DAC noise contribution at amplifier output.

amplified DAC noise = sqrt (164^2 - 112^2) = 120 uV

From past measurements I know that this amplifier has a gain of ~25.6 dB. Subtract this gain to determine noise at DAC output.

noise at DAC output = 120 x 10^(-25.6/20) = 6.3 uV

Convert this noise level to a DR at 5.2 V to match Amir's measurements.

DAC DR = 20 x log (5.2 / (6.3 x10^-6)) = 118 dB

This exactly matches Amir's measurement.

Here is a table of all the DACs.

Everything looks about where it should be. Okto should be a bit better but at that level of performance just a few uV difference in measurement or baseline noise makes a huge difference.

Next I was interested in seeing if I could predict the noise measurement by taking a baseline measurement of the ADC noise and adding noise based Amir's amplifier and DAC DR measurements.

Here is what the ADC looks like with inputs shorted, 102 uV residual noise.

Based on Amir's Buckeye NC252MP review I would expect the amplifier to have a DR of ~99 dB at 5 W in to 4 ohms. From this I can calculate the expected amplifier noise contribution.

V = sqrt (5 x 4) = 4.47 V

amp noise = 4.47 x 10^(-99/20) = 50 uV

Using the M4 as an example I can calculate the DAC residual noise based on Amir's measurement of 118 dB DR at 5.2 V.

DAC noise = 5.2 x 10^(-118/20) = 6.3 uV

amplified DAC noise = 6.3 x 10^(25.6/20) = 121 uV

Now I can sum these noise values with the ADC noise and get a predicted measurement.

predicted measurement noise = sqrt(50^2 + 121^2 + 102^2) = 167 uV

167 uV is pretty darn close to my measurement of 164 uV! This basically means that the amplifier DR is right around 99 dB as expected.

The other takeaway I had from this exercise is that I had originally thought mid-100 uV noise at amplifier output was pretty good performance, however this was a bit of a misinterpretation. After subtracting ADC baseline noise that number is actually more like low-100 uV, although there is a pretty big gap between the 2X4HD and the M4 so it is a bit difficult to draw the line. Here is a table of all DACs with the 102 uV removed from the noise measurement.

All in all this has been a very good learning exercise for me and I now have much more confidence in the absolute values of these measurements.

Michael

 

[mdsimon2]

Also tested a Focusrite 18i20 2nd gen today. I had not listened to it with higher sensitivity speakers (like the SEAS tweeter used in the previous tests) but knew it had a good amount of hiss on my LXmini system which is not very sensitive. As a result I did not have great expectations and it turns out it was not very good.

This is definitely the worst of the bunch so far.

DR spec is 108 dB(A) which at first glance does not sound terrible but once you dig in you see that it is not that great. At full output of 4.86 V I get an unweighted DR of 106 dB. Using this calculator -> http://www.inertance.com/engineering-calculator/dba/?ckattempt=1 and assuming a roughly flat noise spectrum I see that you get another 2-3 dB when using dB(A) so that is right online with spec.

When you normalize everything to DR at 2 V you see that this performance is worse than the miniDSPs at 98 dB DR. Here is the updated table with the DACs I have measured to date.

Michael