Review and Measurements of Purifi 1ET400A Amplifier

[MJT]

I'm running some numbers of the Okto DAC8 PRO vs Motu UltraLite-mk5 for use with Purifi amps. Because the UltraLite-mk5 has a decent output signal voltage of 8.6v, it could in theory be used to feed the 1ET400A without pre-gain at 81.2% power (8.6v/10.59v), and also in theory, the net SINAD would be improved by 1.638dB over the Okto DAC8 PRO?...

---------------------
Motu UltraLite-mk5
Channel 1 SINAD: 0.000226%
Channel 2 SINAD: 0.000229%
Averaged SINAD: (0.000226% + 0.000229%)/2 = 0.0002275%

Purifi 1ET400A (No Pre-Gain)
Channel 1 SINAD: 0.000391%
Channel 2 SINAD: 0.000392%
Averaged SINAD: (0.000391% + 0.000392%)/2 = 0.0003915%

Combined SINAD: 0.0002275% + 0.0003915% = 0.000619 = -104.166dB

---------------------
Okto DAC8 PRO
Channel 1 SINAD: 0.000127%
Channel 2 SINAD: 0.000124%
Averaged SINAD: (0.000127% + 0.000124%)/2 = 0.0001255%

Purifi 1ET400A (With Pre-Gain)
Channel 1 SINAD: 0.000627%
Channel 2 SINAD: 0.000617%
Averaged SINAD: (0.000627% + 0.000617%)/2 = 0.000622%

Combined SINAD: 0.0001255% + 0.000622% = 0.0007475% = -102.528dB
---------------------

SNR would also have a net improvement but both combinations would already be beyond audible.

So, provided 81.2% power is sufficient, it's possible the UltraLite-mk5 may be a better match for the Purifi?

Could also go with a custom output voltage for the Okto DAC8 PRO, but everything I'm seeing indicates the noise penalty with custom output voltage is worse than pre-gain.

It would be ideal if Amir happens to test a custom input buffer that doesn't add as much noise, but so far they seem to test worse than stock Purifi buffers.

 

[Killingbeans]

Combined SINAD: 0.0002275% + 0.0003915% = 0.000619 = -104.166dB

Doesn't quite work that way :

How to interpret SINAD numbers of separate DAC + Amp combos

Let's say I have a SINAD of a DAC and a SINAD of an Amp. When comparing combos, should I look at Avg(SINAD_1, SINAD_2) or Min(SINAD_1, SINAD_2) as a more representative number of the performance of the combination?

www.audiosciencereview.com

 

[MJT]

Doesn't quite work that way :

Thank you! I thought converting to percentages would add linearly, but incorrect. Amir also points out the combined effect is negligible when the two devices vary greatly. You can see how this plays out when using the calculator he linked. The DAC8 excellent SINAD of -118dB has little bearing relative to the Purifi with gain..

-118.027dB (DAC8 Pro) & -104.124dB (Purifi with gain) --> -103.951dB

Yes, that is the point to have least effect, but if we pair up Ultralite with the same Purifi with gain, the difference between the DAC8 and Ultralite isn't even within JND because everything is brought down by the weakest link being -104dB..

-112.860dB (Ultralite) & -104.124dB (Purifi with gain) --> -103.579dB

Thus even with Ultralite's much less SINAD, it's significantly beneficial to get that lower number up with "no gain"..

-112.860dB (Ultralite) & -108.145dB (Purifi no gain) --> -106.881dB

Of course Amir describes the noise component of THD+N is affected by spectrum correlation. The Ultralite spectrum does differ from the Purifi more than the DAC8 does, so the noise component less overlaps/compounds.

Not to get O.T. with these particular DACs, however it's insightful when pairing the Purifi in general.

 

[RandomEar]

I'm not sure where to find the SINAD over voltage values for the DAC8 and UltraLite, but I assume the -118 and -112 dB are achieved at their respective maximum voltages, right? If so, I'd say you can't calculate it this way, as the amp would already be far past the clipping point with the gain enabled and >4 V at the input. To make the comparison fair, you'd have to use the DAC SINAD numbers at around 1.5 V, which are significantly lower - probably around -110 and -105 dB. That would give the "full blast + no gain"-combo an even higher SINAD advantage. Likely all in the inaudible territory, but still nice to know

 

[mdsimon2]

The Ultralite Mk5 does not like amplifiers with low input impedance (like you get when not using a buffer), the result is higher distortion. I show some measurements of this here -> https://www.audiosciencereview.com/...xxmp-buffer-bypass-someone.32763/post-1147902.

That being said I think system SNR (combined DAC and amp) as a function of volume position is the more relevant metric for comparison. THD goes down at lower volume levels but noise stays constant with digital volume control, therefore at most volume positions SNR is more relevant than SINAD (not to mentioned that at higher performance levels SINAD is often dominated by noise).

Michael

 

[MJT]

If so, I'd say you can't calculate it this way, as the amp would already be far past the clipping point with the gain enabled and >4 V at the input.

The Ultralite signal is 8.6v, Purifi amp is input without pre-gain 10.59v, so 81.2% power. But now this is made irrelevant by..

The Ultralite Mk5 does not like amplifiers with low input impedance (like you get when not using a buffer), the result is higher distortion. I show some measurements of this here -> https://www.audiosciencereview.com/...xxmp-buffer-bypass-someone.32763/post-1147902.

Very appreciative to know this. Pre-gain required for both options.

That being said I think system SNR (combined DAC and amp) as a function of volume position is the more relevant metric for comparison. THD goes down at lower volume levels but noise stays constant with digital volume control, therefore at most volume positions SNR is more relevant than SINAD (not to mentioned that at higher performance levels SINAD is often dominated by noise).

I was arriving at this as well. DAC8 excels at SNR and clearly multi-tone test. Since I'm already using NC400 for 7.1, I'll upgrade to the DAC8 (when available) and wait for measurements on the new Purifi 1ET7040SA.

I appreciate learning so much through exchange of ideas. This forum continues to amaze me. Being fully informed of component choices makes final selection more assuring and way more rewarding.

EDIT: Just ordered DAC8. Super excited!

 

[RandomEar]

The Ultralite signal is 8.6v, Purifi amp is input without pre-gain 10.59v, so 81.2% power. But now this is made irrelevant by..

My comment was aimed at the case where you use the DAC at full power and the amp with activated gain - I think the maximum input voltage in that case is 1.5 V, right? Anyway, I think your hardware choices will result in a more than capable setup

 

[mdsimon2]

The Ultralite signal is 8.6v, Purifi amp is input without pre-gain 10.59v, so 81.2% power. But now this is made irrelevant by..


Very appreciative to know this. Pre-gain required for both options.


I was arriving at this as well. DAC8 excels at SNR and clearly multi-tone test. Since I'm already using NC400 for 7.1, I'll upgrade to the DAC8 (when available) and wait for measurements on the new Purifi 1ET7040SA.

I appreciate learning so much through exchange of ideas. This forum continues to amaze me. Being fully informed of component choices makes final selection more assuring and way more rewarding.

EDIT: Just ordered DAC8. Super excited!

I've been meaning to write a longer post explaining how to compare noise performance between different DAC and gain settings with the Purifi. Everything shown below is based on values measured by Amir but should be pretty close to spec'd performance for the equipment in question.

DAC8 pro w/ 1ET400A at 27.5 dB gain
DAC SNR = 122 dB at 3.3 V
DAC residual noise = 3.3 x 10^(-122/20) = 2.62 uV
amplified residual noise = 2.62 x 10^(27.5/20) = 62.2 uV
amp SNR at 5 W in to 4 ohm = 105 dB
amp residual noise = sqrt(5 x 4) x 10^(-105/20) = 25.1 uV
system residual noise = sqrt(25.1^2 + 62.2^2) = 67.1 uV
system SNR at 5 W in to 4 ohm = 20 x log (4.47 x 10^6 / 67.1) = 96 dB

Ultralite Mk5 w/ 1ET400A at 12.8 dB gain
DAC SNR = 121 dB at 8.6 V
DAC residual noise = 8.6 x 10^(-121/20) = 7.66 uV
amplified residual noise = 7.66 x 10^(12.8/20) = 33.5 uV
amp SNR at 5 W in to 4 ohm = 110 dB
amp residual noise = sqrt(5 x 4) x 10^(-110/20) = 14.1 uV
system residual noise = sqrt(33.5^2 + 14.1^2) = 36.3 uV
system SNR at 5 W in to 4 ohm = 20 x log(4.47 x 10^6 / 36.3) = 102 dB

So about a 6 dB improvement in system SNR when using the Ultralite Mk5 with the low gain setting. Interestingly enough if you went with an implementation with a mid-gain setting like this -> https://www.audiosciencereview.com/...ics-hpa-s400et-review-stereo-amplifier.32014/ and used the Okto you can get the same noise performance

DAC8 pro w/ 1ET400A at 20.2 dB gain
DAC SNR = 122 dB at 3.3 V
DAC residual noise = 3.3 x 10^(-122/20) = 2.62 uV
amplified residual noise = 2.62 x 10^(20.2/20) = 26.8 uV
amp SNR at 5 W in to 4 ohm = 107 dB
amp residual noise = sqrt(5 x 4) x 10^(-107/20) = 20.0 uV
system residual noise = sqrt(26.8^2 + 20.0^2) = 33.5 uV
system SNR at 5 W in to 4 ohm = 20 x log(4.47 x 10^6 / 33.5) = 102 dB

Although honestly all of these systems are very low noise and unless you are using extremely sensitive speakers you will not have any noise issues. For reference I like to keep total residual at less than 200 uV so these are all well below that.

Michael

 

[MJT]

Very nice. The mid gain setting looks like the perfect match. Max power at 4.2v (DAC8 4.1v). Most helpful.