Review and Measurements of RME ADI-2 DAC

[Sythrix]

I really have to wonder at these tests and their applicability.

Isn't the very act of measuring it at the connection diverting power into the measurement equipment? Thus changing the actually frequency response? What if the benchmark amp simply has enough extra juice to compensate for that diversion completely, whereas the other amps are now playing with a circuit that is not what was intended.

Sorry for being skeptical and I'm certainly not the expert on this, but I would want to see a hell of a lot more data and comparisons before announcing my unyielding faith in Benchmark.

They're a company and want to sell amps. Their test size was very small and did not disclose the full details of their competition.

I would like to see the effect of this ranging by headphone amp total power output and it's performance at those relative output levels... but I know you're busy Amir, so I don't expect you to go nuts on this.

Might be interesting to insert one of these at every review though, so we get a more comprehensive data set.

 

[SIY]

The "diversion" is microscopic. Ohm's Law.

Agreed, though, that there's less than full disclosure because it's a marketing document.

 

[Frank Dernie]

I really have to wonder at these tests and their applicability.

Isn't the very act of measuring it at the connection diverting power into the measurement equipment? Thus changing the actually frequency response? What if the benchmark amp simply has enough extra juice to compensate for that diversion completely, whereas the other amps are now playing with a circuit that is not what was intended.

Sorry for being skeptical and I'm certainly not the expert on this, but I would want to see a hell of a lot more data and comparisons before announcing my unyielding faith in Benchmark.

They're a company and want to sell amps. Their test size was very small and did not disclose the full details of their competition.

I would like to see the effect of this ranging by headphone amp total power output and it's performance at those relative output levels... but I know you're busy Amir, so I don't expect you to go nuts on this.

Might be interesting to insert one of these at every review though, so we get a more comprehensive data set.

If the measuring instrument has an input of 1 megohm, which wouldn't be unusual, then the power going its way would be less than 0.01% even with the highest impedance 'phones, so nothing to worry about for the non-expert!

 

[Soniclife]

Fascinating, those changes look audible.

Have these sorts of tests been done with amps connected to speakers?

 

[dc655321]

And here is the same data for Auralic Gemini 2000 @1 volt output:

View attachment 11961

The baseline (dummy load) is worse than Topping but with real headphones, performance is better.

Another conclusion seems to be that the order does not change. The HiFiman HE400i is least stressful on the DAC and Grado is the worst.

Your plots correlate well with Innerfidelity's THD+N measurements of those headphones:
Grado SR60i
Sony MDR V6
Senn HD 650
Hifiman HE400i

Is this perhaps the signature of strong electro-mechanical resonance?
Trying to think through exactly what is happening at those bass resonances...

 

[Frank Dernie]

Your plots correlate well with Innerfidelity's THD+N measurements of those headphones:
Grado SR60i
Sony MDR V6
Senn HD 650
Hifiman HE400i

Is this perhaps the signature of strong electro-mechanical resonance?
Trying to think through exactly what is happening at those bass resonances...

Or perhaps one other possibility is that the amplifier Innerfidelity use for the measurements has the same characteristics as the RME (or may even be a RME...) and all the distortion plots they have published may be influenced in the same way.

 

[amirm]

Your plots correlate well with Innerfidelity's THD+N measurements of those headphones:
Grado SR60i
Sony MDR V6
Senn HD 650
Hifiman HE400i

Is this perhaps the signature of strong electro-mechanical resonance?
Trying to think through exactly what is happening at those bass resonances...

I was wondering if that was the case for another reason. Namely, what Tyll is measuring may be his amplifier distorting just the same and not necessarily the headphone itself! If so, they need to get a benchmark amp.

Then again he is measuring way higher distortion (around 1% or so) so the amp one probably does not contribute much.

 

[amirm]

Thanks. Could you also measure the IEM output?

Sure. Will do a little later. Would need to do it at much lower output level though which would in turn raise the noise floor more.

I presume you don't have any Benchmark headphone amplifier lying around?

I don't unfortunately. We have one member, @dallasjustice who has their latest unit and he is visiting us in June. Until then, it sure would be good to locate another one to confirm their findings/setup.

 

[amirm]

Isn't the very act of measuring it at the connection diverting power into the measurement equipment?

That is a constant for the dummy load as well as the real headphones. The input impedance of my Audio Precision is set to 100,000 ohms so it has negligible impact.

 

[amirm]

Have these sorts of tests been done with amps connected to speakers?

I have never seen this test done. It is kind of inconvenient to run with speakers due to them creating tons more output. Would have to run the test when no one is home and wear earplugs. The frequency sweep is pretty nasty sounding.

 

[Dro]

I am fairly confident that you measure the distortion of the headphone, turned into an electric signal by the driver. You are still measuring around 110 dB SPL with 1 V RMS, so even the Hifiman might distort a little bit. Hence it would be interesting to test at a siginificantly lower level, say 95 dB SPL.

What I cannot explain is why different amps would measure so differently though.

The amps measure much better with the HD650, which will be a good bit quieter at 1 V RMS and does fairly well with distortion overall.

 

[amirm]

OK, let me get my acoustic headphone measurement gear up and running and I will calibrate to 95 dB using that, and then remeasure. I may also move this discussion to a more general thread to get everyone participating since it is a broader topic than just RME.

 

[dc655321]

Then again he is measuring way higher distortion (around 1% or so) so the amp one probably does not contribute much.

That was my intuition as well.

 

[Dro]

OK, let me get my acoustic headphone measurement gear up and running and I will calibrate to 95 dB using that, and then remeasure. I may also move this discussion to a more general thread to get everyone participating since it is a broader topic than just RME.

Much appreciated! I hope the noise floor won't be too high. You might get a better result with the IEM output.

Separate topic for this would make sense as well.

 

[Soniclife]

I have never seen this test done. It is kind of inconvenient to run with speakers due to them creating tons more output. Would have to run the test when no one is home and wear earplugs. The frequency sweep is pretty nasty sounding.

I'd really like to know if it affects speaker amps the same way, or if this is a problem for low output amps only.
Are the sweeps any worse than what you do with REW? I put my fingers in my ears when I run those.

 

[amirm]

Are the sweeps any worse than what you do with REW?

In this instance they are since it is a linear sweep. REW uses log-sweeps/Chirps so last much less time. Since benchmark uses linear sweeps in that test, I like to stick with that for now.

 

[pkane]

I was wondering if that was the case for another reason. Namely, what Tyll is measuring may be his amplifier distorting just the same and not necessarily the headphone itself! If so, they need to get a benchmark amp.

Then again he is measuring way higher distortion (around 1% or so) so the amp one probably does not contribute much.

Very interesting results. I wonder how a decent headphone amplifier would fare on these tests. I'm using a TEAC HA-501 with HD650's that has a switchable 'damping factor' control. Would love to see some measurements of this amp, especially with different damping factors.

 

[DonH56]

Late to this, sorry, have some personal stuff going on (wife's mother passed away).

Spread spectrum changes (spreads) the clock frequency so less energy is exhibited at any one frequency. Instead of appearing as a single line (spike, tone) in an FFT, it appears as a broad "hump". The usual reason is to pass EMC testing and that is probably why it is greyed out in the BIOS settings. Doubt it has any impact on audio signals, however, since they tend to use other references (e.g. run asynchronously to the system clock).

 

[amirm]

Very interesting results. I wonder how a decent headphone amplifier would fare on these tests. I'm using a TEAC HA-501 with HD650's that has a switchable 'damping factor' control. Would love to see some measurements of this amp, especially with different damping factors.

I will measure more amps and see how they do. Something tells me they all behave similarity in which case, we really need that Benchmark amp to confirm their results.

 

[RayDunzl]

Something tells me they all behave similarity in which case, we really need that Benchmark amp to confirm their results.

Looks like Benchmark uses the TI LME49600 as the headphone amp. Here's a DAC1 HDR:

https://benchmarkmedia.com/blogs/application_notes/37983617-the-hpa2-headphone-power-amplifier

http://www.ti.com/lit/ds/symlink/lme49600.pdf