Review and Measurements of RME ADI-2 Pro (comparison to ADI-2 DAC)

[Sythrix]

According to this, there are no differences between the Pro and Pro FS besides Auto Dark and new clock.

Exactly what someone who didn’t want a massive surge in returns would say. Only way to be sure that there’s no major difference would be to measure it. If they’re making the new boards, it makes sense to start using all the same analog output components as the DAC version. Why wouldn’t you put your better performing output stage on your top of the line product? I can’t see them honestly discussing it and deciding to leave it as is since they "had to work on the PCB anyway".

Unfortunately if they have this mindset, they will probably resist any attempt by Amir to exchange for the FS version.

Edit: @amirm Do you have any kind of return window or policy with them and the accommodation pricing?

 

[Dro]

True, they might be bullshitting here to not anger existing customers.

 

[amirm]

Edit: @amirm Do you have any kind of return window or policy with them and the accommodation pricing?

I don't know. I just sent them a message and will see what they say.

Meanwhile I looked in the manual for the new version and look at what I found:

It seems that the new clock definitely fixes the issue I saw with random low frequency noise. Their graph is much more zoomed than mine and hence the reason it still shows some of it.

 

[dallasjustice]

“I thought about the Army . . .”

 

[c1ferrari]

Hmm...very interesting results, Amir.
Thank you.

 

[nefilim]

Thank you Amir, interesting indeed. I literally just got one 2 days ago (new open box for $1575) - I'm using it to feed my Hugo2 (over SPDIF coax right now) and using it's internal DAC to feed my subwoofers (doing a crossover, time alignment & room correction in software).

I'm wondering if you had a chance to look at the digital outputs? Or the ADC section? Definitely curious how the ADI2-Pro shapes up as a measuring device (which they discuss in the owner's manual also).

Thanks!
Peter

 

[dir]

i was measure outputs of DAC and Pro via Adi-2 Pro Input (24/44 & 24/48 mode of RMAA)
they are both shows similar very light pulsing on the top of the frequency response

when i route RME's to the old LynxTwo input - it shows worse noisefloor of cource, but response are smooth

 

[XpanD]

i was measure outputs of DAC and Pro via Adi-2 Pro Input (24/44 & 24/48 mode of RMAA)
they are both shows similar 3rd harmonic domination and very light pulsing on the top of the frequency response

when i route RME's to the old LynxTwo input - it shows worse noisefloor of cource, but 2nd harmonic domination and smooth response

There's a reply to this here, now:
https://www.forum.rme-audio.de/viewtopic.php?id=27049

This deviation is well documented, even in several reviews of the Pro and its own manual. It is just a matter of zooming in. Every AD and DA has it , some more, some less, so depending on your zoom you notice it or not. This is from the AD chip itself and can not be changed.

 

[dir]

i know because it was my question
anyway i'm not satisfied by Matias answer
i use same scale with smooth lynx response so its seems not like just 'zooming' point of view

 

[svart-hvitt]

I guess this shows that femto is better than pico clocks?

Why doesn’t everybody use femtos?

 

[Wombat]

I guess this shows that femto is better than pico clocks?

Why doesn’t everybody use femtos?

Or attos, or zeptos or yoctos.

https://www.petervis.com/electronic...ico_nano_micro_milli_Kilo_Mega_Giga_Tera.html

There comes a point when enough is enough.

 

[svart-hvitt]

Or attos, or zeptos or yoctos.

https://www.petervis.com/electronic...ico_nano_micro_milli_Kilo_Mega_Giga_Tera.html

There comes a point when enough is enough.

My point was two-fold:

1) Why not implement whatever yields higher measurable performance (as long as cost is not a hurdle)?

2) «Objectivists» are - based on impressions I have from reading comments on the internet - skeptical towards femto clocks, which is subjectivism disguising as objectivism, cfr. point 1 above.

 

[Wombat]

My point was two-fold:

1) Why not implement whatever yields higher measurable performance (as long as cost is not a hurdle)?

2) «Objectivists» are - based on impressions I have from reading comments on the internet - skeptical towards femto clocks, which is subjectivism disguising as objectivism, cfr. point 1 above.

1. You should be justifying your point, not asking others to defend it.

2. An unsubstantiated generalisation.

Whatever, the ball is in your court to support your supposition.

 

[Soniclife]

Is the ADC side of this going to get tested?

 

[svart-hvitt]

1. You should be justifying your point, not asking others to defend it.

2. An unsubstantiated generalisation.

Whatever, the ball is in your court to support your supposition.

Let me lay out my reasoning once more:

Here we have a product before and after. The same product after new clock (and one other factor) has better measurements. So it is natural
to entertain the assumption that the better measurements are due to the new clock. After all, RME uses «fs» to brand the «after» product.

We don’t have that many examples of «before» and «after» in products, where the «after» product is identical except a couple of changes. «Science» wise, such examples are of interest for outsiders without insight into all of the design in a product.

 

[JohnPM]

i know because it was my question
anyway i'm not satisfied by Matias answer
i use same scale with smooth lynx response so its seems not like just 'zooming' point of view

The plots reflect different filter design choices. Some ripple in the passband can be a tradeoff for a narrower transition band for a given filter order (or length, depending on implementation). The magnitude of that ripple, the attenuation in the stopband and the width of the transition between the two are key inputs to the filter design.

 

[Thomas savage]

Let me lay out my reasoning once more:

Here we have a product before and after. The same product after new clock (and one other factor) has better measurements. So it is natural
to entertain the assumption that the better measurements are due to the new clock. After all, RME uses «fs» to brand the «after» product.

We don’t have that many examples of «before» and «after» in products, where the «after» product is identical except a couple of changes. «Science» wise, such examples are of interest for outsiders without insight into all of the design in a product.

I don’t thing we can really be certain therefore drawing the conclusion this performance enhancement was due to the clock and the clock only is ill advised imo.

It would be nice to have more information, find out precisely what brought the improvement into being.

 

[dir]

The plots reflect different filter design choices. Some ripple in the passband can be a tradeoff for a narrower transition band for a given filter order (or length, depending on implementation). The magnitude of that ripple, the attenuation in the stopband and the width of the transition between the two are key inputs to the filter design.

the point is that specific "signature" of the chip will now "draw" the graphs of measurements of other equipment ((

 

[Jimster480]

Well I think the only option is to exchange for the RME Pro FS and re-measure it again. Or maybe try to get a RME Pro FS and measure side by side to see the differences.
If RME really did make real changes then I would say that they would certainly be open to having Amir test and prove these changes yield better performance.

 

[svart-hvitt]

I don’t thing we can really be certain therefore drawing the conclusion this performance enhancement was due to the clock and the clock only is ill advised imo.

It would be nice to have more information, find out precisely what brought the improvement into being.

@Thomas savage

I understand your "we don't know anything" attitude when there's absence of solid facts. But in this case we do have facts.

What drew my attention was this quote from member @Dro :

"According to this, there are no differences between the Pro and Pro FS besides Auto Dark and new clock".
Source: https://www.audiosciencereview.com/...-pro-comparison-to-adi-2-dac.2682/#post-75929

According to RME: "I am surprised that Thomann treats this as totally new product, when there are only two smaller (minor) changes. For us it is a slightly updated ADI-2 Pro".
Source: https://www.forum.rme-audio.de/viewtopic.php?pid=129982#p129982

So I am not speculating here. I am basing my reasoning on RME's own words.

And because "fs" is featured on the front of the RME box, I think that may be the most important update of the two. A third possible change is the somewhat altered PCB board, but RME doesn't say that this change is material for sound or measurements.

In other words, we have pretty good reason to speculate that the femto clock is the most significant driver behind better measurements.

Interestingly, it seems like lots of "objectivists" are not very interested in clocks. Despite for example Paul Miller documenting a clock's contribution to significantly better performance in this review:

https://mutec-net.com/downloads/manuals/MUTEC_MC-3plusUSB_-_HFN.pdf

See the middle graph on page 2 of that review. It's the Oppo BDP-105D with and without reclocking.

According to Geoff Martin, "[t]hese days the weakest links in a digital audio signal path are typically in the signal processing software or the clocking of the devices in the audio chain". Source: http://www.tonmeister.ca/wordpress/2018/03/25/typical-errors-in-digital-audio-wrapping-up/

I think that statement is worth spending some time on.