• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

RME ADI-2/4 Pro SE - RIAA mode measurements

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
2,682
Likes
4,220
Location
Liège, Belgium
Hi

here are some measurement of the new RME ADI-2/4 Pro SE in RIAA (phono) mode

The RME ADI-24 Pro SE is RME's last addition to their ADI-2 converters range.

RME explains that SE means "Special Edition", and they advertise it as a Special, Limited edition, on top of the existing ADI-2 serie.

Externally, the family relation with the ADI-2 range can't be denied.

_R9A6153_DxO.jpg


Here-above, from top to bottom,
RME ADI-2 Pro fs BE
RME ADI-2 Pro fs R BE
RME ADI-2/4 Pro SE

The most obvious difference with the ADI-2 Pro fs (R) is the presence of an additional 4.4mm Pentacon connector, to allow direct connection of balanced phones.
On the rear, there is also a Trigger output, to remotely switch on a power amplifier, as an example.

The new ADI-2/4 is also deeper than the ADI-2 pro (160mm vs 130mm) and it includes those large "HiFi look" silver feets instead of the small rubber feet of the Pro,
In fact, it shares more of the look of the Audiophile-targeted ADI-2 DAC fs.
For functionalities, it is of course closer to the Pro, with its 2 stereo DACs and the inclusion of a stereo ADC.


Another difference is the new Power Supply, which, while still an external 12V DC PS with a lockable connector, is now more powerfull and grounded.


Internally, we read that the DAC is using ESS chips, similar to the one used in the ADI-2 DAC
and a new ESS ADC (the previous Pro were all using AKM ADC) with improved performance.

Also, the selectable DAC and ADC ranges are modified:
The previous "Pro" allowed a selection between +4dBu / +13dBu / +19dBu /+24dBu ranges,
The new ADI-2/4 has now +1dBU / +7dBu / +13dBu / +19dBu / +24dBu
That's one more range.

On top of this, the ADI-2/4 has a new RIAA mode "for directly digitizing vinyl discs"

This is not really a Phono Preamp.
But given the low noise level of the ADC, RME seems to think that the performance of the RIAA mode could compete with some very good phono preamp.

Let's find out if this is true.

Measurements below were performed using the RME ADI-2/4 Pro SE as a generator, Behringer Monitor1 and Shure A15AS to attenuate the level, and Virtins Multi-Instrument 3.9.7 software.


RME ADI-2/4 Pro SE - RIAA Mode

The RIAA mode can be selected from the Analog Input Menu.
It then allows selecting an additional "Gain" and, of course, will activate RIAA equalization / correction.

Here are the specs, copied from the manual

RME ADI-2-4 Pro SE - RIAA Specs.png


To compare with other measurements here, I chose 5mV / +26dB
Here is what the SINAD / Noise Dashboard looks like at that level

RIAA Mode SINAD / Noise Dashboard 5mV

RME ADI-2-4 Pro SE 1dBu Mono - Monitor1 -19.6dB Shure AT15AS -25dB ADI-2-4 RIAA +26dB L_crop..png


80.8dB SNR
(and we get the same figure for SINAD: distortion is way below, as it measures 23dB below the noise at that level)

This is a pretty respectable figure.
We have 15dB headroom above 5mV, which is probably more than enough.

Measuring the other "Gain" settings for their respective specificied sensitivity, I got this

RME ADI-2-4 Pro SE - RIAA Measurements.png


This pretty well matches the Specifications.

We see the SNR decreasing with the test signal.
(This most probably is partly due to the noise in the test signal, as the level is lowered)


If I measure all settings at 5mV, I get this:

RME ADI-2-4 Pro SE - RIAA 5mV Measurements.png

The SNR remains the same for +20dB up to +38dB.
That just means that the difference in "gain" is purely done in the digital domain.

But for +14dB, you see a lower figure.

There is a distinct relay-switching sound when you go from +14dB to +20dB mode (and none above)
This means that some analog level stage is modified between those 2 levels.
This is not just digital compensation.
And we see this in the measurements.

RIAA Mono Bass
RME includes a setting called "RIAA Mono Bass"
The idea is that, on a vinyl, bass is mono anyway.
Therefore, noise may be decreased without loss by just averaging both channels for low frequencies.

Indeed, we can even measure some benefit on the ADC side.
(NB: a much more dramatic benefit can be expected on the Vinyl noise itself)

RME ADI-2-4 Pro SE 1dBu Mono - Monitor1 -19.6dB Shure AT15AS -25dB ADI-2-4 RIAA +26dB L Mono B...png



RIAA Mode - Frequency response after RIAA correction

One of the most critical measurements is the RIAA correction accuracy.

To perform this one, I went back to the
original RIAA correction formula

RIAA formula.PNG

(Where t1=3180µs, t2=318µs, t3=75µs)
I then computed a compensation curve, including a point for each of the FFT bins, for Virtins Multi-Instrument.


I usually perform Frequency response measurements with REW generated Periodic noise at 192kHz tuned for 128k FFT.
I use the same +/-5dB scale that Amir is using.


RME ADI_2-4 Pro SE RIAA +26dB 5mV - FR 192kHz +- 5dB range.png


This looks like text-book perfect to me.

But RIAA compensation the way it's done here may have a different accuracy for different Sampling rates.
So I compared the correction for each sampling rate (44.1kHz, 48kHz, 96kHz, 192kHz) and compared with the flat Frequency Response of the ADI-2/4 DAC+ADC with no RIAA correction at 192kHz - In black in the plot.

RME ADI_2-4 Pro SE RIAA +26dB 5mV - FR Comparison +- 5dB range.png


Now we see a very (VERY) small amount of difference around 50Hz between the sampling rates.

Let's zoom in a bit:
(Warning: This is a +/- 1dB scale - a microscope look)

RME ADI_2-4 Pro SE RIAA +26dB 5mV - FR Comparison +- 1dB range.png


We see that the 192kHz sampling is slightly more accurate.

OK, that was fun, but let's be serious: there is absolutely NO audible difference between those correction curves.


RIAA Mode - Distortion vs Frequency at 5mV (90kHz BW)

I added the input level (dashed blue), so you may understand why the distortion is higher in lower frequencies: that's because the level is.
NB: This is measured at a -relatively- high level of 5mV.

RME ADI-2-4 Pro SE - RIAA +26dB 5mV - THD vs Frequency.png



RIAA mode - Conclusion

When I first read about the addition of the RIAA correction, I thought it was kind of a marketing gadget.
Now after measurements, I have to say this is quite serious.

If you try to rank this with Amir's ranking, here is where it would be located versus the very top of the chart.
Black arrow is where the "standard RIAA" mode +26dB result of 80.8dB would land,
The red arrow shows the "RIAA Mono Bass" mode ranking.

best budget phono stage review Ranking.png



On top of the very serious performances we see here, we also have the powerfull parametric EQ at hand to fine tune any cartridge-specific FR difficulties.
And the benefits of the "Mono bass" mode are probably much more dramatic on the reading noise of an actual vinyl.

So no, it's definitely NOT a gadget.

Well done !


EDIT August 2023:

RME just published a very informative video on how this works here.
Given the technical details they added, I thought it was worth adding the link here.
 
Last edited:

morillon

Major Contributor
Joined
Apr 19, 2022
Messages
1,336
Likes
263
what about the input impedance?
if faced with a mm card, we most often consider a 47k
(see also the capacitance in mm... low we can "play" but too high we are blocked)
;-)
 
OP
R

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
2,682
Likes
4,220
Location
Liège, Belgium
what about the input impedance?
if faced with a mm card, we most often consider a 47k
(see also the capacitance in mm... low we can "play" but too high we are blocked)
;-)
I didn't measure it.

From the specs, we can read
  • Input impedance: 45 kOhm @ 1 kHz, input capacitance 150 pF
 

G-K

Member
Joined
Jan 31, 2022
Messages
51
Likes
7
Let's zoom in a bit:
(Warning: This is a +/- 1dB scale - a microscope look)

A scaling of +/- 1 dB means +/- 12.2 % difference. How can you call it microscopic look?
 

MaxwellsEq

Major Contributor
Joined
Aug 18, 2020
Messages
1,628
Likes
2,426
Great measurements, thank you
 

G-K

Member
Joined
Jan 31, 2022
Messages
51
Likes
7
This is audio and by eyeballing the graph it looks pretty good, don’t you think?

This a measuring of an electric signal in a frequency band between DC and 100 kHz AC. However its a loopback measuring and we dont know where the differences are coming from. Either from ADC or DAC. I didnt say that it will be audible. If we would look at +/- 0,1dB (approx 1%) then we are far away from a rular flat response. If we will look for measurements and look what is only audible, then we dont need to measure distortions above 10kHz fundamental frequency, because most of humans are not able to hear
 

Trell

Major Contributor
Joined
May 13, 2021
Messages
2,752
Likes
3,285
This a measuring of an electric signal in a frequency band between DC and 100 kHz AC. However its a loopback measuring and we dont know where the differences are coming from. Either from ADC or DAC. I didnt say that it will be audible. If we would look at +/- 0,1dB (approx 1%) then we are far away from a rular flat response. If we will look for measurements and look what is only audible, then we dont need to measure distortions above 10kHz fundamental frequency, because most of humans are not able to hear
The interesting part in the graph is from 20Hz to 20kHz, so I don’t quite understand why you bring up 10kHz or 100kHz.
 
OP
R

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
2,682
Likes
4,220
Location
Liège, Belgium
A scaling of +/- 1 dB means +/- 12.2 % difference.
The full "zoomed" plot scale is +/-1db.
Here, we see less than 0.05dB variation in total.
That's completely impossible to hear.

NB: Amir plots are +/-5dB, like my non-zoomed ones.
This +/-5dB scale is more relevant, on an audibility point of view.
 
Last edited:

G-K

Member
Joined
Jan 31, 2022
Messages
51
Likes
7
We do.
That's why I added the loopback without RIAA for comparison.
Your graph shows a loopback curve for 192 kHz sampling rate (the black curve if l interprete correctly), so its only valid for the 192 kHz measuring curve. By the way, for all of the other sampling rates the frequency response curves are not normalized to the level of 0 dB at 1 kHz. I think no pick-up system or venyl record will deliver such a "flat" frequency response. In that case the RIAA frequency response is perfect.

index.php
 

Newman

Major Contributor
Joined
Jan 6, 2017
Messages
3,449
Likes
4,211
Compared to Puffin?
 

Mnyb

Major Contributor
Forum Donor
Joined
Aug 14, 2019
Messages
2,641
Likes
3,618
Location
Sweden, Västerås
I’m also impressed that RME only chooses to add this feature when they could realise some advantages that’s rare in other riaa implementations.

Not just extending the feature list .
 
Top Bottom