Thank you for your answer. I'll put your tests in the next issue of the list of the results.

Has anyone had a chance to get any of the new converters released at NAMM?

* Audient iD48
* Ferrofish Pulse 8
* SSL 18
* Apogee Symphony Studio

I'd love to see the results for the SSL and Apogee especially.

The apogee should be the same as the symphony mk2 non special edition, which I think is listed in the thread.

The test needs to take into account two details to be useful:

1. Both DA and AD conversion stages have high pass filter stages in them. This does roll off measurable signal peak level quite a lot and can vary a lot between audio interfaces. The audible difference is non existent though, but can easily be measured.

2. The signal themselves between what goes out and comes in the audio interface can vary depending on the electronics themselves. I would assume the manufacturers would try to make the output/input levels match fairly closely, but I'm not sure they can vary or not.

Just some food for thought.

Those could skew the null test results quite a lot.

The level discrepancy between the original and its loopback copy is compensated for so that it has no impact on the test measurement. The corresponding gains to be applied to the left and right tracks of the copy are the two first figures of a test result.

The list shows the input DC filter on/off, but essentially also audio output usually does have DC filtering by design. So that will affect the results a lot when comparing output vs. input signals.

Out of band high pass as well as low pass filters are very much audible. Not because of the audio content they remove, but because of phase shift, passband ripple and other time domain related phenomena with psychoacoustic implications.

I have converters that allow me to choose filters and they do sound much better with DC frequency based filtering (simply adding or subtracting the proper amount of the DC in the digital domain is not a problem).

Also, real time high pass filters fully out of the audio band (e.g. only affecting signal above 20khz at 44.1khz samplerate) can drastically alter the perceived sound. Only filters with massive amounts of taps, requiring large processing windows and significant processing power impractical for real time use, can be made to be psychoacoustically neutral at the steepness required not to affect the audio band. So a compromise filter that actually affects the audio band but is less steep is preferable for real time audio (= converters) at low sample rates.

Finally, the currently preferred way of doing conversion, using delta sigma modulation at higher orders, has drawbacks in the real world because of the complexity of the high frequency feedback networks and susceptibility to jitter. And the frequency dependence of jitter, especially the effect of very low frequency jitter on audio, doesn't seem to have registered much with manufacturers.

Here are the files for my Cranborne Audio 500ADAT lunchbox (which integrates AD/DA converters). It was hooked via ADAT to the RME HDSPe AIO Pro audio interface. Channels 5/6 were used.

I've tested with both the internal clock of the 500ADAT and the clock of the RME. I made 3 runs of each (because why not).
Here's the link to the files: https://1drv.ms/f/s!AmeW3L6PgCIFhOVY..._d93w?e=30m1gu

500ADAT Int Clock - Run 1.wav
9.117 µs, 0.3518 dB (L), 0.3740 dB (R), -46.1318 dBFS (L), -46.8746 dBFS (R)

500ADAT Int Clock - Run 2.wav
9.117 µs, 0.3519 dB (L), 0.3741 dB (R), -46.1321 dBFS (L), -46.8748 dBFS (R)

500ADAT Int Clock - Run 3.wav
9.117 µs, 0.3517 dB (L), 0.3739 dB (R), -46.1322 dBFS (L), -46.8749 dBFS (R)

External clock tests to be measured later on.

500ADAT Ext Clock - Run 1.wav
9.119 µs, 0.3532 dB (L), 0.3751 dB (R), -46.1294 dBFS (L), -46.8728 dBFS (R)

500ADAT Ext Clock - Run 2.wav
9.118 µs, 0.3528 dB (L), 0.3748 dB (R), -46.1300 dBFS (L), -46.8732 dBFS (R)

500ADAT Ext Clock - Run 3.wav
9.117 µs, 0.3519 dB (L), 0.3740 dB (R), -46.1312 dBFS (L), -46.8741 dBFS (R)

To be added to the next issue of the

Would really like to see the results of the RME ADI-2/4 Pro SE.


Cheers!

Hi @ Diegel , can you please double-check your ADI-2/4 SE loopback recording? It shows nearly a 14dB worse null difference with mine under the same conditions with the same device. Comparing your frequency response to original2.wav and to my own capture, it appears you may have had some DSP enabled at the time. Here's the frequency response of your recording:



And here's mine:

See also the post here above from pkane.


Loopback tests requested by forum members: Eventide H8000FW (for validating or invalidating the test from jeamsler), Pacific Microsonics Model One and Model Two, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro, RME Fireface UCX II, SSL BiG SiX, Prism Sound Dream ADA-128, anything from Qes Lab, JCF, Burl and Linn, PreSonus Studio 192, iZ RADAR, Avid HD OMNI, Cranborne Audio 500R8, Fractal Audio Systems Axe-Fx III, NEVE MBC, Sound Blaster X7, Focusrite ISA 428, Elektron Octatrack MKII , Burl B2 Bomber ADC ---> DAC, dCS, Antelope Amari, Black Lion mods (e.g. Apollo X mods), Lynx Hilo 2, new (2024) MOTU 828, Forssell MDAC-2 ---> MSB Studio ADC as master . Of course any other one welcome!

Thanks, funny enough I couldn't see those result, but the moment after I posted I saw it! LOL

I have a question regarding the DC offset being turned off for the these test. Turning it on would score a lot lower, correct? If so will that be an issue when actually working with DC off?

Cheers.

Evaluating AD/DA loops by means of Audio Diffmaker


DC filters are audible. Turning them off can cause problems if the AD converter creates DC by default (most do since they work with an offset applied or drift through varying DC during normal operation) or your DAC causes DC by default or your signal already has DC on it and your following stage doesn't deal with it.

Thanks for that. Hmm so these tests would really make sense and more true by having the DC filter on as other convertors I take it have them on with no option to change that. If true these results are not really a fair comparison, right?

Yes, DC filters cause phase shift and may take away low end, resulting in an elevated difference signal with this test.

I don't think it's unfair to test a device using a configuration option that is provided, as long as this is documented. There are some loopback recordings in the list with a reported DC filter setting, and many that are not documented at all. We can't assume that all of the undocumented ones had the DC filter turned ON (or OFF).

RME ADI2/4 Pro SE has built-in capacitors to block DC offset at the ADC input. That's not defeatable. The option to turn off the DC filter is to disable the additional 0.5Hz digital high pass filter that is used to further reduce the DC offset to -130dBFS.

You can't turn off DC filters in most production units anyway. And the caps have the added benefit of protecting the converters from accidental destruction by phantom power.

An electrolytic blocking DC can be far less objectionable than a digital filter operating in real time. Or even a DC servo.

That's one of the reason why this test is ultimately rather limited in predicting the usefulness and (psychoacoustic) quality of a converter.

Right. Which is why I applaud RME for implementing a switch to enable/disable the digital DC blocking filter. I have it turned off permanently in my setup.

I also agree that the test, as implemented here, is limited from psychoacoustic perspective. If only because a single number can't fully account for audibility when differences over the entire bandwidth are included equally, and when frequency, level, and time masking effects are not factored in. The RMS null metric is a good engineering metric, but is insufficient to compare audible differences between any two devices.

This is good to know. Thanks. I was told from RME that the ADI-2/4 Pro SE is the best they do and upgraded from ADI-2 Pro FS R BE. I can see the results show this 👍

Would love to see Solid State Logic 18 results!

After seeing Julian Kraus measurements of the little brothers

https://www.youtube.com/watch?v=6zBe2tiD680

Loopback tests requested by forum members: Eventide H8000FW (for validating or invalidating the test from jeamsler), Pacific Microsonics Model One and Model Two, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro, RME Fireface UCX II, SSL BiG SiX, Prism Sound Dream ADA-128, anything from Qes Lab, JCF, Burl and Linn, PreSonus Studio 192, iZ RADAR, Avid HD OMNI, Cranborne Audio 500R8, Fractal Audio Systems Axe-Fx III, NEVE MBC, Sound Blaster X7, Focusrite ISA 428, Elektron Octatrack MKII , Burl B2 Bomber ADC ---> DAC, dCS, Antelope Amari, Black Lion mods (e.g. Apollo X mods), Lynx Hilo 2, new (2024) MOTU 828, Forssell MDAC-2 ---> MSB Studio ADC as master, Solid State Logic SSL 18. Of course any other one welcome!

Hi,

This is my first time contributing. I hope I did it ok.
Link below for Focusrite Scarlet Solo 3rd Gen

It only has a mono balanced line input so I had to do L & R channel outputs separately in Audacity and then used Audacity to simple make stereo version.

I reduced Audacity output volume to 97% (screenshots are in my zip file).

Focusrite input gain reduced a little on the device and output gain reduced a little on the device.

As you can see in screenshots, the Solo model doesn't allow adjusting gains in Control app.

https://drive.google.com/file/d/1g-W...ew?usp=sharing

You did not. Your recording is noise only.

2nd attempt looks better. Feeding balanced output into single balanced line input. Separate for left and right channels and then Audacity used to make stereo track from 2 x mono's
Recorded on Macbook, running on battery power, no connections to AC (best case setup, no ground loops possible)

Focusrite Scarlett Solo 3rd Gen
Link:

https://drive.google.com/file/d/1sDA...ew?usp=sharing

Thank you for your effort!

-153.671268 ms, 0.9545 dB (L), 1.0034 dB (R), -45.8006 dB (L), -44.7109 dB (R)

To be added to the next issue of the

So my RMS gain needs to come down a bit more?

1dB seems higher than everyone elses tests

Unfortunately the Solo model doesn't let me seen digital gain in software. I have to use the knobs. I think the bigger models do allow this

Not at all. Your test is perfect. There are a lot of tests in the list of the results for which the corrective gains are larger than 1 dB. For instance
Doing a new test at slightly higher level so that the corrective gains would be closer to 0 dB would not result in lower difference levels.

Thanks! It was actually easy in Audacity. I can see the waveform and any clipping clearly

Just hit Transport and Record As New Track

I will be able to test Tascam US-16x08 soon

Very affordable interface that has been around so long - surprised it hasn't been tested yet