MOTU M4 Loopback Measurements

[L0rdGwyn]

Here are some preliminary loopback measurements of the new MOTU M4. I purchased this device to do some measurements on current and future DIY projects, I will be pairing it with the Linear Audio Autoranger. Please bear in mind this is my first sound card audio measurement implementation, I welcome any feedback.

Here is the product info link: https://motu.com/products/m-series/m4/
Product manual with more detailed specifications: https://cdn-data.motu.com/manuals/usb-c-audio/M_Series_User_Guide.pdf


Photos:


The unit has two sets of line inputs, variable gain XLR-1/4" TRS (balanced or SE) combo jacks on the front panel, and fixed 1/4" TRS (balanced or SE) jacks on the back panel. It has two sets of line outputs on the back panel, RCA and 1/4" TRS (balanced or SE).

Quoted Specs

Front panel inputs:
XLR: THD+N -97 dB < 0.0014%
1/4" TRS: THD+N -100 dB < 0.0015%

Back panel inputs:
1/4" TRS: THD+N -106 dB < 0.0005%

RCA outputs: THD+N -108 dB < 0.00039%

1/4" TRS outputs: THD+N -110 dB < 0.00032%

Measured Output Voltage

SE RCA - 2.510Vrms
SE 1/4" TRS - 2.681Vrms
Balanced 1/4" TRS - 5.350Vrms

Loopback Measurements

Single-ended RCA output into TRS front panel input, gain adjusted to ~3dBFS.

Left channel

Right channel


Single-ended TRS output into TRS back panel input.

Left channel

Right channel


Balanced TRS output into TRS front panel input, no gain adjustment.

Left channel

Right channel


Balanced TRS output into TRS back panel input.

Left channel

Right channel


Results are somewhat inconsistent across the two channels. The noise floor of the left channel also rises at high frequencies. Overall seems to be a good showing given its price of $220 USD. I do have to wonder though if a different unit would show more consistent results across L and R channels. Interested to hear the thoughts of other members, thanks for reading.

 

[Blumlein 88]

According to specs sheet and manual it has 16 dbu max output on RCA and line outs. That should be about 4.9 volts. Unless you have measured it and found the spec sheet wrong (which it could be).

Also are the TRS inputs on the front panel switchable between line and instrument in the software? If not those have a 1 megaohm input impedance. For testing purposes the rear line in jacks will have a lower input impedance and slightly lower noise and distortion. So using those for testing should work the best. Your results indicate the best performance with those rear line ins.

 

[Deleted member 2944]

Did you perform a device calibration on your ARTA setup?

Dave.

 

[Blumlein 88]

This might make you feel better. This is a loopback on my Antelope Audio Zen Tour. Your best results appear to have a slightly lower noise floor and lower distortion. Now from testing various things I know most of the problem is the Zen Tour DAC which is okay, but not great. The ADC is much better. Still this is more expensive than your device with your Motu showing better results.

Same 64K FFT, I used Blackman Harris window as I didn't have a Kaiser in this software.

 

[L0rdGwyn]

According to specs sheet and manual it has 16 dbu max output on RCA and line outs. That should be about 4.9 volts. Unless you have measured it and found the spec sheet wrong (which it could be).

Also are the TRS inputs on the front panel switchable between line and instrument in the software? If not those have a 1 megaohm input impedance. For testing purposes the rear line in jacks will have a lower input impedance and slightly lower noise and distortion. So using those for testing should work the best. Your results indicate the best performance with those rear line ins.

I remeasured the max SE and balanced outputs with a DVM, I had mistakenly listed the -3dB values above, it has been fixed. The 16dBu spec seems consistent with the balanced output voltage, it is listed for both single-ended and balanced outputs in the manual though. I see no means to change the input impedance of the front panel connectors and only the 1Mohm spec is listed in the manual, so I assume they are fixed. That may be a good reason to go for the M4 over the M2, as the M2 only has front panel line inputs. I'll reach out to MOTU on both fronts to be sure.

Also, I forgot to mention, the measurements above were done with MOTU's WDM driver. The included MOTU ASIO driver is incompatible with Arta, and I experienced the same issue when attempting to use it with Virtins Multi-Instrument.

Did you perform a device calibration on your ARTA setup?

Dave.

Yes, of course. I recalibrated in Arta for each combination of inputs and outputs.

 

[Blumlein 88]

You should be able to use REW with ASIO.

Also if your Motu came with Ableton lite that will definitely work with ASIO. Just create yourself a 1 khz or other test tone. Play it via the Ableton, and also record the loopback, then use other software to analyze the recorded tone.

 

[L0rdGwyn]

Okay, I have REW up with the ASIO drivers. These are repeats of the best results from Arta above with the WDM drivers.

Balanced TRS output into TRS back panel input.
FFT Length 65536
Hann window
16 averages

Left channel


Right channel


The distortion figures agree quite well with the Arta measurements, although in REW I am now getting a pretty consistent 0.0017% noise.

Edit: the increased noise floor does not appear to be an artifact of REW or ASIO drivers, repeating the Arta measurements now shows the same, and across different devices (desktop, laptop). Curious.

 

[Blumlein 88]

I think for a bit over $200 those are pretty nice results. THD in one channel of just nearly -110 db, and remember that is combined distortion for both ADC and DAC. Not much worse in the other. I can't find a recorded file of my Scarlett 18i20 loopback, but all your results are at least a few db better than it.

 

[L0rdGwyn]

Sir, please, I am doing my best to be grumpy and dissatisfied here, you are making it difficult with your facts.

Kidding of course thanks for your input. Considering what I will be using this for, the results are stellar. It will serve me well until I need digital I/O.

 

[restorer-john]

Here's the Focusrite 2i2 V2 loopback, same parameters.

 

[restorer-john]

Here's the M-Audio Audiophile 24/192 loopback in my old computer (PCI card):

(997Hz)

Multitone loopback (TRS to TRS balanced):

This left channel is amazing.

 

[Severian]

I acquired one of these last week and there's no way that the RCA output voltage hits 4.9V; that must be a mistake in the manual. Based on my speaker system's behavior, I would have guessed that it maxes out at around 2V.

Thanks for the measurements! I have yet to record anything with it, but I've been quite enjoying it as a combination DAC/preamp/headphone amp!

 

[L0rdGwyn]

I acquired one of these last week and there's no way that the RCA output voltage hits 4.9V; that must be a mistake in the manual. Based on my speaker system's behavior, I would have guessed that it maxes out at around 2V.

Thanks for the measurements! I have yet to record anything with it, but I've been quite enjoying it as a combination DAC/preamp/headphone amp!

Yes, I think it is a mistake. I measured it at 2.510Vrms at full output. Glad you are enjoying it, nice little package for a good price

 

[Blumlein 88]

Here's the M-Audio Audiophile 24/192 loopback in my old computer (PCI card):

(997Hz)
View attachment 45882

View attachment 45883

Multitone loopback (TRS to TRS balanced):

This left channel is amazing.
View attachment 45885

View attachment 45886

Wow on the Maudio card. I had/have an Audiophile 96 version of that. It was not even close to this good. It had okay distortion, but the jitter on it was horrendous and raised the noise floor rather high across the whole audio band. Maybe it was defective.

 

[Blumlein 88]

Here are some loopbacks I did at 1 khz in REW. 32 k FFT size. As close to -1 dbFS as I could get. Lower levels of course are a bit better on distortion.

Focusrite 18i20
Somewhat high noise, pretty good harmonic distortion. From testing with other gear I know the ADC and DAC are about equal.

This is a Focusrite Forte. The DAC is a few db in distortion and noise better than the ADC.
If not for the 3rd and 5th it would be stellar on distortion. This unit has some quiet microphone preamps with up to 75 db gain.

The Antelope Audio Zen Tour. The DAC is weaker than the ADC. From other testing I know the ADC is lower noise and much lower distortion having low 3rd, and almost no 5th harmonic with all others below the noise floor. For a portable recording interface the ADC and microphone preamps are the important parts anyway.

 

[restorer-john]

Wow on the Maudio card. I had/have an Audiophile 96 version of that. It was not even close to this good. It had okay distortion, but the jitter on it was horrendous and raised the noise floor rather high across the whole audio band. Maybe it was defective.

That left channel was so good, I doubted my analog loop back was real, but pulling the loop real-time showed it was. I haven't tried a digital loopback but I fail to see how it would get much better. Clearly having the A/D and D/A synced together perfectly is the key.

I have 2x of the M-Audio AP (audiophile) 24/96 PCI and 1x Delta 410 (24/96) cards. You are right, they have some random jitter related issues. My three are all the same, so I doubt yours was defective. It's a pity because the THD was pretty good and for a single ended input/output and they are pretty quiet. Here's the 24/96 I had in a computer the other day:

I did get hold of the last driver for the cards and it improved the performance significantly, but I would still get random 'glitches' which I blamed on the OS. Also, a strange ~16KHz spike that varies from small to quite large. It's there on that loopback above, but at a low level.

 

[tuxaflux]

Hello People!

I need some advice for choose my next audio interface, some one can help me?

I’m really focusing on the quality recording for my synths.
So my choice is Motu M4 or Apollo Twin.
Both are excellent but the Apollo a bit expensive.
Do you think, their DAC in therms of quality recording are similar?
To be honest, I really love the Apollo but I pretty much don’t care about their plugins and UAD circle, and they justify the price around it, I’m correct?

 

[Minigun]

I wrote a review about the Motu M4 from a Linux user's perspective, along with some practical specs and info which I couldn't find anywhere else, if someone is interested: link to the review (my personal blog). Unlike apparently some people, I didn't have to do anything to my kernel which is a self-compiled version 5.4.15. The M4 worked right away.

I did some ARTA measurements in Windows and got about -106 dB THD and -103 dB THD+N using loopback via balanced connections. There are so many things to configure in ARTA, though, I'm not sure if the results are comparable. But I think it gives a good enough general idea.

I also measured latency in Linux using the same loopback setup. I was running JACK at 96000 Hz with 64 frames per period, 6 periods per buffer. I deemed this stable by running it overnight with some duplex audio data. I measured an RTL of 6.5 ms. The smallest possible RTL I could achieve (without immediate XRuns) was 4.5 ms @ 192 kHz.

 

[JohnYang1997]

Is there anyway for you to measure the adc and dac independently? I would like to know which one is the limiting factor. As well as the distortion with various levels.

 

[Minigun]

Is there anyway for you to measure the adc and dac independently? I would like to know which one is the limiting factor. As well as the distortion with various levels.

This would require some kind of reference card, such as a high-end RME. However, from the specs of Motu M4:

• DAC: ES9016 SABRE32 Ultra DAC
• ADC: AKM AK5552VN
• Preamp: THAT Corp 6263 preamplifier

The ADC has a THD+N of -106 dB and the DAC -110 dB.

If by various levels you mean volume levels, in my review I mentioned setting the potentiometers (mix and monitor) to 12:00 o'clock, which means something like 30 dB attenuation IIRC. It only had a 1 dB effect on the THD+N. The volume levels are controlled digitally in the ESS chip and therefore I think there's practically no quality loss.