Measurements of Motu UltraLite mk4 (AKM DAC Version)

[pos]

Nicely done!
Shouldn't this be part of the DAC?

 

[sfabio]

Nicely done!
Shouldn't this be part of the DAC?

I think so, but it seems that Motu, unlike RME, has not implemented it. Surely with another amplifier or less sensitive speakers the noise doesn't appear (I've tried with my Mosscade 502, deadly silent!). It's just a small detail that differentiates well-engineered products like RME or March-audio amplifiers

 

[ppataki]

I've read your post and made an RF filter like Alan - March audio suggests and... no more noise!
Here two wave files recorded at the mouth of my mid-highs horn, the background hum is my pc...
Motu before
Motu After
I've the Motu output directly connected to my NC500MP inputs, so the filter is very simple: one 5,6 Kohm resistor in series and one 680pF ceramic capacitor in parallel to both + and - signal.
Hope this helps

Thank you @sfabio !

 

[howard416]

I've the Motu output directly connected to my NC500MP inputs, so the filter is very simple: one 5,6 Kohm resistor in series and one 680pF ceramic capacitor in parallel to both + and - signal.
Hope this helps

What does this end up creating? A first-order low-pass?

 

[sfabio]

What does this end up creating? A first-order low-pass?

Yes, a first order lowpass at about 50KHz in my system

 

[howard416]

Yes, a first order lowpass at about 50KHz in my system

Thanks. Interesting that it would make an audible difference seeing as how the attenuation is so far above audible spectrum.

 

[sfabio]

Thanks. Interesting that it would make an audible difference seeing as how the attenuation is so far above audible spectrum.

It's an intermodulation issue like @March Audio explained here

 

[sfabio]

A little update. I have not studied electronics so forgive the inaccuracies; after doing some research I found an article that described the filtering of balanced connections and in which it was noted the importance of the tolerance of the components on the two branches of the line: at least 1% resistors and 5% capacitors, with the addition of a capacitor with a capacity of at least 10 times greater between positive and negative. I then changed the filter as per the scheme:

now the noise is totally inaudible up to a few cm from the horn mouth!

 

[AnalogSteph]

Well yeah, but have you calculated what the bandwidth of this filter is? I'm getting 1.37 kHz, which I dare say probably is a tad too aggressive if you want to suppress ultrasonics but pass audio. I am very much assuming that you do want to hear some highs.

Are you sure you don't actually have a gain staging problem as well?

What's the sensitivity of your horn?
Your amplifier is a NC500MP, gain ca. +25.5 dB, right? Minus the interaction of your filter + input impedance, maybe -3 dB or so. No external dropper resistors on the output, I guess?
Ultralite Mk4 FS output is 6.25 Vrms, noise is -112.9 dBFS @48 kHz, so probably ~-116 dB(A).
My gut feeling says you have way too much gain.

The maff says:
6.25 V = +6.88 dBW (ref. 2.83 V), so:
6.88 dBW
-112,9 dB
-3 dB
+25.5 dB
=
-83.5 dBW (~-86 dBW A-wtd)

That is an adequate output noise level for a speaker of normal sensitivity (up to 90 dB / 2.83 V / m), but probably not for anything horny (sorry).

At the same time, full-scale output on the MOTU would translate to +6.88 dBW - 3 dB + 25.5 dB = 29.38 dBW = 867 W. That is at least 6 dB above the output capability of an NC500MP, and I am very much assuming that your horn would need maybe 10 W for 110 dB @ 1 m tops, that's only +10 dBW. You're basically wasting the top 20 dB of the MOTU's dynamic range entirely.

I would suggest the following change to your filter:
Remove 10 nF
Parallel each 680 pF with a resistor of 620 or 680/681 ohms or whatever other value in the vicinity you have in stock in 1% or better tolerance, picking two closely-matching ones out of your stock with the multimeter. (Check your 5k6s as well.)

You will obviously have to increase digital levels for the horn in the crossover accordingly.

Your filter cutoff frequency is going to shift well into the longwave range but this may not be an issue. You can still put in something in place of your 10 nF again if need be, though I would suggest more like 2.2 nF (some sort of polypropylene film cap perhaps, FKP or MKP >63 V or so).

 

[sfabio]

Well yeah, but have you calculated what the bandwidth of this filter is? I'm getting 1.37 kHz, which I dare say probably is a tad too aggressive if you want to suppress ultrasonics but pass audio. I am very much assuming that you do want to hear some highs.

Are you sure you don't actually have a gain staging problem as well?

What's the sensitivity of your horn?
Your amplifier is a NC500MP, gain ca. +25.5 dB, right? Minus the interaction of your filter + input impedance, maybe -3 dB or so. No external dropper resistors on the output, I guess?
Ultralite Mk4 FS output is 6.25 Vrms, noise is -112.9 dBFS @48 kHz, so probably ~-116 dB(A).
My gut feeling says you have way too much gain.

The maff says:
6.25 V = +6.88 dBW (ref. 2.83 V), so:
6.88 dBW
-112,9 dB
-3 dB
+25.5 dB
=
-83.5 dBW (~-86 dBW A-wtd)

That is an adequate output noise level for a speaker of normal sensitivity (up to 90 dB / 2.83 V / m), but probably not for anything horny (sorry).

At the same time, full-scale output on the MOTU would translate to +6.88 dBW - 3 dB + 25.5 dB = 29.38 dBW = 867 W. That is at least 6 dB above the output capability of an NC500MP, and I am very much assuming that your horn would need maybe 10 W for 110 dB @ 1 m tops, that's only +10 dBW. You're basically wasting the top 20 dB of the MOTU's dynamic range entirely.

I would suggest the following change to your filter:
Remove 10 nF
Parallel each 680 pF with a resistor of 620 or 680/681 ohms or whatever other value in the vicinity you have in stock in 1% or better tolerance, picking two closely-matching ones out of your stock with the multimeter. (Check your 5k6s as well.)

You will obviously have to increase digital levels for the horn in the crossover accordingly.

Your filter cutoff frequency is going to shift well into the longwave range but this may not be an issue. You can still put in something in place of your 10 nF again if need be, though I would suggest more like 2.2 nF (some sort of polypropylene film cap perhaps, FKP or MKP >63 V or so).

Thanks for the hints,
I'm not an EE and I'm sure your math is ok, but I still measure flat 20Khz at the listening position, how it's possible if the filter bandwidth is 1,37 KHz? I've had no differences in measurements at the listening position with or without the filter I need to filter out frequency above 100KHz to prevent intermodulation with Ncore switching frequency (360-400 KHz)

 

[AnalogSteph]

Weird. Something in practice obviously does not match theory. (LTspice says -3 dB should be at 3.2 kHz or so - not sure where the discrepancy comes from, but <<20 kHz either way.) But as drawn, you've got the MOTU connected on the left and the NC500MP on the right, correct? And your 10n is in fact a 10n, right?

I looked up you compression driver + horn combo, turns out it's brutally sensitive. Around 114 dB at 1-2 kHz, slowly dropping to around 109 dB in the 10 kHz vicinity. So this is even worse than I previously thought. A bit of ultrasonic filtering won't go amiss (one may want to consider a series LC to form a notch filter), but you definitely need level attenuation more than anything else. Around 25 dB, depending on your maximum level needs. Noise contributions of MOTU and NC500MP should be about equal by the time you hit 17 dB, and 27 dB would definitely be diminishing returns territory, so something around 20-23 dB perhaps. Not even a +4 dBu output setting would have gotten you quite all the way there!

So I'd suggest 5.6k / 390-470R x2 or so, and seeing where that gets you. It may already eliminate your interference along the way, and if not, a 2.2n or 3.3n capacitor would.

 

[sfabio]

Weird. Something in practice obviously does not match theory. (LTspice says -3 dB should be at 3.2 kHz or so - not sure where the discrepancy comes from, but <<20 kHz either way.) But as drawn, you've got the MOTU connected on the left and the NC500MP on the right, correct? And your 10n is in fact a 10n, right?

Yes, correct. Maybe NC500MP input impedance (47K either input to ground) plays a role here?

I looked up you compression driver + horn combo, turns out it's brutally sensitive. Around 114 dB at 1-2 kHz, slowly dropping to around 109 dB in the 10 kHz vicinity. So this is even worse than I previously thought. A bit of ultrasonic filtering won't go amiss (one may want to consider a series LC to form a notch filter), but you definitely need level attenuation more than anything else. Around 25 dB, depending on your maximum level needs. Noise contributions of MOTU and NC500MP should be about equal by the time you hit 17 dB, and 27 dB would definitely be diminishing returns territory, so something around 20-23 dB perhaps. Not even a +4 dBu output setting would have gotten you quite all the way there!

Actually I've a BMS4594HE driver, even more sensitive! (16 Ohm drivers but I don't think it substantially changes the problem)
I need the MOTU because my system is 8 way active and in the near future it will become 10 ways, so Ultralite MK4 is perfect and cheap with near SOTA performance

So I'd suggest 5.6k / 390-470R x2 or so, and seeing where that gets you. It may already eliminate your interference along the way, and if not, a 2.2n or 3.3n capacitor would.

so if I correctly understand:

to create a divider that lowers the level, right?

 

[AnalogSteph]

Yes, correct. Maybe NC500MP input impedance (47K either input to ground) plays a role here?

That would do the exact opposite if it were even higher than I thought, so no. And if it had a very low input impedance like a few hundred ohms, you wouldn't have issues with hearing the grass grow in the first place. It's a mystery.

so if I correctly understand:

Yep, that looks good.

Should be a lot quieter than before. (About 20 dB with 470R and 22 dB with 390R, says my L-pad calculator.)

 

[ParadoxDelta]

Hi, I have a question regarding the multichannel capabilities of the Motu Ultralite devices (mk3, mk4, mk5, avb). Is it possible to configure any of them as e.g. 7.1 devices under Windows (not just in the Motu software) or are you bound to multiple stereo devices?

I want to set it to 7.1 in order to connect multiple amps to the different channels, each with the mirrored stereo stream plus individual EQ in EqualizerAPO. I'm currently using Voicemeeter Banana for the mirroring to multiple devices, but this solution seems rather unstable. Thanks in advance!

 

[ppataki]

You can use Motu's ASIO driver in Jriver - multichannel up to 7.1 will work that way
I was doing the same (but only 4.0) with my Clarett and my Apollo X4, now I am doing the same with Okto DAC8 Pro

 

[ParadoxDelta]

You can use Motu's ASIO driver in Jriver - multichannel up to 7.1 will work that way
I was doing the same (but only 4.0) with my Clarett and my Apollo X4, now I am doing the same with Okto DAC8 Pro

ASIO would strip me of the EQ capabilities of EqualizerAPO, if I'm not mistaken the mirroring to 7.1 way would be my prefered solution at the moment, so I wonder if the Motu devices can be configured as such under Windows.

 

[ppataki]

I don't know, maybe Motu customer service can answer that, or somebody here who owns such a device
If you by any chance get 'stuck' with ASIO you can use any EQ VST plugin in Jriver (that is what I do btw)

 

[1231rq32r1qw32r]

I've got a lot of MOTU gear, which tends to be looked down upon on other 'non-measurment' forums. Good to see that they are measuring reasonably well.

I'm running 4x 24I/O (got that 96IO BDE). Would love to see some testing of the 24I/O (the old one).