Motu M4 - Tear down, bit of internals analysis and few in-house measurements

[trl]

Wouldn't be easier to use an oscilloscope to measure the true RMS value? I know I use a cheap but good enough PicoScope for this.

The pots might be swapped with rotary switches, but I'm guessing some shielded wires might be used to connect the new SP4T to the motherboard, right? So the new pots will probably need to be left outside the case, otherwise I don't know how you could fit them inside the M4.

Looking forward for your answer, but in case you'll go with changing the original pots it might be indeed challenging and interesting as well.

 

[richlooker]

It's not like I'm using the M4 to measure true RMS value, I have a Fluke 287 for that. But I want to know that when Audiotester reads a signal at -6dBu, the input is actually -6dBu. If the input signal exceeds 0dBu, I want to be able to attenuate it in 10dB steps, and then just apply a correction of 10dB in Audiotester.

I will only do this mod if I can put the SP4T where the pots are now, and use the same knobs - no visible changes on the outside.

Including the leads, it builds 8.5mm behind the front plate. I'll have to measure the available space before I proceed, but the idea is to make a PCB that extends a little more to the sides than the current pots do, and place SMD resistors facing forward.

 

[richlooker]

Gain is impressively consistent between the channels. The difference between min and max gain is almost exactly 57dB in both channels. Measured with Tektronix ASG-100 audio signal generator as input, wired to the TRS inputs on the M4, and values read out with Audiotester 3.0.

Left input / Channel 1
Generator Gain Reading
-40dB min -61,41dBu
-40dB max -4,45dBu

Right input / Channel 2
Generator Gain Reading
-40dB min -61,42dBu
-40dB max -4,45dBu

@trl Did you by any chance see the resistance value of the gain pots?

 

[richlooker]

Considering the almost 60dB range with the original gain pots, I think I'll go for SP6T rotary switches instead, with 0 - 12 - 24 - 36 - 48 - 60 dB attenuation.

 

[MCH]

Couldn't it be done with an external attenuator?

 

[richlooker]

Couldn't it be done with an external attenuator?

It could, but I want an elegant solution. Plus, to preserve the 1Mohm input resistance with an external attenuator, either an active/buffered circuit is needed, or resistors in the many-megohm range would have to be used, opening for lots of noise.

I want to replace the gain pots with something that does not change the exterior and does not require any external stuff, that performs the same function over the same range, but with a few discrete steps instead of continuous adjustment. Either that, or nothing

 

[richlooker]

There's 10.6mm clearance between front plate and PCB.

Not very crowded on the PCB here, and soldering should be doable. According to measurements, there's room for a 14x14mm PCB to mount the switch and resistors. Diameter of the switch is 9.2mm. This should be possible...

I can see the marking on all pots except the gain ones; they are all B taper 10K. Assuming for now the gain pots are also 10K.

 

[richlooker]

So there's a 10K pot that adjusts a gain range of 57dB. In order to replace this with a rotary switch giving discrete, equal-dB steps, I need to understand the configuration of the gain control. The block diagram indicates it not a buffered passive attenuator; the pot controls the actual gain of an amp stage:

How to dimension resistors correctly depends on how this gain control is implemented. Is it an inverting or non-inverting opamp? Is the pot adjusting the feedback, the coupling or feedback to ground, the input resistance, or something else?

 

[AnalogSteph]

If it's really a pot rather than a mechanical encoder, I would assume it's being used to tap off a varying percentage of a DC voltage which is then filtered and fed into an ADC (8 bits as commonly available on micros would do and provide 256 discrete values, in which case I'd guess it would be a linear pot). Prototyping may involve fiddling with a bunch of pots to get the set points just right, followed by playing with resistor ladder values in simulation.

Gain control itself is probably a hybrid of THAT6263 control in 3 dB increments + digital gain adjustment for the intermediate steps plus extension on top (the chip itself only has a 42 dB gain range, -8 to +34 dB).

Looks like M4 engineering was contracted out to these guys. Interesting.

 

[richlooker]

Thanks @AnalogSteph

I have just established that the exact middle of the gain range - approx 28.5 dB up from min / down from max - is also the middle position of the pot, and the gain seems to be quantized at approx. 1dB steps. Which means i should - instead of a logarithmic ladder - divide the 10K of the pot evenly among the switch steps.

Or, even that exactly 10K is not required, as the pot is used as a voltage divider. This makes things easier. An SP7T switch and 6 x 1.6K resistors will do the trick. That should get me something like 0, -9.5, -19, -28.5, -38, -47.5 and -57 dB steps. I may even be able to solder SMD resistors directly between the pins of the switch; will definitely try that.

 

[richlooker]

Actually, I can do better than that. I have established that in the min gain position, the signal is attenuated by 17dB. Which means I can get the steps -17, -10, 0, +10, +20, +30 and +40 dB. A 1.4K resistor for the first 7dB step and five 2K resistors for the 10dB steps.

I have ordered parts; 2 subminiature SP7T switches plus SMD resistors in two sizes: 3.2x1.6mm and 2x1.25mm; I'll have to see if any of these can be soldered directly to the switch pins, or whether I'll need to make PCBs. To be continued.

 

[nowonas]

The front display is fancy and eye catchy, a very good tool to get an idea about the input/output levels, although the OLED VU-meters are kind of slow, much slower than regular analogue lights from, let’s say, Scarlett interfaces. They’re displaying more of a top-averaging of the SPL of the sound, instead of rapid peaks, but if clipping occurs then the red light stays lit on top of the display for few seconds, so there should be no problem in getting a good recording after all.

Are you sure that this is an OLED display and not a normal LCD display?
Does it have a manufacturer's name and if so, what does it say?

 

[richlooker]

I have measured and verified that @AnalogSteph is right; there is 3.3VDC across the potmeters, and the center tap varies linearly between 0.0 and 3.3.

@nowonas : This is definitely not an LCD. There are no visible markings to identify the display, though.

This is going to be an interesting challenge. I have to;

• Make a PCB that holds 6 SMD resistors, and that will actually fit in the available space
• Find a way of adding the equivalent of the two stabilizing/mounting pins on the pots; it will have to withstand a bit more torque than the pot
• Adapt the 4.2mm shaft of the rotary switches to the 6mm knobs

This is BTW my first foray into SMD stuff. I can't believe how tiny these resistors are...

I have ordered a couple fine-line lacquer pens, a UV lamp and some photo resist PCB. Will try to draw stuff by hand first, and have the UV exposure / photo resist as plan B.

 

[MWKMWK]

I just got an M6 and I've been trying to find others who have experience with it, the M4 or the M2 to check something for me. The gain knobs on my unit are making a crackling noise when I increase/decrease the gain volume on a given channel while a signal is being fed into it and monitored. This makes me think they are digital encoders and not analog potentiometers. Any reason why the volume isn't smoothly increased/decreased? Would love to hear from you all if you are experience a similar issue with your units. I attached a file to for sample.

Wouldn't Motu be able to program some sort of code in the firmware to smoothen the transition in volume if they are digital encoders? Shouldn't it be possible for them to remove the 'stepping'?

Audio file.
http://sndup.net/vxf3

 

[trl]

Interesting finding, feel free to read the first page of this review, maybe you can figure out what is causing this, but let's also review the volume control chips THS4522 and THAT6263 which might be causing this "timber pitching" sound. However, I would say that impact is nil if you properly adjust the input volume controls prior to beginning the recording, although adjusting it slightly should not create any audible changes in sound. From what I could hear in your above example this only happens when you max out the volume quickly or when you revert it to the min, right?

 

[-Matt-]

I just got an M6 and I've been trying to find others who have experience with it, the M4 or the M2 to check something for me. The gain knobs on my unit are making a crackling noise when I increase/decrease the gain volume on a given channel while a signal is being fed into it and monitored. This makes me think they are digital encoders and not analog potentiometers. Any reason why the volume isn't smoothly increased/decreased? Would love to hear from you all if you are experience a similar issue with your units. I attached a file to for sample.

Wouldn't Motu be able to program some sort of code in the firmware to smoothen the transition in volume if they are digital encoders? Shouldn't it be possible for them to remove the 'stepping'?

Audio file.
http://sndup.net/vxf3

Crackle noises can also be caused by selecting too low a buffer size, try a larger buffer and see if they go away.

(Just something for you to try).

And crackly analogue pots are very common (caused by intermittent contact between the wiper and plates) not sure why this would suggest a digital control to you.

 

[MWKMWK]

Crackle noises can also be caused by selecting too low a buffer size, try a larger buffer and see if they go away.

(Just something for you to try).

And crackly analogue pots are very common (caused by intermittent contact between the wiper and plates) not sure why this would suggest a digital control to you.

The issue is related to the audio interface alone and when I monitor directly from the unit. The 'stepping' shifts and noise heard is present when I slightly increase/decrease of the gain volume of that channel.

I've uploaded another file https://sndup.net/hxnf/ In this example, I'm feeding a constant tone and any noise or crackle that you hear is caused by the change in gain from the unit. This is a new unit and it's the second unit I tested. If it were a dirty analog potentiometer, wouldn't the noise be more random and inconsistent? This noise (which is why I think it's a digital encoder) is creating a 'stepping' or clipping noise at a specific rate dependent on the speed at which I increase or decrease the gain. I'm no expert when it comes to this so I can't know for sure, but I'm hoping you all can help determine the cause of this and how I can contact Motu to get it fixed (or at least, maybe there's something they can do to the firmware that would resolve that harshness of the stepping and smoothen it out). That way this audio interface can also be used as a mixer in live scenarios. Thanks for your comments and help.

 

[chaeyoung]

Cool,
Thanks for this well explained deep dive of the motu m4!

I can now rest easy having bought it.

Also I compared it to the ifi zen DAC and found the difference to be staggering for only double the price you have; 2 more outputs; midi connectivity and for 4 inputs (it is miles better than the zen DAC too).

cheers!

 

[AnalogSteph]

Wouldn't Motu be able to program some sort of code in the firmware to smoothen the transition in volume if they are digital encoders?

The problem is that the steps are generated in the actual input amplifier with its 3 dB gain step size, in the analog domain. They would have to determine how long the delay between being commanded and the actual switching is, and how many samples later it comes out in the digital domain at a given sample rate - assuming that value is constant at all. Of course, at this stage the step has already been lowpass filtered by the ADC anti-alias filter, too.

This may be asking a bit much from the DSP capabilities of what I assume is some sort of XMOS job. Coordinating a PGA and digital section in such a way is a feat best performed within a single chip - see ADCs and DACs with DRE / ARE - but even then it's fairly advanced and only a handful of makers have mastered it (AKM and TI/BB come to mind).

 

[MWKMWK]

The problem is that the steps are generated in the actual input amplifier with its 3 dB gain step size, in the analog domain. They would have to determine how long the delay between being commanded and the actual switching is, and how many samples later it comes out in the digital domain at a given sample rate - assuming that value is constant at all. Of course, at this stage the step has already been lowpass filtered by the ADC anti-alias filter, too.

This may be asking a bit much from the DSP capabilities of what I assume is some sort of XMOS job. Coordinating a PGA and digital section in such a way is a feat best performed within a single chip - see ADCs and DACs with DRE / ARE - but even then it's fairly advanced and only a handful of makers have mastered it (AKM and TI/BB come to mind).

So, I should accept the current behaviour as normal and that there's not much they can do 'practically' to make the knob behave more like a normal smooth volume dial, correct? This stuff is a bit beyond me so I was thinking that maybe it was a simple tweak/fix on their end, and that they just didn't really consider it since they weren't making a mixer but instead an interface (where you adjust your gain once or twice and move on to the DAW). I'm always finding different ways to use my gear and always reveal these types of problems as I go. Maybe I'm the one needing tweaking/fixing, haha! Thanks for your input!