Microphone-Measurement-System (Open Source)

[juliusbusch]

Hello folks.

I am currently planning a somewhat larger open source project and would appreciate feedback on my thoughts.
In order to measure my self-built microphones and to generate measurement data that is as close to the standards as possible, I would like to build a platform that is as adaptable as possible.

Basically, there should be three modules:

1. analogue front-end with:

1x microphone input​

1x line input​

1x line output​

1x amplifier output​

2. digital calculation stage with motor control:

1x Teensy 3.6 or 4.1, which is accessible via​

1x USB-B on the enclosure and the​

1x SD card slot of which should also be available from the outside.​

1x Teensy Audio Board serving as ADC and DAC​

1x stepper motor driver with output via a CAT connection​

1x screen​

1x rotary encoder for software control​

3. power supply unit with Powercon connection

1x transformer with linear generation of +-15V, +48V, +12V, +5V​

After I have already planned a few solutions in Eagle and Fusion, I would like to present two variants for discussion.


1. a very (perhaps too) compact unit with three boards on top of each other in a "Fischer Elektronik KOH 6100 housing kit". Here, all the components are very close to each other. I am particularly concerned about the proximity of the motor driver to the analogue electronics. In addition, there would be hardly any space to accommodate all the connections. I will only use Neutrik connectors. I would have to install an XLR-TRS combo jack in this cabinet, which would have to be switched via a relay either to the microphone input with phantom power or to the line receiver. Close to it would be the Powercon connection and the CAT connector for controlling a stepper motor.
On the front of the unit, there would be little room for the SD card slot, rotary encoder, display and USB port.
This means that everything would be very cramped in the housing. The transformer in particular is difficult to accommodate.

Alternatively, the entire electronics could be accommodated in a 19-inch 2U enclosure in a very relaxed way. Here, dedicated XLR connections for separate line and microphone inputs could be realised. The same goes for the line output and amplifier output. The assemblies would all be further away from each other and there would be enough space.
This solution would be more comfortable to plan and build, but incomparably more expensive, due to the large enclosure that is now necessary.

I can't decide. What do you think?

Here are a few more details.
For the microphone amplifier I plan to use the THAT 1512 chip with a fixed 0 dB gain. However, for all those who absolutely want to set a gain, the possibility to connect a corresponding potentiometer or a rotary switch is to be prepared on the board.
The THAT 1240 chip is planned as the line input receiver. Here, too, 0 dB gain.
Of course, the THAT 1606 chip is now also used as a line output driver. Once again without gain.
The loudspeaker output is to be realised via an XLR connection. The amplification itself I want to implement with a small 5 Watt class-D amplifier. I am thinking of a very compact module from the Chinese market. It is based on the PAM8403 chip.
The motor driver is a Trident Silent Driver whose functions should all be controlled by the Teensy. However, it will hardly be necessary to use these additional functions. Even microstepping will hardly be used.
I am not yet sure about the screen, but I am thinking of a colour display that is as large as possible.
The unit should actually also be usable as a standalone, but I like the idea of not attaching a power switch to the unit, but instead switching the power via a relay that would be supplied via the USB bus voltage... However, the idea is more of a snapshot and stems from the space-saving considerations for variant 1.

The device should work with REW and possibly take over all its functions independently in a later software version.
Measurement data should then always be stored on its own SD card and also sent to a PC connected via USB.

These are my thoughts.

Greetings from Germany
Julius

 

[Tubetec]

Whats the purpose of the motor ?

 

[juliusbusch]

Whats the purpose of the motor ?

Is for a fully automatic turntable for measuring the mic's polar pattern.

 

[rock soderstrom]

This sounds like a cool and ambitious project. I can't help you with it but I keep my fingers crossed that you can realize it successfully. Keep us up to date.

 

[Tubetec]

Ahhhh I see .
Any ideas about how to create a sound source of equal magnitude across the audible range ?
I notice B&K use a pistonphone directly coupled to the mic capsule to calibrate sensitivity , but these are single fixed frequency transducers .

 

[12afael]

I don't think you need a perfect source but a good reference. You need to do some IR compensation for the room/box speaker source and the reference. Substracting all that you get your freq response.

 

[RuudNL]

That is how it works in REW. You do a reference measurement with a calibrated microphone and the speaker you are going to use for the measurements, and from that moment you have a reference measurement.
When measuring a microphone under test, you subtract the reference values from the measurement data.
I doubt if a rotating speaker would be useful for DIY measurements, because nobody has a completely 'dead' room.

 

[iampoor1]

I doubt if a rotating speaker would be useful for DIY measurements, because nobody has a completely 'dead' room.

What about bringing it into the middle of a flat open field to make the measurement?

 

[RuudNL]

I don't see myself doing that. Besides that: wind noise, traffic noise etc.

 

[Tubetec]

How about the Bruel & Kjaer pistonic tube methodology ?
A suitable headphone driver with a steel tube coupling it to the capsule ,
Gets rid of the room from the equation at least for an on axis measurement .
I havent taken a B&K Pistonphone apart so I dont know whats inside it , maybe piezo
or dynamic , Im not sure.
Some kind of acoustic resistance at the back of the capsule and drive unit end needent take up an entire room ,just a small box .

Presumably it would be the mic that was rotated , not the sound source , otherwise we might need to re-cal due to speaker positioning within the room itself ,thats not say things wont vary to some degree if the mic istelf is moved in relation to the room , unless its anecoic .

 

[Tubetec]

Its the on axis response of LDC's were primariliy concerned with in this case , thats a thing we can do something about when we choose the electronic circuit , by lending a favourable eq curve . Off axis response is mainly due to capsule 'porting' and headshell , ie your stuck with it unless you change the mic parts . Once you've chosen a donor body and a capsule the off axis is more or less a done deal .

 

[juliusbusch]

So, I got an anechoic chamber at my proposal and I built a rotating platform for the mic, not the speaker. To get a full range single point sound source is not easy, but subtracting a reference measurement in REW is. So I will use a just good point source speaker.

 

[ccaudle]

How about the Bruel & Kjaer pistonic tube methodology ?

Because the tube couples to the front side of the capsule only, that method is only for calibrating SPL for omni microphones. The piston moves at a single frequency (it is a mechanically coupled piston which moves on a small cam):
B&K pistonphone calibrator info
https://www.grasacoustics.com/products/calibration-equipment/reference-calibrator/product/255-42aa

So useful for SPL meters, but different use than described here.

 

[JohnRoberts]

This a picture from one of my dad's old notebooks. This is from an article about "good" loudspeakers printed in 1929 in a Bell Labs journal. Back in those days they called microphones "transmitters"....

Looks like the ancients are still stealing modern ideas...

JR

 

[swpaskett]

Sounds like a fun idea. Yes, it is a lot easier to rotate the mic than the speaker. I tried doing something similar, testing mics in my backyard at 3AM looking for some particular characteristics that we hoped were going to resolve some issues with a project at work. I found that even in the middle of a large yard I got too many reflections and a suburban neighborhood was still not quiet enough to get repeatable measurements, but it worked better than the antenna chambers we had at work which were "dead" to RF but not to audio.
It's just hard to beat a good anechoic chamber sometimes. I still wonder how large a chamber has to be. The one we had at school was a 20x20x20 foot cube, real luxury for a 20-something kid with interest in acoustics (then I went to work for a radio company, go figure). But does it need to be any bigger than large enough for a full range speaker, a microphone, and a test stand? Are there other minimums? For RF there are, but for audio? Would a 4 foot cube (64 ft^3) be big enough, assuming the absorber may eat up all but 8 ft^3? I could jam a 4" speaker, a little test stand with a stepper motor, and a couple nominally sized microphones in that space. I won't be testing an A7 in there (dating myself terribly -- ever see an A2?), but microphones and teeny bookshelf speakers (maybe?).
I suppose you can get too small and reflections off what you are testing would start to impact the measurements.
And while I am abusing the privilege of commenting, why is acoustic absorber so blamed expensive? RF absorber is loaded with carbon, but acoustic absorber, isn't it just foam?
Sorry, i got carried away (again).

 

[PureBasic]

Hello Julius,

Here are my thoughts:

Basically, there should be three modules:

1. analogue front-end with:

1x microphone input​

1x line input​

1x line output​

1x amplifier output​

Yes! Also have at least one buffered mic output to be able to make measurement with a calibrated RMS voltmeter. That's making the whole chain calibration easier.

2. digital calculation stage with motor control:

1x Teensy 3.6 or 4.1, which is accessible via​

1x USB-B on the enclosure and the​

1x SD card slot of which should also be available from the outside.​

Yes ! Excellent choice

1x Teensy Audio Board serving as ADC and DAC

The teensy audio is based on the SGTL5000 which is a consumer grade, low-power codec. It's 16bits 44.1kHz. Clearly not enough imho. Noise figures are half decent but there are better options out there.
I'm thinking about Cirrus Logic and AKM. You'll need to get your hands dirty and make a board, but we could collaborate on this. I need a good analog frontend board myself, with line in, line out, mic preamps etc...
Working with Kicad v6 tho.

1x stepper motor driver with output via a CAT connection

You mean 8P8C?

1x screen​

1x rotary encoder for software control​

Alright, this is getting complicated

3. power supply unit with Powercon connection

1x transformer with linear generation of +-15V, +48V, +12V, +5V​

After I have already planned a few solutions in Eagle and Fusion, I would like to present two variants for discussion.


1. a very (perhaps too) compact unit with three boards on top of each other in a "Fischer Elektronik KOH 6100 housing kit". Here, all the components are very close to each other. I am particularly concerned about the proximity of the motor driver to the analogue electronics.

They are not as noisy as that. If you use Trinamic stealth drivers I don't think you'll get that much noise (acoustic and EM). I would be as worried with the step command line between the MCU and the driver. Also, measurement should be made with the stepper driver OFF:
Stepper ON -> Make your move -> Stepper OFF -> Measure -> Repeat

In addition, there would be hardly any space to accommodate all the connections. I will only use Neutrik connectors. I would have to install an XLR-TRS combo jack in this cabinet, which would have to be switched via a relay either to the microphone input with phantom power or to the line receiver. Close to it would be the Powercon connection and the CAT connector for controlling a stepper motor.
On the front of the unit, there would be little room for the SD card slot, rotary encoder, display and USB port.
This means that everything would be very cramped in the housing. The transformer in particular is difficult to accommodate.

Alternatively, the entire electronics could be accommodated in a 19-inch 2U enclosure in a very relaxed way. Here, dedicated XLR connections for separate line and microphone inputs could be realised. The same goes for the line output and amplifier output. The assemblies would all be further away from each other and there would be enough space.
This solution would be more comfortable to plan and build, but incomparably more expensive, due to the large enclosure that is now necessary.

I can't decide. What do you think?

I would make everything work as naked boards first, then re-design for a box (with I/O placement, shielding etc...).

Here are a few more details.
For the microphone amplifier I plan to use the THAT 1512 chip with a fixed 0 dB gain. However, for all those who absolutely want to set a gain, the possibility to connect a corresponding potentiometer or a rotary switch is to be prepared on the board.
The THAT 1240 chip is planned as the line input receiver. Here, too, 0 dB gain.
Of course, the THAT 1606 chip is now also used as a line output driver. Once again without gain.

Good choices. But I wouldn't fix the mic input to zero dB gain. I would love a steppable gain so low sensitivity dynamic mics could get in the ADC optimal range. Also, with those good ICs, a better codec is clearly justified.

The loudspeaker output is to be realised via an XLR connection. The amplification itself I want to implement with a small 5 Watt class-D amplifier. I am thinking of a very compact module from the Chinese market. It is based on the PAM8403 chip.
The motor driver is a Trident Silent Driver whose functions should all be controlled by the Teensy. However, it will hardly be necessary to use these additional functions. Even microstepping will hardly be used.

You mean trinamic?

I am not yet sure about the screen, but I am thinking of a colour display that is as large as possible.

If you want something basic, take a look at the screens used for 3d printers: screen - Shop Cheap screen from China screen Suppliers at FYSETC Official Store on Aliexpress.com
There are some graphic LCD with a rotary encoder, or touch screens.

The unit should actually also be usable as a standalone, but I like the idea of not attaching a power switch to the unit, but instead switching the power via a relay that would be supplied via the USB bus voltage... However, the idea is more of a snapshot and stems from the space-saving considerations for variant 1.

The device should work with REW and possibly take over all its functions independently in a later software version.

That's a lot of work...

Measurement data should then always be stored on its own SD card and also sent to a PC connected via USB.

A way simpler thing to do would be to design the device such that is presents itself as an USB AC2.0 compliant audio frontend and to add an additional USB HID endpoint so that you can send keystrokes from your board. Not sure REW have keyboard shortcuts to launch a measurement, but it would work.

These are my thoughts.

Greetings from Germany
Julius

Nice project, Cheers from Strasbourg!

 

[ccaudle]

wonder how large a chamber has to be

The effectiveness of acoustic absorption varies with thickness and frequency, so the size of the chamber depends on the lower frequency limit, and the absorption limit you will accept. The traditional fiberglas panels like Owens Corning 703 only have measured absorption coefficients down to 125Hz. There is probably some industrial standard that specifies that as the lower limit for practical reasons, but if you look at how required thickness changes with wavelength, you can see that the thickness quickly becomes impractical if you want high absorption down below 40Hz.

 

[juliusbusch]

The teensy audio is based on the SGTL5000 which is a consumer grade, low-power codec. It's 16bits 44.1kHz. Clearly not enough imho. Noise figures are half decent but there are better options out there.
I'm thinking about Cirrus Logic and AKM. You'll need to get your hands dirty and make a board, but we could collaborate on this. I need a good analog frontend board myself, with line in, line out, mic preamps etc...
Working with Kicad v6 tho.

I'm afraid you are right. The SGTL5000 is not great, but will do for prototyping. The big advantage of this is the Teensy's Audio Library. I Already have a working prototype which I built during my bachelor thesis. I use the Teensy and its audio shield to realize a class compliant USB audio device which was recognized by REW. As you proposed, this device handled the motor driver in parallel. And yes, I always did send the stepper motor to sleep during measurements. There is no way around this, even with my silent (yes) Trinamic drivers.
Software is not my strength, so I am quite scared to build my own ADACs and prefer using the Teensy audio library.
If you are developing a compatible ADAC board with the required software tools, I'd be so glad to incorporate it into this project.

You mean 8P8C?

RJ45 or 8P8C yes of course.

Good choices. But I wouldn't fix the mic input to zero dB gain. I would love a steppable gain so low sensitivity dynamic mics could get in the ADC optimal range. Also, with those good ICs, a better codec is clearly justified.

That's why I am designing an option for this into the board.
Btw. here is a quick picture of the audio board as it is planed by today.

If you want something basic, take a look at the screens used for 3d printers: screen - Shop Cheap screen from China screen Suppliers at FYSETC Official Store on Aliexpress.com
There are some graphic LCD with a rotary encoder, or touch screens.

That's a great idea. I will look into that.



For now, here is a picture of the IO on the back (but actually in 2U not 1U):

 

[PureBasic]

I'm afraid you are right. The SGTL5000 is not great, but will do for prototyping. The big advantage of this is the Teensy's Audio Library. I Already have a working prototype which I built during my bachelor thesis. I use the Teensy and its audio shield to realize a class compliant USB audio device which was recognized by REW. As you proposed, this device handled the motor driver in parallel. And yes, I always did send the stepper motor to sleep during measurements. There is no way around this, even with my silent (yes) Trinamic drivers.
Software is not my strength, so I am quite scared to build my own ADACs and prefer using the Teensy audio library.
If you are developing a compatible ADAC board with the required software tools, I'd be so glad to incorporate it into this project.

I have a teensy 4.1 at home but I'm not using it. I'm looking at it for the IEEE1588v2 compliance, but it's fairly undocumented and quite niche.
I went for the ESP32 (programmed in c with the ESP-IDF). Audio is handled by a DSP so the ESP32 is only here for management, webserver etc...
For your project, it should be quite easy to get audio In and Out of a custom audio frontend. The hard part is hacking a driver but it shouldn't be that hard to do if you know a bit of c.

 

[bigben83]

@juliusbusch This looks a great project, have you progressed any further with this project.