Audio capture card options

Posted by: @askolds11

I've found that the Behringer LINE 2 USB is amazing (like, seriously perfect for this project and it will be used in version 2) but it may be above your budget. Might be worth dissecting one and seeing what chipset it uses.

This seems to be unavailable in my country, although there is a GUITAR 2 USB, that looks the same just without the 2 leads coming out of it, but it's 20 euros, and I'd much rather get something for much cheaper (5-10 Euros).

 

I have tested that one too and it seems to be basically the same chip etc so performs the same. Not sure about any cheaper options to be honest... I'm sure some of the cheap ones will work well but I've just not gone through and tested them extensively.

I'm gonna give the CM108 card another try - it comes with the CM108B chip, which, according to the datasheet, can have an external EEPROM attached (the CM108 can too), and you can (hopefully) disable the gain (can't do this on the normal CM108, which might be discontinued anyways). The EEPROM chip is about to arrive, but I'm going to have to learn how to use the memory addresses and how to program it, but I'm sure I'll make it haha.

It also could be that the chip could be quiet enough even with the gain enabled (of course not as good as the audio capture cards or that cable), but my circuit was the problem - I'm using my own circuit with a LM4562 op-amp and a dynamic capsule, and I imagine the noise could be from the somewhat lenghty (physical size) feedback loop for the op-amp, but I've bought a THAT1512 (hopefully not fake), which should solve that problem, although I still have to get the isolated +-15v chip, although I'll probably start with a +-12, since that's available in my country.

Keep us posted on that. I've never heard of the CM108B before now and it certainly sounds like an interesting option!

Finally got done with exams and tests, could try some things.

Unfortunately, I haven't tested the EEPROM option, as I'm still working on it - found an Arduino sketch with half of the necessary EEPROM data, successfully programmed the EEPROM chip, but didn't get it to connect to the CM108B (soldering to the tiny pins itself was terribly hard!).

I did however try it out without the microphone circuit component (in my case - LM4562 op-amp, very likely a lot of noise from the circuit) - I recorded some audio on my computer - I've got some filters on it, so it's pretty noiseless -, and played it back on my phone connected to the CM108B sound card microphone jack.

I tested 2 configurations - unmodified and removed bias voltage (don't need it for the microphone either) by removing R13 from the circuit on page 28 in this datasheet (I am not sure how accurate the rest of it is, but I've confirmed that at least the components from the VBIAS pin are exactly the same). I left R10 and C15 there, as I'm not sure how good it is for the pin to be left floating. I also left C11 (confirmed this is the same as well), as I do not want to accidentally fry the chip.

Here are the results (I've tried to normalize the audio levels, so they're the same):

Original audio:

Sound card unmodified:

Sound card removed bias:

There is some minor noise I can hear from the original audio, in the unmodified card I can definitely hear some added noise though - when talking, I can hear the typical buzz, but by removing the resistor it sounds almost the same as original! There is maybe some added noise, but I can't say for sure, as the original audio has some noise.

I was honestly surprised by these results - it must be my circuit that's causing the noise, as these results are not even in an enclosure - I found that my whole desk is a disaster for interference and the front of my computer, where I got these results from, is not any better. I'm going to try to put a circuit together with the THAT1512 (already tested with a dynamic mic capsule and it works, but I have to rethink it a little) and try to find a metal box to minimize interference and feed it into the CM108B with no bias resistor and maybe EEPROM modification, if I manage to finish it, to test the results out!

Here is the stock card with the resistor you have to remove. I did it with my soldering iron and tweezers, but I've seen people do it on YouTube by just snipping it off with flush cutters.

Note: I seem to have forgotten to turn off AGC, so the differences are more noticeable than they would likely be normally, but the comparison should still be fine, as they're under the same conditions.

Hi,

I'm new here and I may start doing this mic project soon. I'm still in the process of checking all parts prices and availabilities before doing anything concrete (and also I want to mention that I have zero knowledge in sound electronic so i'll have to triple check everything on every steps 😅).

I just realized that I have an old sound card that I don't use and so I may recycle it in this project.

This is a TechRise external sound card. It costs ~$15 on aliexpress and amazon US, but it is surprisingly cheaper on amazon FR where it costs only 8€.

It looks like this:

Sadly the rotary knob is for headphones volume not for mic volume 😢 

When opened it looks like this:

Do you think it may work?

If I want to test it without altering the sound card, can I just put a male jack on the DIY mic side?

Sorry, I haven't had much time, hope this is still relevant.

Yeah you can test it by simply putting a mic jack on the DIY mic side, just make sure you have put a DC blocking cap (that is, a simple capacitor which in the context of the schematic blocks the DC voltage) on the input as the microphone jack usually (almost always) has bias voltage which could damage the DIY mic.

I'd recommend testing it with a cheap electret microphone (I have a bunch of laravel mics - electret capsule at the end of a wire - from aliexpress for like 0.5€ a piece) on a laptop without it plugged in and away from other electronics to isolate as many variables as you can.

In my experience these sound cards usually suck, it's hard to find one that doesn't have a lot of noise. The CM108 cards seem to be a rare exception. I've found some on AliExpress that adveritise CM108, but the chip is epoxied so you can't find out what it is, however they sound good for electret mics - I wouldn't recommend them for a DIY mic thought as it's harder to understand the circuit without the chip pinout.

In my case most of the noise seemed to come from the bias voltage, which on my board can simply be removed by desoldering that one resistor.

Agreed, those CMedia chips are a real exception, it's a real shame we can't get them.

@marcdraco Do you mean the CM108 or some different chip from them? The CM108 (although only the B variant, if I remember correctly, they were very similar though, I know that the eeprom differs) as a sound card is readily available on AliExpress.

I'm not 100% sure to be honest. I've only seen it mounted direct to the PCB and then covered in epoxy which rather cuts out the middle-man. Compared to all the other "low-cost" ones I've tried they seem to stand head and shoulders above.

I snagged a Focusrite Scarlet yesterday to make sure my designs work correctly with "high end" gear.

And it almost did. I overlooked how little IC regulators like the TL431 starts motorboating when the current drops below a certain level (this is predicted) but I didn't allow myself enough headroom and I ended up with a 10K oscillator. 

Even with this fairly mature design, there is a little bit of background hiss if you boost it hard enough. On balance the CMedia cards I've used (over the years) have been superb. I should really plug it in and see if the drivers help ID them.

Posted by: @askolds11

↑

Sorry, I haven't had much time, hope this is still relevant.

Hi,

Actually I already did some progress and test that card, you can see my first test here: https://forum.diyperks.com/postid/4530/ 😊 

For now there is some white noise on my tests, but I still have to complete the wiring and shield every cables. I'll keep you in touch when everything will be done correctly and when i'll do final tests.

@askolds11 

Hoping that you will still see my inquiry. How did you connect the audio out of the mic preamp to the CM108B mic input? where will the audio out + and audio out - to the TRS of the CM108B mic input? Thank you!

The CM108B has its own mic input and bias voltage for a standard electret. So it expects quite modest voltage perhaps 200mV at the most. This is far less than the external mic amplifiers can produce so it would be a good idea to put a pair of anti-parallel signal diodes across the input before the input capacitor.

The input impedance is 10K so you don't need a massive decoupling capacitor for the high-pass (DC blocking).

@marcdraco 

Should I use the VBias and GND from CB108B PCB to route the audio out of Matt’s preamp, or the Mic In and GND of the CM108B?

Here's the relevant part of the schematic (blanked out the other parts). On my units the tip and ring are shorted, so to remove noise I had to remove the Vbias, which is done by removing R13 (also seen higher in the thread on the board itself). The connections from the preamp should be done as marked in the schematic and as @marcdraco said, a pair of anti-parallel diodes would be a good idea. The DC blocking cap is already there, probably not worth the effort changing it.

That looks quite reasonable, yes.

Although you don't even need a separate amp (unless you already have one) as the CM108 has an internal amp up to (from memory) a +20dB gain controlled through the USB HID.

There's no harm in using a separate amp because on-chip PGAs are often quite noisy compared to a decent op amp so you'll probably want to set the gain as low as practical and use the gain from your amp to provide the necessary lift with less noise.

Having a couple of series caps (blocking DC) will lift the low-end of the frequency response a little though.

As it is (without anything else) you've got a high-pass filter with 1 uF into a 10k input impedance which gives you an f -3dB of 15Hz which is sufficient to capture pretty much anything a normal mic can throw at it - many will fall off naturally below 100 Hz or so.

You can calculate the actual value of C using:

C = 1/(1/C1 + 1/C2)

Capacitors in series work like resistors in parallel hence why there are so many reciprocals.

@askolds11 Thank you for the confirmation!

@marcdraco I’ll be using the CM108B USB soundcard as the digitizer for my USB-C mic project. I’ve decided to move on from the wireless microphone project as it ended up costing me too much time, effort, and money. HAHA

Awesome. It's a great chip. The only "gotcha" you might face is the USB input which (strictly) should be impedance controlled. Most of the impedance comes from a pair of 22R resistors so keep the lines from the USB D+ and D- as short as practical and as close as you can to the same length. USB audio seems to be a leisurely 12 Mbps which is USB 1.1 speed although I haven't verified that as yet.

I could add wireless to the V2 but it seems like a high cost for no real benefit, save for not requiring a differential pair to kill off common mode noise.