Combating TDMA noise on a wireless microphone.

[Creep]

Hello everybody.

I am working on a DECT (1.9GHz) based wireless microphone product.
However I am struggling with noise on the microphone line of around 100Hz (the packet transfer rate), meaning I am dealing with TDMA noise. Moving the microphone around also changes the characteristics of the noise (mostly the amplitude).
I am using a TI TLV320AIC3204 audio codec and a an electret microphone. The microphone is attached to the PCB using a 2 wires (around 20mm in length), on the PCB side the two lines wires are connected using a connector where the continue as differential lines with LC filtering in series as shown in the attachment. The traces are short (10-15mm in total from the connector to the codec input), with a reference ground plane directly under in the PCB layout. The signal is clean with no microphone attached, which points to this not being a power supply issue.
Googling around, one can find plenty of articles like this one:
https://product.tdk.com/en/techlibrary/solutionguide/suppression-filter_varisotr_microphone.html
where manufacturers suggest in series noise filters (ferrite beads, inductors etc.) at the codec input. However the thing that eludes me is: how do they work? I have tried using a few of the filters with no success, but I also can't understand the basic principle of operation.
From my understanding, the filters are circuit elements that change their impedance based on frequency (maximum impedance at a certain frequency, usually in the range of 500-2000Ohm). Looking at the insertion loss numbers, they seem to coincide with placing a 50Ohm load at the other side of the filter (forming a voltage divider), however the codec input is a high impedance input (10kOhm in this case), meaning that from a pure voltage divider point of view, the filters basically do nothing, and yet, plenty of articles show them being used exactly like this. Am I misunderstanding the magic of how these filters work? I can't exactly just terminate the codec input with a 50Ohm load either, since the microphone output is also fairly high impedance (in the order of around 2kOhm), meaning I would be crippling my signal levels. Is there some mechanism that I am not aware of that can allow such devices to do their job even with a high impedance load as the articles suggest?
Any other tips on dealing with TDMA noise would also be much appreciated.

[tggzzz]

Start by realising it isn't "noise", it is "interference" - an EMI/EMC issue.

When the RF power is coupled into a non-linear element, it is rectified and/or mixed and produces a DC level shift. That DC level shift will turn on/off at 100Hz and will be amplified by an audio chain.

Non-linear elements can be diodes, transistors, rusty bolts, some types of capacitor, poor connections, magnetic components, etc etc.

[Creep]

That much is understood, I apologize for the misuse of terminology.

The non-linear elements at play here (aside from ESD protection) are the codec analog input and the microphone internal FET amplifier (not quite sure how much of an influence the MICBIAS bias voltage generator has here though). Replacing the microphone with an equivalent impedance resistor (at the end of the cable same as the microphone would be) makes the noise go away. Assuming that the noise induced in the wires and traces stays the same, that should rule out the codec input being the root cause here. Same goes for the microphone bias voltage generator. That leaves just the microphone capsule itself. So the question still remains: what can be done to reduce the influence of EMI on the audio signal?

[Terry Bites]

The EMIF10-1K010F2 is chock full of diodes that will no doubt work as great RF detectors. You dont want any diodes near the mic amplifier inputs.
See Figure 22. Analog Input Connection With Pull-down Resistor in the data sheet. Apply this to unused inputs.

[Creep]

Assuming you mean the EMIF02-MIC03M6:
Alright, I will give it a go and try removing it, however: what are my options in terms of ESD immunity in that case?
Also: it is a purpose built IC meant to combat EMI and ESD problems on microphone circuits. If any diodes ruin you day when it comes to microphone circuits, then what is the point of such devices?

Regarding grounding unused inputs through a capacitor: thank you, will give it a try.

[Marco]

Maybe filter the mic bias?

https://e2e.ti.com/support/audio-group/audio/f/audio-forum/706274/tlv320aic3204-tlv320aic3204-has-high-frequency-noise-on-micbias-will-can-cause-a-audible-noise-in-system

[Creep]

In the topic linked, a TI employee specifically states that there should be no filtering capacitors on the MICBIAS output (the last post).
Also, as stated earlier, replacing the microphone capsule with a resistor causes the interference to disappear  completely. Would that be the case if MICBIAS was what is generating the noise? Also, in this case, the noise is around 100 Hz frequency, that coincides with the RF transmission intervals.

[Marco]

The electret is more like an open circuit than a resistor, so could be. What happens if you leave it open?

The engineer is probably afraid capacitance on the micbias output would cause the internal reference amplifier to become unstable. A RC filter would not necessarily have the same effect as a directly connected capacitor though. You could also take mic bias from the supply rail and RC filter that (So R from the supply rail to the C, then R41 from the capacitor instead of the micbias output on the IC).

[Creep]

The circuit is completely silent without the mic connected.

[dmills]

In which case look at the mic as being the source of the demodulation.

Those capsules generally have a built in buffer amp, which is what the bias supply powers, I would be tempted to try 47pF or so directly between each of the mic connections and the metal case of the microphone (zero lead length).

[Marco]

I would still try to add a 100 Ohm resistor and a ferrite bead in series, with a 100 uF capacitor to filter the IC micbias (for a bodge with through hole components putting a smaller capacitor in parallel with the electrolytic won't accomplish much, it's the leads which add the inductance, but with SMD components it could).

[Creep]

Adding 47pF between each of the signal pads on the microphone and the microphone housing does help deal with the issue. Not the easiest solution to work with (lots of extra manual work for each product), but at least there is a solution. I will test out Marco's proposal when I have the chance since having the fix on the PCB side is so much more convenient for production.

[Terry Bites]

Yes wrong part#
The perils of cut and paste!