LME49721: PDM digital mic to analog conversion

Hello,

Thanks for reaching out. We are looking into your inquiry. 

Regards,
Sydney Northcutt 

Hi Jason,

Unfortunately the low pass filter is the only way to demodulate the signal without going through a codec (PDM -> I2S -> Analog).

I am not sure where you got the 6dB number from for your PSRR. Could you elaborate there? The performance should be a reflection of the spec of the microphone.

Best regards,
Jeff McPherson

Here's how I came up with the PSRR of 6dB.

The PDM signal is going to alternate between 0 (ground) and 1 (supply voltage).
With no output (silence) on the mic, the PDM will be an equal number of 0s and 1s.
This means the output after a low pass filter will be 1/2 of the supply voltage, with an associated 1/2 of the noise on the supply voltage.
Of course, any supply voltage above the cutoff of the LP filter will be attenuated, but any in-band noise will not.

This PSRR is in addition to the PSRR of the PDM source.

Hi Jason,

I see your point. Unfortunately there's no way to remove that noise after the fact, but if you choose a mic with a high PSRR you should have sufficient room to LP the DC value out and any low frequency noise (like 60 Hz line noise). The in band noise that comes onto the PDM signal will be a result of the PSRR of the mic itself.

Best regards,
Jeff McPherson

I don't follow.  What do you mean by "LP the DC value out and any low frequency noise (like 60 Hz line noise)"?
A low pass filter would retain 60 Hz and DC.  Were you talking HP? 

I am aiming for hi-fi frequency range, so I'm looking to retain 20 Hz audio.  50/60 Hz is still in-band and would not touched by a high pass filter.  But the DC offset would be gone.

What about if I run the digital signal through an inverter.
This would give me a differential signal.  Then I could low pass that differential signal.
Assuming no other mismatches, would this cancel the power supply noise?

Of course, "assuming no other mismatches" is a tall order, as the inverter will likely have different drive capabilities as the original PDM source.
That's another concern altogether. 

Hi Jason,

Sorry you're right, I was thinking of a HP filter but typed LP filter.

As you stated though it would be in band since you're going for hi-fi. A differential approach is not a bad idea, and you would definitely cancel out a lot of noise by doing so.

The assumption here would be that no other noise would be coming from the inverter. With some good differential layout you could at least filter out noise from the mic power supply, you would just need to be careful of the inverter you choose that it won't introduce much extra noise.

Best regards,
Jeff McPherson

What about using something like a SN74AUC2G86 Dual 2-Input XOR gate with 1A tied high and 2A tied low and the signal fed to both 1B and 2B?
This should give a differential signal with the same propagation delay, and drive level on both signals.
The VOH and VOL seems pretty close to symmetrical.

These signals could then be independently low pass filtered with an active 2nd order LP filter using the LME49721.
My final input is differential, so I would be relying on that input's CMRR to reject the now common mode power supply noise.

Or, would it be better to use a fully differential op-amp for this?
Would that provide some CMRR that would add to the CMRR of the final input I am driving?

Let me clarify:

If I use a differential amplifier with a CMRR of 70dB, will the residual common mode noise still be common mode noise afterwards, or will it be mixed in with the differential signal?

Or put another way, the input I will be driving has a CMRR of 50dB.  If the differential amplifier has a CMRR of 70dB, will I have a total CMRR of 120dB?