Post Go back to editing

AD1938 55 Hz noise

Category: Hardware
Product Number: AD1938

Hello again all!

I'm having a bit of trouble with my AD1938 setup. I have an board design that has a ADAU1462 in conjunction with AD1938 that feeds audio into PAs and then into speakers.

My problem is that the audio has a huge amount of noise in it. Upon inspection with an oscilloscope, It looks like a brief DC that is in the signal at a rate of 55Hz. This appears right at the output pin of the AD1938. I have tested out the power rails and everything to see if there's a similar signal elsewhere, but there is no indication that this is an external noise interference. It looks like the AD1938 is generating it.

I'm not sure if something in my code is incorrect, either in the DAC or the DSP (the DSP program is thus far a pass-through for S/PDIF plus 2 sinusoid sources for testing. I have some screenshots below. I just don't see these values corresponding to any kind of frequency from either of the chips, and it doesn't seem to be from the power rails. The power supply is an SMPS, further regulated by an LDO down to 3.3V and a Linear regulator for 5V.

The waveform you see the output from the Analog Devices' recommended op-amp interpolation filters. I copied the design exactly as specified in the AD1938 datasheet - I do not think these op-amps are to blame for this noise. And yes, I did test the output pin of the AD1938s to verify the noise starts there.

I'm hoping some of you AD1938 veterans might be able to sniff out what's wrong here. I do know that my SMPS is a cheap acquisition for quick and dirty testing, and I wouldn't be surprised if it were the source of the noise, however as I mentioned, I checked the SMPS power rails (which directly feed power to the PAs on the board), as well as the two regulator outputs. None of these sources show noise anywhere near this low frequency (as you'd expect, they do have a few mV here and there in the upper kHz region). I tried to put big bulk capacitors on the SMPS output terminals to try and snub anything that might be coming out - there was no appreciable difference in noise level from the speakers. I'm inclined to dismiss the power supplies as culprits.

Thank you in advance!

Parents
  • Hi   ,
    The issue you're experiencing with the AD1938 could be related to several factors. Here are some 
    Clocking Issues:
    • Jitter: Ensure that the clock signals (MCLK, BCLK, etc.) are clean and stable. Jitter in the clock can cause artifacts that might manifest as noise in the audio output.
    • Synchronization: Check that the clocks for the AD1938 and ADAU1462 are properly synchronized. Any mismatch in clock rates can cause glitches or unexpected DC offsets in the audio output.
    2. Grounding Ground Loops:
    •  Improper grounding can lead to noise issues, all the ground pins on the Codec are internally connected It is best to use a ground plane and connect the parts to this plane               with traces as short as possible
    3. Op-Amp Circuit:
    • please refer the below circuit which is used in our Eval Board design
    Additionally, Dave explains a few more points regarding the noise in the link below. Please have a look.
    Regards,
    Harish
Reply
  • Hi   ,
    The issue you're experiencing with the AD1938 could be related to several factors. Here are some 
    Clocking Issues:
    • Jitter: Ensure that the clock signals (MCLK, BCLK, etc.) are clean and stable. Jitter in the clock can cause artifacts that might manifest as noise in the audio output.
    • Synchronization: Check that the clocks for the AD1938 and ADAU1462 are properly synchronized. Any mismatch in clock rates can cause glitches or unexpected DC offsets in the audio output.
    2. Grounding Ground Loops:
    •  Improper grounding can lead to noise issues, all the ground pins on the Codec are internally connected It is best to use a ground plane and connect the parts to this plane               with traces as short as possible
    3. Op-Amp Circuit:
    • please refer the below circuit which is used in our Eval Board design
    Additionally, Dave explains a few more points regarding the noise in the link below. Please have a look.
    Regards,
    Harish
Children
  • Thank you for the detailed response Harish, much appreciated!

    I believe all the pulses coming from the DSP into the AD1938 are clean and stable, from what I recall. I have not checked the MCLK pins though, I will do that sometime this week.

    But your mention of synchronization is making me think - on my board, AD1938 and ADAU1462 are both powered from two separate crystal oscillators. They're both the same frequency, same load capacitances, but they're independent of one another. Do you think this may be a potential source of problems? I did leave provisions on the PCB to connect the MCLK OUT of the DSP to the DAC, but left them unconnected, and turned off MCLK OUT on the DSP.

    The grounding could be a potential issue as well, I made a mistake on the PCB, and one of the analog grounds was left unconnected - I jumped with a short piece of copper wire to a nearby ground. The rest of the grounds are going to a dedicated ground plane. I did expect noise from the layout, but this issues seems to be closer to your first point about mismatching clocks, it feels like the DAC just drops out every now and again.

    As for the output filters, I think I followed Figure 31 from the AD1938 datasheet (I pasted it below) It looks a little bit different from the Eval board setup you show. Do you think this could be a potential culprit?

    Thank You,

    -Ernest

  • Hello Ernest,

    The master clocking is the problem. You are seeing where the drift of the frequencies are such where occasionally there are too many or too few master clock cycles per frame so the DAC will mute. If the frequency was way off you would get nothing but the two crystals are close enough where it is right for a short time but wrong at about a 55Ha rate.  

    You MUST send out a bit clock and frame (LR) clock that is divided down from the master clock that is being fed into the codec. 

    Regarding the filter. The difference is that the AD1937 is a differential output and the AD1938 is single ended so the filters are slightly different. Your filter circuit is fine. 

    Dave T

  • Hello Dave,

    Thank you for the fix! This did it, I deleted the oscillator circuit on AD1938, and activated the MCLK_OUT pin on the ADAU1462, sounds like butter now!

    Many thanks,

    -Ernest