• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 2 answers
  • Subscribers 31 subscribers
  • Views 864 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

PCM1863 unused analogue inputs

Peter Stewart
Prodigy 20 points

Other Parts Discussed in Thread: PCM1863

Clarification please : what is the effect of leaving "unused" analogue inputs floating vs terminating to GND through a cap.  To my knowledge, both could be valid depending upon chip-internal circuitry.

Datasheet section 9.3.1 says "Do not connect unused analog input pins."  (Section 10.1.5 talks about anti-aliasing filters on used inputs.)

A couple of support question/answer threads recommend terminating unused inputs : e.g.

https://e2e.ti.com/support/audio-group/audio/f/audio-forum/849308/pcm1863-pcm1863-mic-bias-and-vin/3142142?tisearch=e2e-sitesearch&keymatch=PCM1863%20VIN1M%20floating#3142142

"Note that ALL unused inputs should be terminated with a 0.47uf Cap to GND. "

https://e2e.ti.com/support/audio-group/audio/f/audio-forum/403581/pcm1863---unuse-pins/1432241?tisearch=e2e-sitesearch&keymatch=PCM1863%252525252520VIN1M%252525252520floating#1432241

it is recommended that you AC couple with a cap to ground, but you can also use a high value resistor to ground. You can read more here: e2e.ti.com/.../2810.pcm186x-unused-inputs

(Incidentally, the "read more here" thread mentioned cannot be found.)

Background and further questions please :

I have an audio PCB which has a single TRS socket for stereo unbalanced inputs and provides a header to add XLR or 2xTRS sockets for dual mono balanced inputs : in this instance, these extra connectors are not present so the header pins are floating/unconnected.

I believe the PCM1863 pins VINL1/VIN1P and VINR1/VIN2P are connected to TRS Tip and Ring (for stereo Left and Right), and PCM1863 pins VINL2/VIN1M and VINR2/VIN2M are connected only to header pins (hence floating/unconnected in this instance).

1.  Obviously the ADC inputs should configured as Single-Ended for this use case, but I can hear a reduction in noise level by configuring the ADC inputs for DIFF mode instead.  Is there any internal chip-internal coupling between VIN1P and VIN1M which would explain the perceived noise reduction ?

2.  I have noticed that the ADC Right channel input is shorted to Ground due to using a 2-conductor plug with the 3-conductor TRS socket.  (Ring is shorted to Sleeve).  Since the ADC channel is active, the input is not "unused" : is there any long-term problem with having the ADC input shorted to Ground ?

Thanks,

Peter Stewart

  • 0
    TI__Mastermind 23050 points

    Hey Peter,

    It's not necessary to terminate the unused analog inputs, but I do not believe it would hurt to have AC coupling capacitors to terminate unused channels. All of the pins are muxed to the respective ADC inputs.

    The dynamic range of the device is improved in differential mode since each pin can swing 2.1Vrms and so you have twice the input range. You will typically see 6dB better noise floor for differential vs. single-ended. If you're applying a single-ended input in differential mode though then your ADC output will also be reduced by 6dB.

    All of the inputs are internally biased to AVDD/2 so it is not recommended to ground any of the pins even if they are unused, I wouldn't expect any long-term degradation from this, but it may impact the bias voltage and performance of other channels.

    Best,

    Zak

  • 0
    Prodigy 20 points

    Hi Zak, thanks for the response.  I had not fully taken into account the statement in section 9.3.1 about DC blocking caps being required due to AVDD/2 biasing.  Your comment made me reasses my understanding, and I think that explains why I can hear a reduction in 60 Hz hum on the output when I switch the input to DIff mode.

    Just for completeness : is the ADC input biased to AVDD/2 in all modes i.e. Single-Ended mode as well as Diff, whatever input mux'ing selection is applied ?  (section 9.3.4, Table 2)

    Re the input which is shorted via plug&socket path to ground : would it be better to set this ADC input to "No Select " on-chip ?

    Thanks again,

    Peter Stewart

  • 0 in reply to Peter Stewart
    TI__Mastermind 23050 points

    Hi Peter,

    If you have 60Hz noise in your system differential inputs would probably help if it is picked up as common-mode noise, but this could also point to some layout concerns that you may need to resolve. 

    The input pins are always biased to AVDD/2, regardless of single-ended vs. differential or whether the pin is routed to an active ADC channel. I confirmed that grounding one of the pins does not drop the bias at other pins at least.

    If you are only using a single ADC though then it is probably best to leave the other path set to no select.

    Best,

    Zak

  • 0
    Prodigy 20 points

    Hi Zak, just one more wrinkle if you would.

    If the AVDD/2 biasing still applies when the ADC input is set to "No Select", what impedance would be seen looking into that pin ?  (i.e. if something does actually try to drive that input while it is set to "No Select")

    Obviously one would not expect any output from that channel, but understanding what the rogue driver would see as a load impedance would be useful.

    Cheers,

    Peter

  • +1 in reply to Peter Stewart
    TI__Mastermind 23050 points

    Hey Peter,

    Each pin will still see a 20kOhm input impedance so the loading should be pretty light.

    Best,

    Zak

  • +1
    Prodigy 20 points

    Thanks Zak, you have answered all the questions I have on this topic  (at least until someone asks me another which I can't answer myself (:^))