• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 1513 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TLV320AIC3106 with dc coupled inputs

Lajos Balázs
Intellectual 650 points
Other Parts Discussed in Thread: TLV320AIC3106

Hi All,

I was redirected from the C67x Single Core DSP Forum with this question:

I am using the LCDKC6748 dev. kit with a TLV320AIC3106 codec on it. Unfortunately the codec uses capacitive coupling on LINEx/MIC inputs but I need dc coupling. This seems not possible with this codec because the LINEx inputs of the codec have DC bias voltage (around +1.3V). I removed and short cut the input caps but the CODEC measured -32768 only. Questions:

  1. Can the TLV320AIC3106 codec work with DC coupled single ended inputs? (non-capacitive coupling)
  2. If not then is there any TI CODEC (24-bit, 96kHz) which can be used for dc coupled inputs? I just need the simultaneously started 2xADC and 2xDAC. No DSP, no filtering. A codec which can be used for instrumentation application.
  3. It is shocking but I could not find the reference voltage of the ADC and DAC (TLV320AIC3106) in any manual. I also could not find the resolution of the ADC and the DAC in any manual. I think the resolution is 24-bit. Can somebody tell me in which manual specifies the resolution and the VREF?

Many thanks for your help!
Best regards,
Louis

  • 0
    TI__Mastermind 28585 points

    Yes, the AIC3106 can be used with a DC input. You would need to level shift your DC input and perhaps change the gain to match the input range of the ADC. As you mentioned, the 0 level is at 1.35V and the signal goes up and down 1 V from there.

    The internal bandgap reference is not brought out to any pin, so the voltage is not specified. The output common mode is programmable in Register 40.

    This is a 24 bit device however, we spec only THD and SNR so you can get an idea of the performance from those numbers.

  • 0 in reply to Converter Apps
    Intellectual 650 points

    Dear TI,

    Thanks for the answers. The DC voltage is clear, I have to shift the 0V to +1.35VDC and I can measure +/-1V peaks ((+2.35V and -0.35V). Is this true?

    If I understand your answer right the CODEC has 24-bit ADC, but I’d better if I use it in 16-bit word size... That's okay.

    But I MUST KNOW the reference voltage because otherwise I can’t calculate the measured voltage... This is why I am using ADC: to measure amplitude of the input signal. But how do you calculate the amplitude without knowing the reference voltage?

    Also there is new doubt with AIC3106: Are the ADCs started in synch (at the same time)?

    Do you have a CODEC dedicated for instrumentation (not audio) applications? I mean:

    • 2xADC and 2xDAC
    • ADC started at the same time,
    • Optional external start input
    • 24-bit,
    • +/-1V peak input DC coupled
    • variable internal gain,
    • 96 kHz or higher sampling rate,
    • SPI port controlled, (No need McAsp or McBsp port)
    • Conversion complete interrupt,
    • Without 3D filtering or DSP
    • cheap.

    Many thanks for your help!
    Best regards,
    Louis

  • 0 in reply to Converter Apps
    Intellectual 650 points

    Hi,
    Hopefully you did not forget to answer my earlier questions. I can’t imagine why the value of the voltage reference is so "top secret weapon" for TI? :-)

    I have new details about the AIC3106 codec. I am not too impressed about the quality of a 24-bit codec. As I mentioned I am using the TI LCDK6748 development kit which has an AIC3106 CODEC on it. I removed all the input caps from the baord and replaced them with a wire. So I am using DC coupling.

    On the image below you can see the input signals on the pin 3(LINE1L+) and pin5 (LINE1R+) of the CODEC. Both inputs are DC coupled shifted by +1.33 and +1.29V, amplitude around 1.2V, frequency 100Hz, and the blue signal has 1 ms phase delay to the yellow.

    Now look at this! This is a screen shot how the same signals look like in Code Composer Studio 6’s graph window. In the McAsp port's read interrupt service routine I just read the data measured by the CODEC and store them into two 16-bit signed integer arrays. The I started the CCS6 unique graph function which draws a graph from the selected array. (Very nice tool)

    Basically they are the same... But why do we have slopes? It seems to me there are still serial caps in input path inside the CODEC??? Any advice what could be wrong?
    Best regards,
    Louis

  • 0 in reply to Lajos Balázs
    TI__Mastermind 28585 points

    Hi Louis,

    No big secrets, but, i don't know why you need it. As I stated in a previous post, the input is centered on 1.35V. This level (+/- some offset) will give you a zero output code. The max output code will occur at 2.35V (0x7FFF), The min code will occur at .35V (0x8000) and it will be linear between those points. If you know the number off bits you will be using, you should be able to divide that input range by your resolution and pretty much know what voltage you are looking at. I am not sure where the voltage reference fits into that equation.

    With the standard EVM and Goldwave, you can record a much better waveform that what you are seeing so I would assume that something is wrong in your setup.

    Best Regards,

    dave

  • 0 in reply to Converter Apps
    Intellectual 650 points

    Hi Dave,
    Thanks for your answer!

    Why do I need to know the reference voltage? How else do you calculate the measured voltage from the code ratio (Code / 65536) to display it?

    You need to multiply the code ratio with the value of the voltage reference. (In this case - because the codec can not measure ground symmetrical signals - we have an offset voltage)

    Vout = VRef * Code / CODE_MAX +VOffset

    With stating that the 0x8000 means +0.35V and 0x7FFF means 2.35V you specified the value of the voltage reference. It msut be +2V and we have +1.35V offset. Otherwise the Vref = +2.V is interesting. I calculated the Vref = around +1.7V.

    So we have this transfer equitation: Vout = 2.0V * Code / 65536 + 1.35V

    I missed this equ from the specifications + the temperature/power supply dependency + accuracy (min/max value) of the voltage reference and of the offset voltage. Like it is in any other TI’s ADC user’s manual.
    So
    •    if the code is +32767 (0x7FFF) then Vout = 0.999998 + 1.35V =~ +2.35V (-1bit voltage)
    •    if the code is -32768 (0x8000) then Vout = -1.0 + 1.35V = 0.35 V

    If it helps then I can send the demo project to you. I do not think that I can turn on a serial capacitor in the codec during the setup. I am using the setup of the McAsp example of the C6748 StarterWare. Filtering can do things like this but the input is DC coupled, all the gains/attenuations are 0dB and there is no filtering set.

    Q: The EVM means the EVM kit of the CODEC? What is Goldwave? (I know these are obvious for you but not for me)

    By the way my earlier question whether TI has a CODEC for instrumentation applications. See the spec above at the end of my earlier post.

    Many thanks for the info and for your help.
    Best regards,

    Louis

  • 0 in reply to Lajos Balázs
    TI__Mastermind 28585 points

    Hi Louis,

    What you are calling Vref is what I am calling the input range (0.7V rms as stated in the datasheet i.e. 2V). the input range * code/max code range +1.35 V is correct.

    Goldwave is a shareware audio package for windows. We use it with our EVM to process audio data.

    You would have to forward any specific questions on the DSP side to the appropriate DSP team. Being an analog guy,  I am not familiar with their tools.

    In this forum, we are focusing on the audio codecs. you would have to check with the industrial side to see if they have anything better for your app. The precision data converter team will have more info on their offerings. In audio we don't usually spec any DC parameters like offset, drift etc...

    Best Regards,

    dave

  • 0 in reply to Converter Apps
    Intellectual 650 points

    Hi Dave,

    Okay, many thanks for your help. I know this is an audio codec but I got it with the TI’s LCDK C6748 development kit. I thought the codec is just an ADC and a DAC with some DSP features so every parameters are specified like industrial ADCs/DACS.

    I also understand that the Vref is not an issue (transparent for the user) for an audio app, and the 1ms delay between the input and output signals also means nothing when play back an MPEG file. But these are very important in a test & measurement application.

    The idea of the CODEC is nice. TI should design CODECs for DSP nad T&M applications. This codec can not be used for test & measurement applications or it can but with many-many restrictions.

    Thanks again,
    Louis