• Resolved

PCM4204: ADC Full Scale Input Voltage Discrepancy

Prodigy 40 points

Replies: 4

Views: 52

Part Number: PCM4204


Differential 0dBFS analog input level is specified as 6Vp-p (typical) on the PCM4204 datasheet (9/2004, p.3). Here it approaches 10Vp-p differentially, a greater than +4dB difference, indicating the '4204s input pins need to swing nearly rail-to-rail before 0dBFS is achieved.

Input voltages were confirmed via oscilloscope, audio analyzer, and DVM. Analog rail voltages VCC1 and VCC2 measure 4.99VDC. Digital levels were confirmed via the CLIP3 pin and two different digital analyzers. The attachments show the single ended and differential Vins and an FFT. Vin is 8.1Vp-p with a resulting digital level of approximately -2dBFS. These tests were carried out on our own board, not the TI evaluation board.

Please explain this discrepancy. 


  • Hi Tony,

    Would you measure the VREF outputs of the PCM4204 device in question?  Does this occur on all boards?  

    I don't think it particularly applies but wanted to mention that the input amplifier on the EVM is configured for a differential gain of 270/1kΩ = 0.27V/V (-11.3dB).  I believe the amplifier was configured this way to utilize a larger input signal from the Audio Precision and lower the noise contribution from the OPA1632 input drive amplifier.  I don't think you copied this schematic and probably would have figured this out but wanted to verify the measurements are being taken at the pins of the device.

    Collin Wells
    Precision ADC Applications

  • In reply to Collin Wells:

    Hi Collin,

    Thanks for the fast reply! Your question helped to identify the underlying problem.

    The Vref+ outputs were being treated as inputs and connected to the +5V analog supply. After cutting traces to the Vref12+ and Vref34+ pins, they both output 2.97V. I could not find Vref voltage specs on the datasheet, but a 3V Vref helps to explain why the 0dBFS analog input voltage is specified at 6Vp-p.

    Quick background: this board was redesigned for the PCM4204 after the original A/D became unavailable. The original A/D also had Vref+/- pins except they were inputs instead of outputs. Although page 15 of the PCM4204 datasheet is clear about their application, the original connections were inadvertently propagated to the new PCB layout. It looked correct to all during design reviews and initial testing, especially considering a +2.5V Vcom voltage.

    Fortunately, this is only a first-pass PCB. The Vref+ pins are on the corners of the package so cuts and jumps will be easy prior to re-spinning the PCB. Analog gains will be adjusted as well assuming a 6Vp-p 0dBFS target and then performance can be reevaluated.

    Input current limit on any pin a specified as ±10mA; current into the pin when pulled up to +5V is only a few microamps. Nevertheless, do you think there is any chance the PCM4204 was damaged and needs to be replaced prior to proceeding with mods and testing?

    Also, the datasheet does not list production limits for the full scale input voltage, just 6Vp-p typical. What full scale input voltage variations can be expected lot-to-lot?

    - Tony

  • In reply to Tony Scott:

    Hi Tony,

    Thanks for the updates.  I will mark your post above as the answer to this thread. If the device is still functioning properly then I'm guessing it was not damaged and you were simply over-driving the modulator reference input which is likely not a strong-drive output as you found so damage is a little unlikely.

    If your next measurements end up being substantially off in the higher performance SNR / THD+N measurements then perhaps something did get improperly biased in a damaging way but as mentioned I wouldn't guess so.

    Collin Wells
    Precision ADC Applications

  • In reply to Collin Wells:

    Thanks for your help and expertise Collin! I have two PCM4204 related questions and will post them to a new thread.

    - Tony Scott