ADPD107 TIA ADC Mode Timing

I am trying to use TIA ADC Mode for the ADPD107, and the data sheet does not explicitly describe how the timing should be changed to properly sample all TIAs during a timing block. The only mention of timing with respect to TIA ADC mode is the following"

"To measure the response from the TIA and verify that this stage is not saturating, place the device in TIA ADC mode and slightly modify the timing. Specifically, sweep SLOTx_AFE_OFFSET until two or three of the four channels reach a minimum value (note that TIA is in an inverting configuration). All four channels do not reach this minimum value because, typically, 3 μs LED pulse widths are used and the ADC samples the four channels sequentially at 1 μs intervals. This procedure aligns the ADC sampling time with the LED pulse to measure the total amount of light falling on the photodetector (for example, background light + LED pulse)."

It seems like the device should be able to sample from all 4 TIAs without issue by extending the length of the LED pulse. I have been successful at making what seems like valid ADC measurements from all 4 TIAs during a timing block, but I noticed that LED pulse width, LED pulse offset, AFE width and AFE offset all have the ability to change this setup. It also appears that AFE fine offset does not have any effect in this configuration. It would be nice if there was an explicit description of how this sampling takes place so that I can optimize it, because playing around with the values leads to some confusing and non-intuitive results.

  • 0
    •  Analog Employees 
    on Oct 17, 2016 10:36 PM

    TIA_ADC mode is not really designed to measure modulated LED signals. This is described in the datasheet. The best use of TIA_ADC mode is for ambient light measurements, measuring PCB parasitics, and other types of signals. The primary reason for this is because when the device is in TIA_ADC mode, it simply takes a single ADC sample per period. The ADC samples at 1MHz. So if you are trying to measure 4 channels of modulated LED signals that are each 3 or 4us long, for example, all you'll get is a sample at one instant in time when the ADC takes its sample. You can certainly adjust timing parameters so that part of the LED pulse will show up while the ADC takes its sample, however, you won't get the full integration of the signal like you do in normal mode with the integrator in the signal path. If I knew more about what you're trying to accomplish then I can provide better guidance.



  • I understand all the trade offs of using ADC mode. It seems that it is difficult to tell if the TIA is saturating in normal mode, which then makes it hard to test designs, optimize sensor layout and geometry etc. I am currently using this mode to more easily understand the dynamic range of this sensor on a range of conditions and users, normal mode makes this extremely difficult and not strait forward. I plan on using normal mode during normal sampling, and potentially only using ADC to understand the current conditions of the sensor response. If it is a supported mode, I wish that it would be properly documented so I can use it correctly. It seems like if I adjust some of the timing parameters, it takes the ADC sample out of the measurement window causing the signal measurement to not be accurate, but I have not been able to reverse engineer exactly what the timing is that makes sense.

  • I also want to mention that I have a board with the ADPD107, and I am also looking at the ADPD103 Eval board. It seems like on the ADPD103 Eval board in normal mode when I increase the LED to the point of saturating the sensor, the signal will hit a high level and max out. On our board with the ADPD107, when you increase the LED current the response will go higher until a certain point, and then start declining, which I thought made sense from the description in the datasheet on what happens when the TIA saturates in the normal mode. Can you help me understand why I am seeing different behavior on my board vs the 103 eval board? 

  • I accidentally responded to the original question instead of your response, so my response should be below.

  • 0
    •  Analog Employees 
    on Oct 18, 2016 6:19 PM

    In normal mode, the integration window is set up such that LED_OFFSET is approximately equal to AFE_OFFSET +9us. This assumes that your AFE_WIDTH is 1us longer than your LED pulse and you use the fine offset adjustment to optimize for the peak signal. The ADC samples occur at the end of the integration period which is approximately 2xAFE_WIDTH + 1us. When you're in TIA_ADC mode, you essentially want to pull the ADC samples back to the time where the LED pulse is actually occuring so you can take a sample at the output of the TIA while the LED pulse is being processed. In order to do this, you would need to back off on the AFE_Offset setting by 2xAFE_WIDTH + 1us. That should put you at approximately the beginning of the LED pulse. For example: if your normal mode settings are LED_OFFSET = 25us, AFE_OFFSET = 16us, AFE_WIDTH = 4us, LED_WIDTH = 3us, when you switch to TIA_ADC mode and you want to measure the TIA output while the LED pulse is being measured you would change the AFE_OFFSET to 16us - (2xAFE_WIDTH + 1us) = 7us. You can then sweep the AFE_Offset from there until you find the peak of the output of the TIA.