In-amp amp for reverse biased photodiode

Discussion in 'The Projects Forum' started by thodoris46, Oct 11, 2014.

  1. thodoris46

    Thread Starter New Member

    Jan 29, 2013
    10
    0
    Hello everyone, i designed a card to Read with an ADC the current of a photo diode InGa .
    I have designed something similar with the following modules
    -TIA (Transimbedance amp)
    -sallen key low pass filter
    -16 Bit ADC
    -In Another card High voltage supply for the photo diode

    The previous design had a problem with 50 Hz humming (I'm pretty sure that it passes through magnetic field) I shielded the system with copper tape and worked fine [​IMG]

    Now this photo diode has a heatsink that the photo diode fits inhttp://www.hamamatsu.com/eu/en/product/category/3100/4007/4175/A3179/index.
    Out of the heat sink comes a coaxial cable where the center conductor is the anode and the shield the cathode , it has another cable for TEC cooling but there is no noise there.
    I have not been able to shield the system and I'm going to try cutting the humming with an instrumentation amplifier.
    The expected signal is dc and the noise is 50 Hz with harmonics in 75 and 100 Hz.

    I made a design (pdf) but it must be wrong .. because when you get the signal from the first op-amp (TIA) it changes it's phase, so on the second op-amp (IN-amp) the 50 Hz noise will have a difference in phase from signal to Ground.

    Another approach is putting the anode-cathode of the photo diode on the inputs of an in-amp, but where am i going to put the 0,5 V reverse bias voltage and how will the 0,5 V bias voltage effect the in-amp [​IMG]

    help guys...
     
  2. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    I can't get your link to work, but usually photodiodes don't have heat sinks because the is very little current.
     
  3. DickCappels

    Moderator

    Aug 21, 2008
    2,659
    632
    http://www.hamamatsu.com/eu/en/product/category/3100/4007/4175/A3179/index.html (The heatsink is the hot side of a thermoelectric cooler for the photodetector -the cooling is to reduce noise).

    First things first: Is the signal coming out of the transimpedance amplifier clean? If it is clean coming out of the transimpedance amp, then the 50 Hz is probably the result of a ground loop between the transimpedance amplifier and the ADC. If that is the case, then it might help to add some voltage gain directly to the output of the transimpednace amplifier so the signal will be lager with respect to the ground loop.

    The differential amplifier seems to be properly connected to reject a ground loop.

    With the wide bandwidths of the amplifiers with respect to 50 Hz, the phase shift is a puzzle. Could it be that your scope probe is AC coupled?

    One thing to pay attention to is power supply hum, particularly in the photodiode bias circuit. Maybe your monolithic regulator can give you 60 or 70 db of ripple rejection, but sometimes, that is not enough. You might have to have two stages of regulation in order to get the 50 Hz low enough.
     
  4. thodoris46

    Thread Starter New Member

    Jan 29, 2013
    10
    0
    If you hit on hamamatsu search A3179 you can see the heatsink it's for TO-8 case.
    InGaAs work in -10 Celsius, they need cooling systems
     
  5. thodoris46

    Thread Starter New Member

    Jan 29, 2013
    10
    0
    I need to clear something in my post. I have a pcb in my lab that is almost the same design as the one attached, instead of the in-amp it has a sallen key filter.
    The design attached it has not been manufactured yet, i'm using the in-amp to cancel the humming of the system.

    Thank you for the answer
    No the signal has humming at the output of TIA. When i take a measurement without the PD by sort circuiting the inverting input with ground, 50 Hz is gone. The system is in a small suitcase to have dark conditions. When i ground the metal parts of the suitcase on true ground or reference ground the 50 Hz is gone. That leads me to think that the pcb card is not responsible for the humming and its because the wire-photodiode-heatsink system.

    How can i measure the phase difference between signal and ground with the scope ?

    I will consider of two stage regulating, thank you
     
  6. DickCappels

    Moderator

    Aug 21, 2008
    2,659
    632
    You have one place where you will do best to concentrate your effort -the inputs to the transimpeance amplifer, its shielding, and grounding. Ok, and the bias supply too. If you get those right, then you will have a clean(er) output from the transimpedance amplifier and you won't have to worry about instrumentation amplifiers.

    Transimpedance amplifiers with high transimpedances that I have seen have the entire transimpedance stage shielded. You might consider that.

    I think your hunch about the cable is a good one. The cable is evil. It adds capacitance which increases the amplifier's noise gain, and in conjunction with other ground paths, such as the thermalelectric cooling supply, can act as a loop antenna that magnetically picks up hum injects the hum into the input of the transimpedance amplifier. For those reasons, it would be a good idea to make the shield as short as possible. If possible, eliminate it and build the preamplifier onto the back of the heatsink.

    To measure the signal phase with respect to ground, ground your scope probe (a differential probe would be ideal, but hey--- this is only 50 Hz) and connect the probe tip to the signal. Note the phase with respect to whatever you want to compare it with. If you have to AC couple, make sure you use a large enough coupling capacitor so that it introduces negligable phase shift. For a 10 Meg Ohm probe, use 1 uf for a 1 second time constant -should be fine.

    One other thing, if your signal has a low bandwidth, you might be able to deal with the 50 Hz by low pass filtering the signal.
     
  7. thodoris46

    Thread Starter New Member

    Jan 29, 2013
    10
    0
    Thank you DickCappels i'll give another shot of solving this by proper grounding shielding.

    By the way the fact that the anode, cathode cable is a coaxial it makes me think what if the shield of the cable the cathode acts like an antenna ??? one side of the shield goes to the TIA input (cathode) the other side goes nowhere ..the PD has no case pin to put the shield..
    Will try that tomorrow.

    What a meant about the phase difference is, if you have a sine wave in the inverting input of the TIA and the same sine wave at ground with phase difference between them 0 degrees.
    At the output of the TIA will there be a phase difference between the sine wave on the output track and the sine wave in the ground track?
     
  8. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    Why the 100k in series with the bias voltage? Isn't this the shield connection?
     
  9. thodoris46

    Thread Starter New Member

    Jan 29, 2013
    10
    0
    It forms a voltage divider with the 28K to get -0,5V
     
  10. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    Without knowing what diode you are using I don't know what the bias voltage can be, but usually higher than .5 volts and just reduces capacitance so increasing speed.
    If it is .5 volts I would make it a much lower impedance and filter it.
     
  11. thodoris46

    Thread Starter New Member

    Jan 29, 2013
    10
    0
    I'm using the InGaAs photodiode,hamamatsu G12183-105K.
    Lower the impendance to draw more current or for noise level?
     
  12. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,292
    1,255
    Ok, I see from the datasheet only 1 volt reverse bias. So to make the shield work better (lower noise) try 28 ohms and 100 ohms for the divider. I don't know if this is your problem, but it should help. Let us know how it works.
     
  13. thodoris46

    Thread Starter New Member

    Jan 29, 2013
    10
    0
    thank you ronv lowering the voltage divider did help.I did three changes.
    1.Made a very small card with the TIA to put it in contact with the PD pin's
    2.I lowered the feedback gain of the TIA to 220 Kohm(i was using 1 mega)
    3.I changed the coaxial cable to a twisted pair unshield and the shield with copper tape the photodiode pins and the small card with the TIA.

    the third change makes a big difference. I found something interesting..
    http://www.ti.com/lit/an/sboa061/sboa061.pdf

    • Shield the photodetector circuit in a metal housing. It is a
    very high impedance, high sensitivity circuit and it requires
    good shielding and effective power supply bypassing. This
    is not optional
     
Loading...