I-V converter for PMT


My name is charles.

I'd already used the the high speed video amp(AD811 and AD8057)and ultra low input current offset amp(AD549) respectively in my simple I-V converter circuit to get a singnal from PMT(2uA max) with Rf(100Mohm). 

Although I-V converter circuit done well, signals from those amps have extra tails.

I thought that it is related to speed of amp. So I decided to change the amp has low input current offset and high bandwidth.

Now, I'm trying to order samples, AD8610, AD8651, ADA5817-1 and AD8065 for converting the current from PMT(2uA maxium) with Rf(10Mohm) to voltage.

(Not yet ordered..)

So, is it ok to use those kinds of amp(AD8610, AD8651, ADA5817-1 and  AD8065)  as a I-V converter without any tails?

Please let me know a tip or pice of advice about it.


Best regards.

P.S. The singnal frequancy from PMT is about 25MHz.

        Photon come from confocal laser scanning microscope system.

  • 0
    •  Analog Employees 
    on May 20, 2013 7:29 PM

    Hi Charles,

    I apologize for the delayed reply on this. What are you referring to when you mention about extra tails? Do you see this at the output of your circuit? Would you mind sharing your scope shots so I can better help you out? Also, what board are you using?



  • Hi Charles,

    I'd say a couple of things. Firstly, 100M is huge for a feedback resistor if the maximum signal current is 2uA. If you think about it, you need an amplifier output swing of 200V at full scale. You need a feedback resistor in the 1M to 5M range. You have to remember that the amplifier must be kept out of saturation for it to maintain the virtual earth.

    Secondly, what PMT are you using, and what speed are you trying to achieve. You need to achieve a good virtual earth across the frequencies of interest, so you will need a suitably fast opamp. This means your circuit capacitances need to be low, specifically the PMT anode output net. Your amplifier needs to be right by the PMT, not at the end of a piece of coax. I used to design confocal electronics, so I know how strongly this affect performance. You will also need to at overload protection as the PMT will easily produce more than its maximum rated output if you get a bit of reflectance from your sample.

    To recap, the tail may be caused by the PMT itself, but I suspect that you are not maintaining a good virtual earth at the inverting input.

  • Hi all,

    After all, thanks a lot for your answers. To tell you the truth, It is too difficult to explain my problems.

    So, I'd like to show some images from my system. I totally agree with OliverSedlacek's opinions, and I changed the feedback resistor into 150k by using the tee-network method.

    ( According to http://www.idc-online.com/technical_references/pdfs/electronic_engineering/High-Gain%20Transimpedance%20Amplifier.pdf)

    It's so simple circuit with AD8065 amp.


    • FET input amplifier 
    • 1 pA input bias current 
    • High speed: 145 MHz, -3 dB bandwidth
      (G = +1) 
    • Low cost 
    • 180 V/µs slew rate (G = +2)

    I'm using a hamamatsu current output type PMT, H7422-40 with a power supply unit(C8137-02) from that commpany, which has a function of overload protection,and connecting the BNC cable to above circuit. From the I-V converter to DAQ card on my computer, an out-put line is also connected by BNC cable. PMT is located above the object lens directley. All cables are short, as possible as I can, to avoid the stray capacitance.

    fig1. 512x512 fluorescent bead image with 535nm bandwidth filter 1.12 sec/frame .

    fig2. 512x512 fluorescent bead image with 535nm bandwidth filter 2.71 sec/frame .

    As you can see, the single bead(a small white particle) of first image has "extra tail" like a shooting star comparing with second one.

    The differences of those two images are "frame rate" for getting 512x512 images, first is 1.12 sec/frame and second one 2.71 sec/frame respectively.

    I used AD549(ultra low Ios Amp), AD811(video rate amp) and AD8605(transimpedance  amp).

    In case of an AD549, it had a logger tail than AD811 and AD8065. AD811 and AD8065 had same result have same tail.

    So, unfortunately until now, I couldn't solve this problem. I'll check the virtual earth again.

    Should I use other circuits with 2 steps Amp.?

    P.S.  The source beam is an ultra-fast femto second laser (two photon excitation imaging system: Ex: 790nm Em: 535nm with IR cut off filter)

    Thanks again,

    Best regards.

  • Hi Charles,

    Are you sure you haven't got the two frame rates swapped when referring to those images. I'm estimating that at 2.7 FPS your pixel dwell is around 1usec and at 1.1 FPS it's 3 usec. I would expect the shorter pixel dwell to show stronger smearing as the tail become more significant.

    Using the T network doesn't improve your virtual ground if you are still looking at the same effective transimpedance. You will have to use two or more stages of amplification so that you can reduce your first stage gain substantially.

    The coax cable also causes these sorts of problems, so if a lower front end gain doesn't work you should look at that. How long is the coax, and do you know what sort it is? Stuff like RG174 really isn't terribly good.

  • Charles,

    I'm a bit mystified by your signal levels. Fluorescent beads are usually pretty bright, so I don't know why you need so much electronic amplification. The H7422 is an excellent choice for your application, with excellent QE, good speed and reasonable gain. I was using an equivalent transimpedance of around 500k when I was designing confocals.