Photodiode amplifier question

[Fishstickilicious]

I'm trying to amplify the response of a silicon photodiode to a bank of IR (875 nm) emitters flashing at 1 MHz with a duty cycle of ~10%. From what I've gathered on the web, a transimpedance amplifier is normally used with the photodiode to amplify the signal. Since I had a LMC6484 quad op amp lying around, I decided to try using it as a transimpedance amplifier with the photodiode configured in photovoltaic mode as follows:

the photodiode faces the LEDs, and is connected between the "-" input of the op amp and ground.
The "+" input of the op amp is connected to ground
A 1M resistor connects the output of the op amp to the "-" input.
V+ is +9V, and V- 0V.

This setup seems to work OK when the LEDs are driven with 1KHz - tens of KHz square waves, and I observe nice square waves on the oscilloscope when I connect the probe to the op amp's output. When I increase the frequency of the square pulses, however, the square waves distort, turning triangular and finally going flat. The LMC6484 appears to have a gain bandwidth product of 1.5 MHz, i'm wondering if this setup is too fast for the op amp to handle. Are there any good models of op amps that can handle this kind of amplification?

Would it also be possible to use a transistor as an amplifier (with a fast BJT like the SS9018) to amplify this kind of signal, assuming I provide the correct bias voltage? I'm only interested in amplifying the signal to the point where I can see the shape of the photodiode's response.

 

[(*steve*)]

1.5MHz gain bandwidth product... There's your problem there.

How much gain do you require? Let's assume it is a total of 1000. A single op-amp would require a gain bandwidth product in excess of 1GHz for your 1MHz signal -- and that is barely sufficient.

However, if you allow yourself three gain stages, each with a gain of 10, you're looking at a minimum gain-bandwidth product of only 10MHz. Get a device with a 100MHz gain bandwidth product and you stand a fighting chance.

Also note that your signal will likely suffer from poor rise and fall times. You might square them up with a schmitt trigger.

 

[Arouse1973]

Yes Steve sounds like a slew rate issue to me. You need to make sure you have a high enough slew rate if using fast rising edge pulses.
Adam

 

[(*steve*)]

Also a slew rate issue (I meant to mention that -- and of course that is a larger issue at higher output levels.). With a gain bandwidth of 1.5MHz, there's not much gain left at 1MHz, certainly not for the fast rising edges.

 

[Arouse1973]

Yes this is a common problem because people think a 1MHz square wave is the same as a 1MHz sine wave. The fundamental frequency is obviously the same but the square wave with have lots of harmonics which is what will slew rate limit the op-amp. The OP might be better off using a comparator if he doesn't what to use the op-amp for anything other than full scale output.
Adam

 

[Fishstickilicious]

Thanks for the suggestions, guys. I'll go ahead and order up a few more specialized op amps with better slew rates (hopefully I can get some this weekend) for this experiment. As for the use of a comparator, what kind of features are an indication of having a good slew rate? I'm looking at the specs for various models on digikey, but I can't find any direct information on slew rate like on the op amp datasheets. The only advertised spec that I could find on fast comparators was a propagation delay time (nothing about rise and fall times, however). Should I be looking for something like the MAX912 or LT1016? Their datasheets advertise them as being suited for fast operations.

http://datasheets.maximintegrated.com/en/ds/MAX912-MAX913.pdf
http://cds.linear.com/docs/en/datasheet/1016fc.pdf

 

[Arouse1973]

Hello
Comparators don't have the internal compensation capacitors that op-amps have so slew rate is of no importance. That's why it is not on the data sheet. One thing that might catch you out is the output stage maybe open collector so you will need a pull up resistor on the output. You will have to change the circuit a bit and put the diode across the inputs. Take a look at the app note for an idea, look at page 2.
http://www.sophphx.caltech.edu/Physics_5/Data_sheets/AN41_lm311.pdf
Adam

 

[Fishstickilicious]

Thanks for the info again. I decided to not to be a cheapo with the components this time and purchased an LT1016 precision comparator, hooked it up the way you described, added some capacitors in parallel with V+ and ground, and saw a nice 1MHz response that matched the driving signal of my LED (I wasn't interested in the amplitude of the signal so much as the response over time). Good stuff!

 

[Arouse1973]

Thanks for the info again. I decided to not to be a cheapo with the components this time and purchased an LT1016 precision comparator, hooked it up the way you described, added some capacitors in parallel with V+ and ground, and saw a nice 1MHz response that matched the driving signal of my LED (I wasn't interested in the amplitude of the signal so much as the response over time). Good stuff!

Glad to be of assistance. Nice one.
Adam