OPA381: Using the transimpedance amplifier in a photovoltaic configuration

Hi Alex,

the advantage of the OPA381 is the ultra low input offset voltage. Do you need this at all? If not, you could take any other OPAmp as transimpedance amplifier.

What are the parameters of your application? Bandwidth, range of photocurrent, photdiode type, etc.?

Kai

Hello Kai,

I am building a infared spot tracker using a quadrant photodiode (OPR5911, 0.45 A/W). I plan on interfacing the outputs directly to a DSP for proccessing, skipping the analog arithmetic circuitry. 

I am shamed to say that I have yet to determine the range of photocurrent as I planned to do so experimentally. The bandwidth shall be very low - in the order of 100 or so hertz - as it only needs to detect a small amount of radiant energy in its absolute presence(AC coupling would be undesirable, and low frequency noise would be my concern).

Would you say that it is suited for the job? I still need to learn more about the auto-zeroing implementation the OPA381 is using, as I am starting to have a grasp on what it's doing.

Thank you very much for the insight.

Morning Alex, 

yes, the way you have this connected will give a negative going output signal - These opamps can work fine with negative supply but it is unlikely your ADC will want that polarity. More likely you will need to operate positive supply only with a positive DC bias on the V+ inputs. 

Oddly, a similar discussion has been going on here, you might look at the quad OPA4388 I was using there. using a chopper for the Zt stage will remove 1/f noise and give great DC precision. No design can proceed without the source capacitance range. You probably can use a much slower device if only 100Hz is required. 

https://e2e.ti.com/support/amplifiers/f/14/t/886112

Hi Alex,

I would first try to find out the photocurrent range you have to deal with and then decide how to proceed. If the range is very large, you might even profit from modulating the IR light source and using a lock-in scheme (synchroneous rectification). Synchroneous rectification can help to suppress the bad influence of ambient light (50/60Hz driven lamps and sunlight).

Kai

Good morning Michael,

Thanks for pointing me to that thread, I read through it and it did clear things up a bit; very convenient that the OP is trying to do pretty much the same thing, although I am using single-ended ADCs. :)

The OPA4388 really does seem suited for my design(except that it is indeed very fast, but I wouldn't mind it as of now), and for the inverting stage (as you mentioned, I do need and planned on having an inversion stage before interfacing the signal to the ADCs), Do you think chopper amps would be necessary? Sorry that although I now know how they work, I still do not know when are the best cases to use them. If not, I would really appreciate it if you could point me to whatever op-amp in your mind so that I could do more learning.

The photodiodes are spec'd for 10 pF at 10 V reverse bias.

Thank you,

Alex

Ok here are some guesses, 

I went to the 900kHz OPA2378 dual, another RRIO chopper - plenty fast but higher noise, 

Here is what I mean about biasing. Just use a single supply and apply the max ADC input voltage to the + input - then, if the current flows into the inverting node your signal will go down from there, hopefully staying range (with the 500kohm I just guessed at, max input I would be 3V/500kohm = 6uA to reach zero volt at output. 

First sim does not have the required feedback C, that peaking is normal but not desirable, 

Then, calculating the min C for an approx Butterworth gave 2.7pF. That works and if you want to bandlimit more just increase the value,

Here is a V7 file, 

OPA378 Zt stage.TSC

And of course from this earlier schematic you showed, if you want 10V back biasing the diodes you cannot ground them, with 3V on the inverting node, you would need +13V to get a 10V reverse bias - usually these are not too sensitive to bias, so if you have for instance +15V use that - make sure it is well decoupled and low noise. 

Just be careful not to forward bias the detector diodes. The current schematic with +3V on the V+ input and the cathodes grounded would do that.