Negative voltages from a non-biased photodiode and transimpedance amp

Hi all,

I'm using a Thorlabs FDS100 photodiode as a light detector for a project I'm working on. The current it produces is fed to a current-to-voltage transimpedance amplifier, the op-amp I'm using is a CA3140E. Now, since the photodiode isn't reverse-biased, I should expect a "dark" current of 0A, and thus a dark voltage of 0V. However, when the photodiode is completely isolated from light, I'm seeing a negative voltage! It's incredibly tiny, around -1.5mV, but this is enough to confuse the amplifiers downstream.

Once the signal is produced from the transimpedance amp, it's fed to an INA126P instrumentation amplifier, and that signal is then passed to another CA3140E which acts as a buffer. From there it's fed into an A/D for processing.

The buffer amp is non-inverting, and I have the signal fed into the positive input of the instrumentation amplifier. The negative voltages are playing tricks on the whole system.
I can't put an inverting buffer anywhere, because the system behaves as expected when exposed to light. It's these dark readings that are screwing everything up, and I can't just neglect them because the whole point is for particle detection; I need an active scanning and recording system.

Thoughts?

Can we see a schematic?

My first thought is op-amp offset voltage. The CA3140E is not a precision op-amp and its datasheet quotes an offset voltage as high as high as 15mV (5mV for the CA3140AE version). So 1.5mV is well within its specification.

Look into using a "precision op-amp", such as the MAX9636AXT+T. 0.1pA of input bias current and 10uV of input offset voltage should work a lot better.

--
The Ruggeduino: compatible with Arduino UNO, 24V operation, all I/O's fused and protected

RuggedCircuits:
Can we see a schematic?

My first thought is op-amp offset voltage. The CA3140E is not a precision op-amp and its datasheet quotes an offset voltage as high as high as 15mV (5mV for the CA3140AE version). So 1.5mV is well within its specification.

Look into using a "precision op-amp", such as the MAX9636AXT+T. 0.1pA of input bias current and 10uV of input offset voltage should work a lot better.

--
The Ruggeduino: compatible with Arduino UNO, 24V operation, all I/O's fused and protected

But what about -1.5mV? I've replaced the CA3140E with an LF351N. Also not a precision op-amp but I'm seeing slightly better behavior. I'll dig up a schematic as soon as I can. Also, when selecting a precision op-amp, should I look for low input bias current or low input offset voltage?

Offset voltage is +/-, so 5mV means +/-5mV.

The LF351 offset voltage is also pretty bad -- you could just be getting a "lucky" part.

In your application (before seeing the schematic) I think both offset voltage and input current are important, though op-amps tend to be better at minimizing their input currents (picoamps) than their offset voltages for a given cost.

--
The Flexible MIDI Shield: MIDI IN/OUT, stacking headers, your choice of I/O pins

RuggedCircuits:
Offset voltage is +/-, so 5mV means +/-5mV.

The LF351 offset voltage is also pretty bad -- you could just be getting a "lucky" part.

In your application (before seeing the schematic) I think both offset voltage and input current are important, though op-amps tend to be better at minimizing their input currents (picoamps) than their offset voltages for a given cost.

--
The Flexible MIDI Shield: MIDI IN/OUT, stacking headers, your choice of I/O pins

Thanks for the input. I just ordered a variety of low offset voltage precision op-amps for testing.

But what about -1.5mV?

It could be the dark current of the device. That is the signal it produces with no light. This is due to thermal agitation of the silicon in the I region of the diode causing electron hole pairs to be produced.
It also could be the DC offset of the amp as RC says. Probably both.

Maybe a 1M pulldown would help that, when I use a regular led I find it neccesary to use for steady results