LM324-N: Temperature problem in photodiode signal amplification

The LM324-N has a noticeable input bias current that varies with temperature (see figure 2). Also, the photodiode itself will have a temperature dependency.

What are the voltages, currents, and frequencies involved? What errors can you afford? The LM324-N is a low-cost device; there are likely to be better choices.

Dear Clemens, light source is current controlled DC so there is no frequency. Problem is not related to photodiode, I have checked with heat source applied to different components of the circuit. Only when applied to feedback resistors I have seen signal change. I didn't see an important change if I heat the amplifier. I need an error of max. %1.

Hello Demir, welcome to the TI forums.

Does the source have to be a fixed light source? A modulated source is easy to detect and measure.  What is the goal, measure the intensity of the fixed source or detect the presence or absence of the fixed light source?

Hello Ron, this is used to detect very sensitive absorbance measurement. Fixed light source seems important to get a standard measurement, this is why I use current controlled LED. According to sample, absorbed light changes so light reaching to detector decrease or increase. I measure light intensity and then convert it to digital value by ADC.

How about a toggled light source on/off at a slow rate?  So two ADC measurements; the delta of the measurements will be the detected light. This will remove offset errors leaving only gain errors.  I expect offset errors are much larger than the gain errors.   

If I understand correctly you want to take difference of ON and OFF states. When light is off I get no signal because I give voltage to non-inverting (+) pin of the amplifier. That prevents noise or other unwanted signal below threshold. I got no important offset error.

Yes, ON and OFF state. However both states need to have both amplifiers in a liner range (not VOL, VOH). 

The non inverting input input voltage and IIB current though the 50M feedback does overdrive the output low so the amount of offset is unmeasurable. IIB * 50M is a substantial voltage. Try LM2902LV for a much lower IIB current for reduced variance (offset error).

With low IIB, pin 1 out has 0.1V OFF and 1.2V at pin 14

Thanks for information Ron. Could you please draw your suggestion if it won't take your time. Do you think it will solve the problem ?

Caner,

50M plus the post amplifier makes this very sensitive. What is the expected range for photodiode current?

For output range, I upgraded to TLV9002

Dear Ron I will check the current again. We take measurements in a very short time (nsec.) and with low light because sample is light sensitive and can affect the measurement if light is bright or have long measurement time. Thank you for the schematic I will try it and tell you the results. 

Caneer,

The provided schematic is not even close to nanosecond compatible. How many nanoseconds?

Caner Demir said:

Dear Clemens, light source is current controlled DC so there is no frequency.

A short time durating counts as frequency.

Dear Ron, input signal is around 1 nA. 

I will give more information in a few days. 

Thanks, Caner.