photodiode AC amplifier

Dear Bruce,

Thank you very much for your comments and reply.

The circuit seems to be very appropriate for my application, but although it worked in a simulation (Tina), I haven't been able to make it work on the protoboard. I am not sure if the photodiode model I used in the simulation is different from the one that I am using. I will keep trying today...

On the other hand, what do you think about using directly transimpedance amplifiers? I have seen them in Pulse Oximeter designs (as described in the Medical Applications Guide, p47 http://www.ti.com/litv/pdf/slyb108e). Specifically:

OPA380 Precision, High-Speed Transimpedance Amplifier, or
IVC102U Precision Switched Integrator Transimpedance Amplifier

Thanks again!!
Roberto

Roberto,

The OPA380 and IVC102 would not seem particularly suited to your application. The OPA380 is effective at achieving accuracy at low light levels with its auto-zero technique--not important in your application. The IVC102 integrates the input signal. This is great for averaging (accumulating) output over a period. You are trying to extract an AC signal. This would require very fast integration and sampling at at least 2x your modulation frequency.

You were not specific in your description of problems with the two-op-amp loop that subtracts the average level. If you can provide details, perhaps I can help.

Regards, Bruce.

Thank you very much Bruce.

Bruce Trump said:

The OPA380 and IVC102 would not seem particularly suited to your application. The OPA380 is effective at achieving accuracy at low light levels with its auto-zero technique--not important in your application.

Although the received signal from the reflection is weak, I hadn't realised that the transimpedance configuration with the OPA380 was aimed for low light levels.

Bruce Trump said:

You were not specific in your description of problems with the two-op-amp loop that subtracts the average level. If you can provide details, perhaps I can help.

I don't have the OPA320, so I am using the TLC2274 while I get the OPA320 (10 days). The simulation with the TLC2274 and the AC signal 3kHz works fine. But I haven't been able to get the design working in a protoboard. Maybe the photodiode I am using is different from the simulated model, and I don't have all specs from my IR photodiode (is a cheap one).

I reduced the feedback resistance R1 to 220k. And I only start to get some AC signal if I use a larger resistance R4 between the T1 emisor and ground of about 50k, however with these values Vout is not hold at an average voltage of 2.5V and instead is almost at 5V.

Regards,
Roberto

Dear Bruce,

I have not been able to get the circuit working properly. I am more convinced that the problem is that I don't know my photodiode specs, so the Tina model is different from the actual circuit.

However, I would like to make sure I am not missing the point, and that the proposed circuit is appropriate for achieving my goal.

1. What I would like to achieve is to detect small changes in the reflectivity of a surface in front of the photodiode (2cms). For example the differences between a card gray 20% and another card gray 40%. A complete change of surface may take between 100ms to 500ms. I would like to detect the shape of the change and how long a surface is placed on front of the photodiode.

2. I am using an IR LED and Photodetector because the light must be invisible to human eye. These elements are placed in paralell, pointing in the same direction, and I am trying to distinguish a different reflectivity depending on the amount of IR light that goes out of the LED and comes back to the photodiode.

3. The main challenges are:
a) the LED radiant intensity is limited to 4,4 mW/cm2 for safety reasons;
b) there are different sources of interference such as light from bulbs (50Hz) and from the sun; and
c) there are some mechanical vibrations in the ambient

3. The photodiode I am using is is Everlight PD438B (datasheet attached), but I can change the LED and Photodiode. Moreover, the final design should have surface mount LED and photodiodes.

4. Depending what card I would like to detect, the current induced in the photodiode is approximately 200nA different

5. Because I don't care the reference level, only changes in reflectivity, I decided to modulate the LED signal in order to be able to filter out the interference (sun and bulb light) received in the photodiode. Then I am planning to rectify the signal in order to quantify the average reflectivity level.

Hopefully, these details help to understand better my problem and confirm if your proposed circuit design is still the suggested one. In that case, I will keep trying with it.

Thanks, Roberto

Roberto,

From what I can understand of your application, I think the recommended circuit is well suited. I have inserted the reference circuit below so that we can refer to circuit values and voltages. You have not been very specific in describing what is "not working" in the circuit. Can you provide a diagram of the circuit you have implemented with circuit values and supply voltages. Probe the critical nodes and give me a sense of what is not working. Are the op amp in their linear range. Are voltages as expected?

I doubt that lacking an accurate model of the photodiode is the problem. I think there is something more fundamental wrong with your implementation.

Regards, Bruce.

Thanks again Bruce.

Below I am attaching a picture of the circuit I am using, which is your adaptation to single supply with minor changes:
R4 10k (instead of 2,2k)
R1 220k (instead of 1M)
C2 47p (instead of 5p)

In addition, I added actual measured voltages in some nodes to illustrate circuit behaviour. If R4=2,2k, there isn't oscilation anywhere. If R4=10k, there isn't any oscilation withouth significant reflected light back to the photodiode. But increasing reflected light, suddenly produces oscilation in both U1 and U2 inverting inputs (as shown in the measured voltage). However, the beahaviour is different from the simulation, and there isn't measurable oscilation in Vout.

Hope this helps you to understand what's wrong with my implementation.

Thank you very much,
Roberto

Roberto,

I'm traveling and cannot devote full attention to this right now. I have made a quick check including simulation with your circuit values and op amp. I see no problems.

I'm not sure I understand a few of your notes. You indicate a 1.3kHz square wave at the photodiode. Is this your modulation frequency? What is the frequency of the unwanted oscillation?

Do you have good bypass capacitors on the supply pins? How is the 2.5V offset derived? Are there bypass caps on these nodes.

I'll look in more detail this weekend but answers to these questions will help.

Regards, Bruce.

Dear Bruce,

Thanks again for your attention!

Bruce Trump said:

I'm not sure I understand a few of your notes. You indicate a 1.3kHz square wave at the photodiode. Is this your modulation frequency? What is the frequency of the unwanted oscillation?

Exactly. 1.3kHz is the modulation frequency of the square wave that feeds the LED. This light is reflected back to the photodiode. The frequency of the unwanted oscilation is 50Hz from lightbulbs and DC sunlight which I am testing now. Also, I would like to be sure not to have problems with other frequencies such as 60Hz in other countries or other type of common interference in the workplace.

Bruce Trump said:

Do you have good bypass capacitors on the supply pins? How is the 2.5V offset derived? Are there bypass caps on these nodes.

I am not sure what you mean about "good bypass capacitors". I am using a 0.1uf capacitor in parallel with the voltage source. Is this what you are asking? What should I take for "good"? In addition, I derived the 2.5V offset as shown here:

Thanks Bruce and I hope you are having a good trip.

Regards,
Roberto

Dear Bruce,

Thanks for your comments and suggestions. I am trying to improve as you suggest including the old remote dark cover, but I have a more fundamental doubt that I would like to clear before trying to optimize the circuit.

If I put a bright surface in front of the LED and reflect back to the photodiode a large amount of IR light, I get (2500mV DC + 2500mV sq) at the photodiode output. So, is it right to get (3100mV DC + 10mV sq) at U1 output? In the simulated circuit, a low sq voltage at the photodiode output (4mV) produces an big volgage at U1 output (400mV). Moreover, DC component does not rise from 2,5V to 3,1V.

Is this behaviour expected with the proposed circuit?

Regards,
Roberto

Dear Bruce,

Thanks for the guidance. I think I was able to get to the source of my problems by following your comments.

I did the test you suggested, but the U2 output (3,1V DC) didn't change with variations of R4 (1k - 30k) nor with variations of DC ambient light. So, I used another Opamp that I had at hand (LT1014), and it worked better. With maximum DC ambient light, the maximum value of R4 was 4,7k before U2 output reached 5V.

Apparently, the main issue was not having the OPA320 so I will get this before any further tests.

Lastly, I would like to ask what do you think of adding a synchronous detection circuit for the next stage using for example the OPA675. As I will have a square wave plus 2,5V DC at U1 output, but I would need a single number representing the level of reflected light, I was thinking in demodulation using synchronous detection followed by a low pass.

Thank you very much for your support and guidance on this project.

Kind regards,
Roberto

Roberto,

A synchronous detector is the ideal way to process the signal.

You are welcome and best of luck,

Bruce