Hi everyone. I am implementing the attached transimpedance amplifier circuit and I have a few issues with it that I would like to resolve. When the LED is incident on the photodiode, the output of the circuit follows the input frequency of the LED. When I remove the LED, the output shows pulses at a frequency of about 2k Hz, and it remains 2k everytime I remove the LED. I am unsure what is causing this, my guess is that it has something to do with the dc-coupling capacitor circuit. This makes it difficult to distinguish when the LED is on or off. I would expect that when there is no light on the PD there should be no current flowing so the output should remain at some constant DC value, not any type of frequency.

In addition, when the frequency of the LED falls below 1 kHz, the output becomes noisy (shakes on the oscilloscope) and remains at a frequency of about 1.3k Hz, and hence does not follow the frequency of the LED. It seems the circuit does not work at low frequencies. Could it be that the dc blocking capacitor is blocking the signal at low frequencies because at these low frequencies the signal approaches DC? Any suggestions on how to alleviate these issues would be great. Thanks!

 

Take a look at this page about photodiodes:
http://home.sandiego.edu/~ekim/photodiode/pdtech.html

Look for the heading:
Zero Bias Operation - Rl<<Rd, load line (b)

If you're using the zero bias circuit, what you should be getting is an output which goes negative (at pin 6 of the op amp) when light hits the diode. But your circuit has no negative power supply! What you need to do is connect the cathode of the diode to a pseudo-ground at 2.5V. Then you should see an output of 2.5V for no light, and for increasing light, the voltage would dip.

As for spurious signal when there's no light on the diode, it has to be noise somewhere. Try removing the .2uF capacitor that carries the signal to the second amp, so that it isn't driven any more--is the noise still there?

You show a 2.5V level going to pin 3, and you're simultaneously generating 2.5V with resistors. I'd generate 2.5V with a voltage follower built from one of the LM324's amps, and use it for all the pseudo-grounds.

The LM324 is a cheap low-quality amplifier. It's slow, and can only swing the output to about 3.5V on a 5V supply. Depending on how fast your light input changes, you might hit the LM324's limitations.

 

The noisy lousy old LM324 has PNP input transistors so its photodiode inverting input floats high. Light on the photodiode causes it to produce a negative signal that is inverted by the first opamp which causes the output of the first opamp to go high.

The second opamp is biased completely wrong. It is inverting with a missing input resistor so its low frequency response is very poor.

 

Thanks for the replies. John-I am configuring it as the zero-bias diagram you suggested. I reversed biased the diode by using a reference voltage on the non-inverting input, however the circuit seemed more sensitive when it was at ground.

I will reconfigure the opamps as suggested. Also, I have a AD817AN. Would this work better than the LM324? Thanks.

 

The AD817 is a high speed video amplifier. It has input bias current too high. It will probably oscillate in your circuit.

Use a Cmos opamp or an opamp that has Jfet inputs then you can use the zero bias or the reverse bias circuit.

 

I thought of the bias current for the amplifier, but then there's dark current in the diode, so to some extent I believe they'd cancel.

Yes, the AD817AN would be better. If you only have one, use it as the first stage. It has an offset input, so you might be able to null it for the no-light situation, or bias it positive so the diode can connect to ground.

And by the way, if there are unused amps in the LM324, you need to disable them by grounding the inputs and leaving the output open. Or what I do is ground the + input and connect the output to the - input.

Audioguru beat me to that one! But experts aggree: not the AD817.

 

the bias current for the LM324 is listed as 45nA.

Nobody sells an opamp that has "typical" spec's. The max input bias current of an LM324 or LM358 is 250nA.

 

True, though the LM324A has max 100nA bias current. Even 250nA probably isn't a serious problem (.25V error) in an AC coupled system. But with such low headroom under the +V supply, it might be getting tight. How about a non-symmetrical pseudoground, at 2V rather than 2.5?

Maybe something like the AD8622 would be right. Still not super fast, but low noise and bias current.

 

Thanks for all the suggestions. I removed the divider after the first opamp and added a resistor to the (-) input of the 2nd LM324. I also increased the decoupling capacitor to 3.3uf, this allowed the circuit to work at lower frequencies. I notice that I have to fine tune the voltage bias of the 2nd opamp, it is very sensitive and if the voltage is not perfectly tuned, the output oscillates terribly if it is slightly off, then either saturates or goes to ground if the voltage is far out of range. In addition, I can only apply a low bias voltage, max .2V on the (+) pin of the first 324 if I want to reverse bias the diode. Anything larger causes oscillation then saturation of the output.

I think I will search for a new opamp to use instead of the AD817. I will see if I can find an AD8622. Thanks!

 

Hey guys. I found a max407 single supply dual opamp in my lab, I think this would be better than the LM324. Is there any reason to think otherwise? Thanks!

 

The MAX407 is probably Cmos. It has very low power supply current because its max output current is also very low. Its max frequency is also low so it will make a mess of audio circuits. Its max supply voltage is only 12V. Low, low and low. Too low.

The LM358 dual and LM324 quad opamps are noisy, have crossover distortion and have trouble above only 2kHz. Their max supply is only 32V.
I replace them with MC33172 or MC33174 or MC34072 or MC34074 opamps with most of the features but none of the problems.

 

Low supply voltage and low frequency are things that I need. I am looking for a single supply op amp that can run on 5V. The frequency range will be from around 100Hz up to 100k. I will not be going any higher than that. Are your conerns that the amp cannot handle high frequency? If it is CMOS than that is what I am looking for, better than BJT right? I will look to see if we have the replacments you suggested.

 

The opamps you posted have very poor high frequency response and hardly any output current.
Keep looking because there are low voltage opamps that have good high frequency response and much higher output current.

Oh, you don't say where in the universe you are located so maybe modern opamps are not available there.

 

I found an opamp, which looks like a special designed device. I was wondering if it would be useful in low level detection from photodiodes:

http://focus.ti.com/lit/ds/symlink/opa128.pdf

In addition, the LMV794, along with the suggested MC34072 seems like it is a good choice for this application. However the LMV794 is only in surface mount. DOes nayone have suggestions on how to test with surface mount devices before making a custom pcb?

Another device, IVC102, has onchip integrating capacitors. How does this device compare to the others? http://focus.ti.com/lit/ds/symlink/ivc102.pdf

I'm basically looking for a device that can measure the smallest current because my signal is very weak.

 

The datasheet for the OPA128 says that it is designed for a plus and minus 15V supply. It says its spec's are derated when its supply is as low as plus and minus 5V.