0
\$\begingroup\$

I am building an amplifier for a phototransistor. It works well under normal lighting conditions.

In my application, the light is very dim and barely produce any readable output.

I tried to increase the feedback resistance potentiometer (R1 in the schematic) from 50k to 1M Ohm and it gave me the gain I wanted but the noise was very high and it was prone to EMI from a nearby power supply.

Would the noise decrease if I used two amplifiers in a multistage configuration?

  • Op amp used: OP07DD
  • Phototransistor: Osram SFH3310

circuit schematic

\$\endgroup\$
5
  • 3
    \$\begingroup\$ For a multistage amplifier, the first amplifier stage is the most critical, system noise performance mostly depends on it. (en.wikipedia.org/wiki/Friis_formulas_for_noise) \$\endgroup\$
    – Long Pham
    Commented Aug 18, 2021 at 9:32
  • 8
    \$\begingroup\$ Is there any reason why you chose that circuit configuration over a transimpedance configuration? \$\endgroup\$
    – Justme
    Commented Aug 18, 2021 at 9:33
  • 3
    \$\begingroup\$ +1 on TIA @Justme. Also, if you have EMI problem from nearby power supply, it’s most probably layout related to both your optical and power supply. \$\endgroup\$
    – winny
    Commented Aug 18, 2021 at 9:34
  • 2
    \$\begingroup\$ @Justme 's transimpedance amplifier suggestsion is worth looking at. \$\endgroup\$
    – Russell McMahon ♦
    Commented Aug 18, 2021 at 11:32
  • \$\begingroup\$ It's a noisy phototransistor so try choosing one that has applications listed that imply a low noise. TIAs come with their own noise problems but you are not at this level yet. Pick a better device. \$\endgroup\$
    – Andy aka
    Commented Aug 18, 2021 at 12:27

2 Answers 2

Reset to default
1
\$\begingroup\$

I would say two amplifiers would be better. You generally want to stay away from resistors in the MOhm range when using op-amps. Also, a lot of noise problems may occur from your PCB design and or lack of decoupling. I would add a 100 nF capacitor between your Vcc of each op-amp and ground. Ensure the 100 nF is as close to the contacts as possible. Then, I would also add a 4.7 uF or a 10 uF on the power rail as well. The position of this one isn't too important as long as your PCB/circuit isn't huge.

From there (assuming you're using a PCB), make sure that you have added a ground pour over the whole circuit to help reduce signal integrity and EMI interference.

\$\endgroup\$
4
  • 5
    \$\begingroup\$ Two cascaded amplifiers will be noise inferior to one. The main benefit of two amplifiers is to gain bandwidth. \$\endgroup\$
    – Andy aka
    Commented Aug 18, 2021 at 12:30
  • \$\begingroup\$ I tried the circuit with two amplifiers with feedback resistors of 50k value and the noise was reduced drastically. I used PCBs in both circuits and will add the decoupling capacitors in the next iteration. I am thinking to add a transimpedance amplifier instead of a pulldown resistor to convert the current output of the phototransistor to voltage. \$\endgroup\$
    – Ahmad Al-Baghdadi
    Commented Aug 19, 2021 at 6:49
  • \$\begingroup\$ Low level signal require high input resistance opamp. I advise you to heed this @Andyaka comments. \$\endgroup\$
    – dsgdfg
    Commented Aug 13, 2023 at 12:05
  • \$\begingroup\$ Modern audio opamps produce less noise than old OP07 opamps. \$\endgroup\$
    – Audioguru
    Commented Aug 13, 2023 at 12:56
0
\$\begingroup\$

A phototransistor outputs a current, so you could wire the opamp as transimpedance amplifier. Link includes all the explanations.

First set the feedback resistor to the gain you want. Then consider the bandwidth of your signal. Then add a capacitor of corresponding value in parallel with the feedback resistor. This will reduce the gain as frequency increases, so it does not amplify noise at frequencies you don't care about. If it is a slow signal, and you're only interested in the DC value, then a -3dB frequency of a few Hz is adequate. You can add another RC lowpass at the output to filter even more.

If it is picking up interference, then you should build it on a pcb with a ground plane. You can put the phototransistor in a grounded metal tube, or wrap some copper tape around it (also grounded) to make sure it doesn't act as an antenna.

You can also add a bit of filtering on the phototransistor power supply, and the opamp input:

enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ A transimpedance amplifier is a great idea, maybe I will add it as a preamplifier then amplify the signal to the desired output using another amplifier so I can get rid of the MOhm resistors. The feedback capacitor looks promising. I will build another iteration with all the comments in mind and will post the updates. Thanks \$\endgroup\$
    – Ahmad Al-Baghdadi
    Commented Aug 19, 2021 at 6:52
  • \$\begingroup\$ If the light to be measured comes from a source that is always in the same place, have you considered using a lens to focus more light on the phototransistor? You could also use more phototransistors in parallel to average out the noise. \$\endgroup\$
    – bobflux
    Commented Aug 19, 2021 at 14:44
  • \$\begingroup\$ The light comes from a fiber optic cable with a 1mm diameter, so I can't put them in parallel. I though about using lenses but I prefer an electronics-based solution than a mechanical one, as I would need to worry about the focal length and the lens diameter. \$\endgroup\$
    – Ahmad Al-Baghdadi
    Commented Aug 19, 2021 at 16:18
  • \$\begingroup\$ Can you get a better coupling between the output of the fiber and the phototransistor, or use a lens at the other end to focus more light into the fiber? \$\endgroup\$
    – bobflux
    Commented Aug 20, 2021 at 10:24
  • \$\begingroup\$ The coupling between them is great, the light coming out is barely visible because the light goes through a liquid sample where most of it get absorbed and I read the remaining passing light. I will try a lens if the amplifier doesn't cut it. Thanks! \$\endgroup\$
    – Ahmad Al-Baghdadi
    Commented Aug 20, 2021 at 17:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.