Hi
I designed a photodiode amplifier, which I put below the schematic and photo board.


I used IC LT1931A to make its negative power supply. I put this part in the datasheet attachment.
My problem is with the noise that is seen on the output. This noise starts from the base of the photodiode, which you can see this noise below.

After amplifying this noise, it amplifies and becomes the following shape.

How can I remove this noise?

pooya

 

Do you see this noise when the unit is placed in 100% dark?

 

Why so little gain at each stage? Op amps are generally more stable at higher gains. What value are your capacitors really? You have them measured in Coulombs!
With three stages, you will have enough phase shift for any coupling between output and input to make an oscillator.

 

What noise are you getting from the LT1931A ?
Thats not the quietest DCDC on the block ,

Can you share the picture of the dcdc switcher ?

 

Nice little board.

Do the noise peaks on the output happen at about the same time as any transitions on the LT1931A?

 

hi pooya,
How are you light stimulating the photodiode.??

I would say your 1st stage OPA has a problem, it is causing some inversion of the negative swing of the signal, ref this image.
Image shows the two waveforms super imposed for comparison.

On the PCB, the -3.3v supplies the power to the OPA's and the PD negative bias.!!
The long cross board wire link is the -3.3V source, it is a very crude way to bias the PD, it will introduce noise into the amplifiers.

E

 

Your noise is at about 1.6 MHz,
is that not the frequency range of the DCDC ?

 

hi andrew,
Is that query for me or the TS.??

E

Aside:
I have observed that by not using a simple addressee header, we can get confused about whom is being asked the question.E

 

The gain of U2 and U3 is only 3.3 times for each one.

 

Do you see this noise when the unit is placed in 100% dark?

No, this is when it is in an environment with normal ambient light

 

Why so little gain at each stage? Op amps are generally more stable at higher gains. What value are your capacitors really? You have them measured in Coulombs!
With three stages, you will have enough phase shift for any coupling between output and input to make an oscillator.

I did not understand what you meant. Can you explain more?

I designed this circuit to be able to receive very low light intensities, but my photodiode does not work exactly according to the data sheet and passes more current, so I had to reduce the gain of each step.
What do you think I should do better?

 

Nice little board.

Do the noise peaks on the output happen at about the same time as any transitions on the LT1931A?

thank you.
I will check and send you the information.
But the frequency as noise on the board is about 2.2 MHz.

 

hi pooya,
How are you light stimulating the photodiode.??

I would say your 1st stage OPA has a problem, it is causing some inversion of the negative swing of the signal, ref this image.
Image shows the two waveforms super imposed for comparison.

On the PCB, the -3.3v supplies the power to the OPA's and the PD negative bias.!!
The long cross board wire link is the -3.3V source, it is a very crude way to bias the PD, it will introduce noise into the amplifiers.

E
View attachment 239366

Hi ericgibbs
I made this circuit to receive pulses of about 100 nanoseconds. And with the TORLABS photodiode I compared, it reports about 20 nanoseconds more bandwidth.
I have already asked for the first part of OPA inside another site and it says that this will create less noise.
I was able to receive the wave correctly by inverting it 3 times.

What do you think about preventing noise from being transmitted from the source to the photodiode?

 

Your noise is at about 1.6 MHz,
is that not the frequency range of the DCDC ?

Yes, my noise is exactly 2.2 MHz, which is the same frequency as the DCDC source.

 

There are some application in which it is just too much trouble to use a switching power supply.

I would suggest at least temporarily moving the LT1931A to a separate assembly separated by about 30 cm of wire, then experiment with various types of filtering to see what it takes to remove the noise from the output. Once that is accomplished, try holding the DC to DC supply close to the preamp and see whether you are going to need shielding.

Given that your input stage is a transimpedance amplifier it will be particularly sensitive to capacitively coupled noise.

Edit: How much bandwidth do you need with that amplifier? Would filtering the 2.2 MHz solve the problem?

 

Yes, my noise is exactly 2.2 MHz, which is the same frequency as the DCDC source.

Cause and effect ?
I'd suggest not to use a noise DCDC in this case,

or if you do , you need to filter its output well, may be using a low noise LDO,

 

No, this is when it is in an environment with normal ambient light

Modern lights often flicker at high frequencies, especially LED and fluorescent lights.

Check it in 100% darkness and see if it goes away.

 

hi pooya,
I will bring your attention again to the layout and routing of the -3.3V supply line.

The -3.3v PD bias line is being used to bias the PD and power the following OPA's.

When the -3.3v current demand for the OPA's change, due to the changing OPA signal levels.
The long wire from the -3.3v voltage regulator to the OPA minus supply causes the bias to the PD changes, which injects a signal into the PD circuit.

If you look closely at the superimposed waveforms I posted, it clearly shows this effect on the negative swing of the signal noise.

Use a separate -3.3v supply for the PD, also rework the layout of the -3.3v line.

I am not saying that the 3.3v generator is not causing the switching 'sine wave' noise, but the layout and by using the -3.3v for PD and OPA's is making the problems worse.

E

 

Most trandconductance amplifiers I have encountered were shielded from stray electric fields. I am not familliar with LT1931 but I would bet that one of those capacitors in the circuit is being switched between being in parallel with the 12 volt input and being in parallel with the output capacitor. With that capacitor hopping up and down 2.2 million times per second it would radiate a very significant electric field.

 

There are some application in which it is just too much trouble to use a switching power supply.

I would suggest at least temporarily moving the LT1931A to a separate assembly separated by about 30 cm of wire, then experiment with various types of filtering to see what it takes to remove the noise from the output. Once that is accomplished, try holding the DC to DC supply close to the preamp and see whether you are going to need shielding.

Given that your input stage is a transimpedance amplifier it will be particularly sensitive to capacitively coupled noise.

Edit: How much bandwidth do you need with that amplifier? Would filtering the 2.2 MHz solve the problem?

Hi DickCappels
Thanks for your complete tips.

I need frequencies less than 2.2 MHz.
What filter do you suggest?