Guitar - Photodiode Amplifier, Signal-to-Noise ratio

[Yarrrrr]

Hi,

If nothing brings an improvement ... maybe there are systematic mistakes or measurement mistakes.

Maybe...Unsuitable wiring, unsuitable PCB layout, unsuitable GND plane, unsuitable measurement equipment...

Klaus

Measurements are made in the computer, Circuit is plugged into XLR on my audio interface, and I am comparing the frequency spectrum between changes. There is a picture of how that looks a little further up on page 3.

Unsuitable PCB layout, sure, most likely. This is the only PCB I have ever designed...

I could upload the DipTrace project files if anyone is interested.

 

[KlausST]

Hi,

No fast power supply capacitors, no bandwidth limiting capacitors.
Not that good GND plane.

How do you avoid "noise" by ambient light?

Did you do some reference measurements with OFF IR LEDs, and with shorted photodiodes?

Klaus

 

[Yarrrrr]

Hi,

No fast power supply capacitors, no bandwidth limiting capacitors.
Not that good GND plane.

How do you avoid "noise" by ambient light?

Did you do some reference measurements with OFF IR LEDs, and with shorted photodiodes?

Klaus

Could faster power supply capacitors help with broadband noise?

Bandwidth limiting capacitors is something I have thought about but I am assuming it would be more applicable if I had an amplifier per string and could use it for "low-pass" at the maximum frequency range of each individual string.

I have tested the circuit in a completely dark room + blocking the diodes. it shows no change in the noise floor.

Could a better ground plane, however that looks like help with broadband noise issues? I already solved all EMI related noise issues since I got this proper PCB with a ground plane + shielded enclosure.

I have measured with a resistor instead of the photodiode before, and that resulted in a worse noise floor than with a photodiode, probably because the resistor value though.

I do wonder about the Phototransistors though, why does it seem like their output(and all different brands and sizes of photodiodes, with different specs) are the same?
I really hoped they could give me a stronger input signal, but perhaps I need to use them differently in the circuit.

Here is how the noise looks like:

 

[Audioguru]

Your noise level of only -12dB at 4kHz is horrible. It should be -60dB to -80dB.
I think your photodiodes are producing "dark current" noise since they are reverse-biased.
See how to fix it here:

 

[Yarrrrr]

Your noise level of only -12dB at 4kHz is horrible. It should be -60dB to -80dB.
I think your photodiodes are producing "dark current" noise since they are reverse-biased.
See how to fix it here:

We tested that theory before in the thread, and the original arrangement was like how you are describing with connection to +V/2. It makes no difference in the noise output.

And the picture was just meant to show it is broadband noise. The yellow dB indicators is for the range of the EQ bands, the grey numbers to the far right is probably what you should look at to determine the actual level.

 

[Audioguru]

Then you have a very low noise level of -90dB at 4kHz.

 

[Yarrrrr]

Then you have a very low noise level of -90dB at 4kHz.

Sure, if we assume the inherent circuit noise is good. How do I get the input signal to be stronger so the actual signal to noise ratio improves.

Any thoughts about the Phototransistor and Photodiode(and different brands of photodiodes) giving me what seems to be exactly the same SNR?

For reference, left side is my circuit with a note playing, right side is a typical guitar sample.

 

[Audioguru]

Your left photo shows all the odd harmonic numbers of a squarewave (severe overdrive clipping) plus the even numbered harmonics of the photodiode producing a DC bias causing the output to have DC offset (one side of the signal is clipping more than the other side).

The right photo shows the same distortion harmonic frequencies at different levels but with less high frequency response and less noise.

 

[c_mitra]

A quick update on a few things I have tried that hasn't worked.

• Tried several different Photodiodes - negligible difference in sound.
  
  
• Different IR LEDs - negligible difference in sound.
  
  
• Tried a Phototransistor (WÜRTH 1540051NC2590)- no difference in signal strength, SNR or noise floor.
  
  
• Lower values for the feedback resistors - no SNR improvement ( as the phototransistor gave no stronger output ).
  
  
• Different resistor values for the voltage reference - no difference in noise floor.
  
  
• Using battery power instead of power supply - no difference.
  
  
• Tested a bunch of different Op amps(LM4562, OPA161

If anyone has any ideas of why nothing I try changes anything, I'd love to hear them.

I do not have clear idea but still...

1. How did you measure the noise? What kind of setup you are using?

2. What kind of output you get when the guitar is not being played: I mean the amplifier is on but the strings are not being played.

3. What the four op-amps in parallel are expected to achieve?

4. How was the circuit layout? Any photo?

A couple of suggestions:

1. Replace the photodiode with a 1M resistor and measure the noise

2. Bypass the power supply pins for the op-amps with a 0.1uF ceramic cap at or close to the mounting point. Ans measure the noise.

You are assuming that the noise is coming because the parts are noisy. I suspect that the noise you are getting is coming from the circuit design and layout.

 

[Yarrrrr]

1. How did you measure the noise? What kind of setup you are using?

Explained at the top of this page.

2. What kind of output you get when the guitar is not being played: I mean the amplifier is on but the strings are not being played.

There are pictures further up on this page showing the noise, the noise with a string vibrating, and the output of a normal guitar pickup with a string vibrating...

3. What the four op-amps in parallel are expected to achieve?

Reducing noise under the assumption that the op-amps were not low noise enough.

4. How was the circuit layout? Any photo?

Shown further up on this page.

The last thing Audioguru said about the noise was that it is very low. And as evident by replacing the op-amps with much lower noise ones, and all other changes, showing no change in noise floor I am inclined to agree that the curcuit is actually low noise already.

But the strength of the input signal is too weak.

There is a significant enough difference in signal strength by increasing the current for the LEDs so they shine brighter. But it has already hit diminishing returns. I had hoped a more sensitive Photodiode / Phototransistor would give stronger output, but in my tests they are all identical.

 

[c_mitra]

There is a significant enough difference in signal strength by increasing the current for the LEDs so they shine brighter. But it has already hit diminishing returns. I had hoped a more sensitive Photodiode / Phototransistor would give stronger output, but in my tests they are all identical.

There is no substitute for a systematic analysis.

Replacing components at random will not produce the desired result.

First you need to locate the source of noise OR why the signal is weak.

Begin at the beginning. Start with the photodiode signal. If the photodiode is covered with a black tape, do you get the low noise?

If the photodiode is replaced with a 1M resistor, do you get low noise?

If the answer is yes, the noise comes from the PD. And that needs attention.

If the answer is no, the noise comes from elsewhere. Focus on the amplifiers.

Do you have bypass caps for the power supply pins of the op-amps? If no, please solder them and see the noise level.

Why you are not using a dual power supply? One half of the signal will be gone.

 

[Yarrrrr]

There is no substitute for a systematic analysis.

I figured systematically replacing one thing at a time and do a measurement after every single change would be sufficient to narrow the issue down.

Do you have bypass caps for the power supply pins of the op-amps? If no, please solder them and see the noise level.

Would be good practice for sure, but not something I can reasonably try without having an entirely new PCB manufactured. And this assumes I am not already at the theoretical noise floor I am supposed to be. Perhaps Audioguru is wrong.

If the answer is yes, the noise comes from the PD. And that needs attention.

What more could I possibly do to give that attention, I have tried 4 different Photodiode brands and one Phototransistor, in all of the two possible biasing configurations. They all give me a SNR difference around ~1 dB.

Why you are not using a dual power supply? One half of the signal will be gone.

Circuit is supposed to work from a single 9 Volt battery and I have had no reason to believe this design choice is a reason for the bad SNR.

Begin at the beginning. Start with the photodiode signal. If the photodiode is covered with a black tape, do you get the low noise?

It is getting kind of old to begin at the beginning 10 different times because 10 different people are not reading through the responses of the thread.

I appreciate that people are willing to help.
But this thread has been:

• A majority of people not reading more than a singular comment, repeating the same thing over and over.
• Saying something completely off topic and irrelevant to the actual question of SNR, leading to less visibility of the actual topic at hand.
• Contradicting each-other and even themselves.

So far the only actual advice on the circuit that led somewhere was very early and had to do with my high-pass filter and input resistors drawing too much current.
The suggestion about reading how optical vinyl players work is interesting, but I can't find any usable information on it.

Unless someone has an eureka moment I suggest we save everyone some time and just end the thread here.

 

[Audioguru]

The IR beams might be too wide. The strings scatter the IR light that is shined on them. Therefore perhaps the IR san be focused on the strings.
Reducing the distance between the IR LEDs and the strings and between the photodiodes and the strings will increase signal strength.

 

[c_mitra]

I figured systematically replacing one thing at a time and do a measurement after every single change would be sufficient to narrow the issue down.

Ok, now you have seen that the origin of the noise is not from individual components. Each component is working as it is expected.

What more could I possibly do to give that attention, I have tried 4 different Photodiode brands and one Phototransistor, in all of the two possible biasing configurations. They all give me a SNR difference around ~1 dB.

This suggests that the component PD is not the source of the noise. There are only two possibilities: how it is connected and/or the noise is coming from elsewhere

Circuit is supposed to work from a single 9 Volt battery and I have had no reason to believe this design choice is a reason for the bad SNR.

Most of the modern op-amps can work at low voltage; you can easily use two electrolytic caps and two 100K resistors to get a -4.5V-0-4.5V supply. But we can wait for that for the last step.

It is getting kind of old to begin at the beginning 10 different times because 10 different people are not reading through the responses of the thread.

It is part of the learning process. For example, you have given the graphs on the noise spectrum but the description is very sketchy. The graph is useless unless I can figure out under what condition it was done.

• A majority of people not reading more than a singular comment, repeating the same thing over and over.
• Saying something completely off topic and irrelevant to the actual question of SNR, leading to less visibility of the actual topic at hand.
• Contradicting each-other and even themselves.

It happens; I have seen that many times even with recognized experts.

Giving up is not good. You have invested time and effort and you deserve some satisfactory closure.

You need not make any new PCB.

Just try these two steps:

1. Solder 100nF ceramic bypass caps to the power supply pins of the op-amps. All four op-amps must be bypassed.

Forget about the signal; just measure the noise floor level. Both before and after. Has the noise come down?

2. Remove the cable connecting to the photodiode (unsolder it from the PCB) and measure the noise floor. Both before and after.

Giving up is easy. Hanging on is tough. I have given up so many times.