Noise in photodiode and optimum size.

Hi:
I am currently working on an application – that includes a photodiode and a TIA. I have a requirement from the resolution of this device, that the signal should minimum be 30db – and I don’t have much power to do good with.
So my question is – what kind of noises should I consider from the photodiode (beside shotnoise, themal noise & internal noise (TIA)).
Also, how can I choose my photodiode and emitter, so I can collect as much light as possible (angles, size etc), while still have a small noisefloor.
Thank you in advance.

Hi,

To be true.. I don't understand a word.
There is only one value: 30dB...it is a relative value..it is equal to a factor of 31.62.
Do you mean SNR = 30dB?
But what to do with this value. With no other value like voltage, current, frequency, bandwith, light level or what ever.

And what is the photodiode used for? Detecting night and day? Data transfer? Object detection?

Klaus

Sorry for not being precise
Yes it is 30dB SNR (to achieve a certain resolution)… It for a pulse oximeter.
This is my specs: Bandwidth 10Hz, samplerate 100Hz – I am modulation the signal with a square wave with a certain duty cycle (as small as possible) and a switching frequency. The drive voltage is 3.3V and the current should be as small as possible – I want a power consumption around the microwatt area.

Well obviously SNR is improved by S(ignal) which is not
stated, the optical power and its delivery and return.
This, more than circuit design, is probably the big deal.

One form of noise that you do not mention is RTS / RTN.
More a concern at very low light levels / photon counting.
Can be suppressed by chopping / pumping, but tough for
staring applications.

If Tx and Rx can be codesigned then you can take
advantage of autozero, chopping, multi-sample
techniques. Sampling multiple times and taking
the difference between "off" and "on" light phases
will let you reject ambient light, LF EMI and so on.

If you want microwatt power consumption then
I guess the optical source is Somebody Else's
Problem. That could be limiting, as far as your
options. Or maybe you are hyperoptimizing a
point feature that is not so important in the
larger picture.

Okay thank you:
Let’s say that I want an average power-consumption of 10uW at the LED. I have a 3.3V, which means that the current should be 303uA, this gives 1mW (peak) and I am modulating this with square at duty cycle of 0.01 (1%) which gives the average power of 10uW.
I expect that the signal will get attenuate 1000 times, before it hits the photodiode – this will give 10nW.
My question is therefore – is it possible to work with this number – or will it be drowned in noise.
Furthermore, what does the photodiode see, is it the peak power or average power?

Hi,

I expect that the signal will get attenuate 1000 times

Click to expand...

Why that much? Use a lense.

what does the photodiode see, is it the peak power or average power?

Click to expand...

"Average" is a mathematical function.
Both, "peak" and "average" is true.

Klaus