Do I need a precise voltage reference for my ADC if the what I am measuring shares the ADC voltage source?

Question

I'm trying to design a basic measurement system using some phototransistors and ADCs to measure light absorption via a voltage drop. I've attached a schematic for a quick and dirty representation of what I mean. I couldn't find a phototransistor in the parts to I placed a photodiode with a transistor, but the point is the same.

^{simulate this circuit – Schematic created using CircuitLab}

My question is about whether I should connect the voltage reference pin on the ADC to Vin. I'm using a 12-bit ADC, and I know that to make good use of that level of precision, I need to carefully select my voltage reference to also be extremely precise. But here's what I'm thinking: My power supply V1 is going to be muddy. If I manage to get a reference 5V that has 0.01% precision, small but significant (relative the 0.025% theoretical precision of a 12-bit ADC) fluctuations in V1 will translate to fluctuating Q1 collector voltage even if everything else stays the same. Therefore, I will still have pretty low precision in determining the actual collector voltage since the reference will be a "perfect" 5V. But what if instead I connect my ADC Vref to V1 directly? I'm thinking that since reference will now fluctuate with V1, V1 fluctuations will not propagate into the measurement since the fluctuations are cancelled-out when comparing the reference to the AD pin voltage.

Is this a valid approach to preserving precision, or am I missing something important in how an ADC fundamentally works? I did a little bit of searching around for discussions on what the best course of action is when the ADC and the subject being measured share a power source, but I couldn't find anything. And I'm assuming that's because it's so plainly obvious no one talks about it, or it's so plainly wrong that no one talks about it.

Would love some feedback and pointers towards resources!

I'm thinking of using the ADS7828 and all 8 channels going to different phototransistors, if that's relevant.

Answer 1

photo diodes mosty pass current proportional to their illumination, there is some dependance on bias voltage, but illumination dominates.

so yeah, you're converting photocurrent into current in R1 ,and thus, into a voltage. so for most accurate results you need to use a stable voltage reference.

But wait there's more!

You are feeding the photocurrent into a transistor base, so the photocurrent is multiplied by the transitors gain, which depends on the temperature, varies between devices, and also drifts with age. you may want to use a more predictable gain circuit here, or possibly provide the user with some calibration method.