Photodiode Monitor Current Range should be Selectable #45
Loading…
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Hardware Revision: Rev0_3
Description:
Laser diode's integrated photodiode (PD) monitor (Mon) can have very different responsivity. STM32F407 ADC's resolution is 12bit (805.6640nA/LSB in 0-3.3mA Full range). This current resolution is not enough for laser diode with a small PD Mon full range.
Therefore, PD Mon Circuit's gain should be configurable. This can be achieved with Pin/Digitally Programmable Gain Amplifier (such as TI's PGA308) or simply some selectable gain resistor.
Solution
Instead of using a PGA. Using multiple TIA Stage of different feedback resistor can achieve the goal to save cost.
This circuit structure is obtained from the Art of Electronics Book.
Note: Required Cutoff frequency = PCLK2 (84MHz) / 8 / 480 Cycles / 2 < 10.9375kHz
A prototype is built with the noise floor and functionalities verified. The TIA noise floor is probably overkill for the system but it may also be useful for other systems later. The bottleneck is on the bit resolution of ADC to be used. The noise floor can support 16 bit ADC @3.3V. During implementation, we can either connect the 3 output range to AD7172-2 with maximum sampling set or STM32 ADC.
Other Measurement Taken from the Prototype
Given the low noise floor. It will be interesting to also measure the limit of the TIA circuit. Trying to measure the measured shot noise from an incandescent lamp would be interesting. Suppose Shot Noise has the following characteristics.
Shot Noise ∝ SQRT(Signal Power)
So, plotting the signal power(PD Voltage) vs Noise should see a exponential curve.
The circuit is connected to be in following configuration. The measurement configuration is very similar to #17
PD → TIA(1.32kohm 1%)(Bandwidth 234.1055kHz) → 1st order HPF (11.2876kHz -3dB) -> 2000x Gain Block → 1st order LPF(56.4379kHz -3dB) -> HPF (6.7726 Hz -3dB) → Signal Analyzer
Note: The LPF is added to avoid saturating the SA DC Coupling Input.
Signal Analyzer Setup:
Input: DC Coupling 50Ohm Impedance(Required)
Sweep Point: 40001 pts(Maximum)
Frequency Range 11kHz-56kHz
BW: 1.0 Hz (Minimum)
AVG Type: RMS
AVG Pts: 30
Atten: 6dB(Minimum)
Fitted Line: Quadratic fitting on the data points
Theoretical Model: y = Shot Noise @ Voltage Value + Measured Noise Floor
Normalized Theoretical Model: Multiply a constant to the theoretical model so that its x^2 coefficient matches the one in the fitted line
Fitted Line: y = 0.006051 x^2 - 0.01389 x - 0.5385
Theoretical Model: y = 0.003265 x^2 - 5.05e-16 x - 0.3907
Normalized Theoretical Model: y = 0.006051 x^2 - 9.36e-16 x - 0.7241
You can see the fitted line curve shape matches the shape of the normalized theoretical model.
When does Shot Noise dominate?
Here is the formula for the theoretical model above: Shot Noise @ x Voltage(A/rt(Hz)) =sqrt(2e * x / Rf)
Theoretical Noise @ x Voltage = sqrt((Shot Noise * Rf)^2 + Noise Floor^2)
Also, noise is added up with root sum squared operation.
Thus, we can know that theoretically, shot noise should dominate at the following optical power in different range.
In 0-2.5mA range, shot noise dominates @0.728V -> 552.07uA
In 0-750uA range, shot noise dominates @0.326V -> 73.9948769647867uA
In 0-250uA range, shot noise dominates @0.212V -> 16.127uA
I remember this circuit :)
However, its DC response will vary from device to device due to different Vth of JFETs; it will also vary with temperature.
Maybe use a switched resistor approach. Low-cost reed relay would do the job.