I have two amplification stages, separated by a high pass filter. The system works fine. The first amplifier is a transimpedance amplifier converting current from a photodiode to a voltage.
The photodiode creates current in response to a light source emitted by an LED (not shown) and reflected by the skin.
There is a low pass filter on both amplifiers to limit response to 700Hz and remove some noise.
The problem is when I try to duty cycle the light source (on time 1ms, off time 3ms). Because the output of amplification stage 1 oscillates, amplifier 2 rails. To overcome this I have used a much faster (about 20khz) PWM signal and tried to attach it to points A and B shown in the diagram below.
During LED off time, in order to maintain the voltage level seen by the second amplifier on its + pin, and avoid railing, a PWM signal is output to point A or B.
This works perfectly when the DC signal level is below VDD/2, but it does not work for above VDD/2 levels. I cannot understand why. It seems that no PWM value can trick amplifier 2 into seeing a somewhat constant voltage at the input, but only when the DC output of amplifier 1 is above VDD/2.
http://i62.tinypic.com/vouej9.jpg
Here are scope readings from 3 points
Left: point A
Middle: point B
Right: amplifier 1 output (where it says 'oscilloscope probe tip')
http://i60.tinypic.com/21kxrg3.jpg
I have tried many things including adding another resistor downstream of C2 and sending PWM to the point after the resistor, and adding an RC low pass filter to the PWM signal.
I have no idea where the problem is coming from but one hint is the following. Because there is some capacitance on the circuit, when the PWM duty cycle is low, the resulting wave looks smoothed out and it does not reach full VDD voltage. Perhaps then the problem is connected with higher duty cycles actually reaching the 3.3V supply voltage, and pushing down the DC signal; but I really can't see how.
Please note that I am well aware that normally in pulse oximeters the DC offset is removed with a DAC; here I have no DAC available, nor space for more amplifiers or layers of filtering. The system works with VDD/2 and I need to hack it to work for at least 4/5ths of the range.
thanks
The photodiode creates current in response to a light source emitted by an LED (not shown) and reflected by the skin.
There is a low pass filter on both amplifiers to limit response to 700Hz and remove some noise.
The problem is when I try to duty cycle the light source (on time 1ms, off time 3ms). Because the output of amplification stage 1 oscillates, amplifier 2 rails. To overcome this I have used a much faster (about 20khz) PWM signal and tried to attach it to points A and B shown in the diagram below.
During LED off time, in order to maintain the voltage level seen by the second amplifier on its + pin, and avoid railing, a PWM signal is output to point A or B.
This works perfectly when the DC signal level is below VDD/2, but it does not work for above VDD/2 levels. I cannot understand why. It seems that no PWM value can trick amplifier 2 into seeing a somewhat constant voltage at the input, but only when the DC output of amplifier 1 is above VDD/2.
http://i62.tinypic.com/vouej9.jpg
Here are scope readings from 3 points
Left: point A
Middle: point B
Right: amplifier 1 output (where it says 'oscilloscope probe tip')
http://i60.tinypic.com/21kxrg3.jpg
I have tried many things including adding another resistor downstream of C2 and sending PWM to the point after the resistor, and adding an RC low pass filter to the PWM signal.
I have no idea where the problem is coming from but one hint is the following. Because there is some capacitance on the circuit, when the PWM duty cycle is low, the resulting wave looks smoothed out and it does not reach full VDD voltage. Perhaps then the problem is connected with higher duty cycles actually reaching the 3.3V supply voltage, and pushing down the DC signal; but I really can't see how.
Please note that I am well aware that normally in pulse oximeters the DC offset is removed with a DAC; here I have no DAC available, nor space for more amplifiers or layers of filtering. The system works with VDD/2 and I need to hack it to work for at least 4/5ths of the range.
thanks
