rcheshire
12-12-2013, 07:51 AM
In an attempt to better understand the problem of pulse width modulation and mosfet switching and the frustrating lines on images, produced by switching events, I started from scratch.
Other than the huge voltage spikes generated by mosfets, pwm switching is the main source of lines in images. I found this to be the case with any form of hard switching on the logic side and opted for a soft solution.
Using a SPICE based simulator, I started with the mosfet, a logic level device that should run comfortably in the palm of your hand. Itracked down some typical N-channel enhancement characteristics, starting with Gate and Drain turn on/turn off.
The drain showed a nasty 300v spike at turn off - a pair of diodes fixed that with a 5k resistor to flatten out residual spikes along with a 10nH inductor. That pretty much took care of the power side of things.
On the logic side, any form of switching is noisy. In the attached circuit, PWM from the Arduino compatible microprocessor soft switches supply from the 12v side to the gate, rather than switching the mosfet gate directly - more or less simulating switching through a transistor.
The mosfet runs cool, and as far as I can see, inteference lines (in the dark frames I take to test noise levels) are imperceptable. I can't pick the difference comparing the fixed pattern noise of the sensor.
The mosfet gate voltage does not go to 0 with the Zener/diode in place, which probably accounts for a large part of the noise reduction - but it's cool to touch, which is a surprise. Temperature control is steady, with indications flicking up and down, as is usual, either side of setpoint.
Time for that mosfet driver and a new design mosfet with plans to go smd.
Other than the huge voltage spikes generated by mosfets, pwm switching is the main source of lines in images. I found this to be the case with any form of hard switching on the logic side and opted for a soft solution.
Using a SPICE based simulator, I started with the mosfet, a logic level device that should run comfortably in the palm of your hand. Itracked down some typical N-channel enhancement characteristics, starting with Gate and Drain turn on/turn off.
The drain showed a nasty 300v spike at turn off - a pair of diodes fixed that with a 5k resistor to flatten out residual spikes along with a 10nH inductor. That pretty much took care of the power side of things.
On the logic side, any form of switching is noisy. In the attached circuit, PWM from the Arduino compatible microprocessor soft switches supply from the 12v side to the gate, rather than switching the mosfet gate directly - more or less simulating switching through a transistor.
The mosfet runs cool, and as far as I can see, inteference lines (in the dark frames I take to test noise levels) are imperceptable. I can't pick the difference comparing the fixed pattern noise of the sensor.
The mosfet gate voltage does not go to 0 with the Zener/diode in place, which probably accounts for a large part of the noise reduction - but it's cool to touch, which is a surprise. Temperature control is steady, with indications flicking up and down, as is usual, either side of setpoint.
Time for that mosfet driver and a new design mosfet with plans to go smd.