Noise problem

[rnielsen]

Greetings!

I have a small problem and would like to pick your brains for a solution.

I have a PIC (18F452) that is controlling a Servo Drive. The PIC is also monitoring and controlling other devices. I am using the ADC to monitor pin RA1. There is no switching inputs so I'm not worried about settling times. The problem is that the Servo Drive is injecting a fair amount of noise into the circuit, causing the analog readings to bounce up and down quite a bit. I am using a single analog power supply for the whole thing. 24V going to the Drive and sensors and the PIC circuit using a DC/DC convertor. I need to get rid of the noise that is affecting the analog readings. I've contemplated using a seperate supply for the PIC/sensor circuits but I'm not sure if the Drive unit would still inject noise through the common grounds or chassis.

Does anybody have any experience/knowledge regarding problems like this?

Ronald

[PCM programmer]

Can you see the noise with an oscilloscope ?
What's the frequency, amplitude, and waveform shape of the noise ?
My first thought would be to put an RC filter on it. But the 18F452
data sheet says the maximum source impedance for the A/D pins
is only 2.5K ohms. So, that limits the RC filter quite a bit.
So it would be better to buffer the A/D signal with an single-supply
opamp, and then you could put a cap in parallel with the feedback
resistor and set the rolloff frequency to whatever you want.

Another thought is: Are you using the proper A/D clock for your
PIC crystal frequency ? The A/D clock divisor value can affect the
accuracy of the A/D and could be causing your symptoms if it's
incorrect.

[cadegga]

Hi,

we had similar problems. Some considerations:

1) use shielded cables, try to shield the motor circuit as well
2) separate your analog inputs as far as possible from all other parts

and, most important,

3) trigger on the falling edge of the PWM you are using for motor control (e.g. use PWM output as digital input on one of the available external interrupt pins, i.e. the B port) before you start the A/D-conversion.


Best regards!

[kender]

Ultimately (and obviously) you need to isolate your servo motor from the PIC with optos. You can still bring analog and digital signals from the isolated area to your PIC.

Nick

[rnielsen]

I got it fixed. I normally don't do much analog stuff so I forgot some of the basics. I installed a second power supply, one for the servo driver and one for the PIC circuit. The servo driver already has it's I/O isolated from it's power/driver circuit. I added a small capacitor across the resistor that I'm measuring (my analog device puts out a 4-20ma and I have a 250ohm resistor in series with it which gives me a 1-5V signal that I can measure) and then I added a nice ground plane over most of the board. The signal is now very stable. Guess I need to get out my Basics 101 book once in a while.

Ronald