High frequency noise problem

[jrz126]

Hi all, I'm working on this simulator at work and I'm having a problem.
I'm trying to measure the 'on' time of this 100Hz PWM signal, but I kept getting a measurment of 2uS (the smallest pulse width that the simulato card can read). I hooked up an O-scope and I'm getting a bunch of high freq. noise. So I think the noise is causing my 2uS reading.

The card that is putting out this PWM signal basically just switches a mostfet with this pwm signal. (so it basically acts as a switch). I know that I am getting a good PW signal because I am able to read it just fine using a handheld fluke.

So whats the best way to filter this?

oh, and this is kinda what the scope looked like. only Im using a 0-5V square wave
**broken link removed**

 

[RadioRon]

In trying to help, I ran into a problem understanding your setup and I can't quite understand what you are seeing. Could you explain it a bit better. Here's the way I understand it:

you are generating a PWM signal from a microprocessor that is intended to convey a 100Hz analog signal. You are trying to measure the pulse width of the PWM output and are puzzled why you are getting 2 uSec.

OK, so first of all let's review the basics. PWM outputs are usually used as D to A converters. You generate a digital pulse stream at a very high frequency (that is, much higher than the analog signal you want to generate) and you modulate the width of each pulse. As you make the pulse longer than 50% then the average voltage goes up, and then when you make the pulse shorter than 50% your average voltage goes down. So, you vary the average voltage by varying the pulse width. In order to actually see the average voltage as an AC low frequency signal, you must do the averaging with an RC filter. This is always done off-chip by adding the necessary series resistor and shunt capacitor to filter the signal. The values for resistor and capacitor must have a cutoff frequency just above the analog range you want and well below the frequency of your pulse stream. The series resistor is usually a fairly high value to keep the current drain from your digital output reasonably low. The shunt capacitor is calculated to suit the cutoff frequency.

OK, so much for the basics. In your case, I wonder if you aren't simply measuring the width of the PWM pulse stream before the averaging filter. 2 uSec sounds like a perfectly reasonable value for a pulse width from a PWM output that is running at 250KHz, not an unusual value at all.

If I'm completely off base and this isn't it at all, then let's consider that maybe you do indeed have outside HF interference, perhaps from an AM broadcast station. In this case, you need to add some low pass filtering that rejects the HF interference but passes your signal. How you do this depends on your circuit topology (the configuration of how your components are wired up) so we need you to explain that a bit better, please. Perhaps you can show us a simple schematic diagram of your circuitry?

 

[jrz126]

Well I got my hands on a scope today, after probing around for a bit, it turns out the noise is caused by the power supply...For some reason, I thought about sticking a cap across the terminals, and it cleared up the noise.
What is the best value to use? I have a .02microfarad (thats all that was available at the time) across it right now, and it seems to work pretty good.

 

[JimB]

What is the best value to use? I have a .02microfarad (thats all that was available at the time) across it right now, and it seems to work pretty good.

If it is all thats available and it works, then that is the best cap!

JimB