Noise Removal

michaelb · Joined: 28 Nov 2005 Posts: 17

Hi,

I'm building something along the lines of four fan controllers on an i2c net, with smoke and temperature sensors on the ADC pins, and MOSFETs being driven by LP filtered PWM outputs from a 16F877A.

I have 0.1uF and 10uF capacitors across the breadboard power rails on each module - one module being one controller + 2 fans + sensors on 1 breadboard.

I'm having trouble with noise I think, as the temperature sensor work accurately until fans are switched on where the sensor reading increase by a degree or two.

With a scope I've noticed harmonics at 420Hz, 840 etc, with the one at 420 being -36dB with NO fans.

WITH fans I get the first harmonic in the range 340-540Hz depending on fan RPM. This is about -42dB. With both fans on, this actually gets reduced. I'm assuming it's destructive interference.

The ADC pin is negligibly noisy, with only a small increase with fans on, but no clear harmonics.

As the ADC pin is not noisy, I'm assuming the error is due to the noisy power lines giving a poor reference.

I'm having a hard time understanding filtering - how the .1 and 10 uF caps I work I don't know. I've read a lot, incl. The Art of Electronics, but I can't find a simple way to remove this noise. Could I simply add another cap across the power rail to remove this? I'd appreciate any help.

Thanks,

Michael

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

The first solution that comes to mind is to put the fans on a different
voltage regulator than the PIC.

Another solution would be to configure the A/D so that it uses an
external reference voltage, and make sure that this reference is
from a clean power source.

Also, the board layout will be important. Don't put the PIC in the
same ground path (i.e., a trace) as the fan. That way, ground
current from the fan won't flow along the same trace that's connected
to the PIC's ground pin. The fan should have it's own separate
ground connection which goes back to the (low impedance) power
supply's ground pin.

Finally, how much of this noise is caused by coupling into the ground
wire of your oscilloscope probe ? Use a very short ground wire on
your probe, and put it directly on the PIC's ground pin when you're
looking at noise on the PIC's Vdd pin.

michaelb · Joined: 28 Nov 2005 Posts: 17

Thanks very much for your reply.

I tried connecting the fans to a different output on the power supply. However, due to the mosfet configuration this meant the fans were always on. I assume this is something to with the mosfet control being the GS voltage, and with no common ground, it isn't pulled down to 0.

I'll try out your other suggestions tomorrow.

Thanks again.

Ttelmah · Guest

The 'key' here, is that you need a ground between the control (PIC), and the fan supply, but the only current that needs to flow in this ground, is the power involved in the _control_, not the actual power of the fans. Ideally, have the grounds all run from a single common connection, and keep the ones with high currents involved, separate from the low current areas.

Best Wishes

michaelb · Joined: 28 Nov 2005 Posts: 17

I have now connected the fan grounds directly to the power supply rather than running it through the breadboard. This has worked perfectly - the fan induces no noise on the power lines at all now.

The harmonics at 420Hz, 840 etc still exist - I'm not sure what they are caused by, any ideas? - but these do not seem to matter for the ADC reading.

Thanks very much for your help.