noise on 0-5v signal line - need filter

Thread Starter

elRey

Joined Feb 23, 2009
65
Hello, I'm trying to use a 0-5v signal from a car's sensor into a comparator. However, there's noise on the signal feed. I've acttached pictures of the noise in question. Multi-meter reads 1.89v. I need ~1.89v into the comparator (which varies as sensor works). The sensor is a Manifold Absolute Pressure sensor. The measurements below are at atm.


The issue I'm having is as the signal gets close to the Ref V on the comparator, is falsely triggers the comparator on/off a wide range around ref V.

I quickly tried a 0.1uF cap on signal line to gnd, and then a resistor in series with signal. No change.

Main Time/Div = .2
Vol / Div = .2

Thanks,
Rey

Noise:

 

SgtWookie

Joined Jul 17, 2007
22,230
Give this a try. See the attached; I just threw it together without spending time calculating things.

The input is just a 2.5kHz 400mV square wave.
R3 & C3 are just a simple 1-stage RC filter. Note the penalty paid for the simplicity in the cyan trace below; slow to respond and more noise.

The yellow trace shows how much noise has been removed by the first stage of a 2-stage filter. The noise on C2 is very low in comparison to the other waveforms, and the response time is much better than the single stage filter.

You can decrease the noise on C2 by increasing R1 and R2, but you will get slower response time.

Or, you could simply add another stage.

Noise in automotive environments is brutal.
 

Attachments

THE_RB

Joined Feb 11, 2008
5,438
You might want to really increase the time constant of the filter. Manifold pressure will pulsate badly as the intake valves open/close and it requires quite a LOT of filtering to get an average pressure anyway.

On a 4 cyl engine there will be 2 intake valve openings per rev, so at idle (say 1500 RPM) there will be bad pulsing at 1500 *2 valves /60 RPM = 50Hz. You probably want to set that filter pole at 5Hz or even slower and/or multistage filter.
 

Thread Starter

elRey

Joined Feb 23, 2009
65
the low-pass filter worked great. I ended up using 3.7K for both resistors. I could have gone smaller.
 
Top