balancing sensors

Software.

I tried that , I substracted a value (in decimal) to the highest input port to make them equal. but when the values get low lighting, the higher of the two gives me 65555 on myLCD screen

two

The sensor's individual response curves may not differ by only a constant value. You may have to calibrate them over their operating range and do a suitable curve fit to have them match over that range.

HI well thats what I was trying to do , calibrate that chip using resistance on the output or in the micro chip. I just thought that if I decrease the input voltage of the sensor that gives the highest output, I might be able to get a conveniant output. I also have to add a low pass filter, there are a lot of noise and the output on each fluctuates about 0.2volt

Greg Neill wrote:

Lets get back to the origin of the signals, to figure out what correction math might bring them into similar calibration (if a simple offset isn't appropriate).

What kind of sensors are these?

IF the problem is from different sensitivity, you could try putting a voltage divider on the sensor output that gives the larger signal.

The resistance of the divider should be:

1. Large enough so that it does not alter the sensor output appreciably, and
2. Small enough so that the microchip does not alter it's output appreciably.

Mark

Yes, how about putting the voltage divider at the input supply. ?

Here is the chip I am using:

Also I saw about 3%-4% of ripple (noise). Maybe just adding a small capacitor with a resistance in parallel (from Vout to ground)? or would a feedback op-amp low pass filter be a better idea?

The output should be fairly independent of the exact supply voltage, so this is probably not a very good method.

Are you sure your light source is perfectly stable (no ripple)?

An RC could be used as a low pass filter. A voltage divider that has a small range (100% to 80% of input signal) followed by an RC might also work. This device should have a pretty good zero output for zero signal, so variable scaling (zero in equals zero out, regardless of the scale factor) is probably a lot better than variable offset.

lerameur wrote:

It really needs to go on the output voltage.

Nice. That 8 Meg resistor makes it pretty sensitive.

A simple way to both have a low-pass filter and make adjustments to the DC signal would be to put this on the output of your sensor:

R1 . o--,/\\/\\/'---+----+---o . | | . \\ | . / | . R2 \\ --- C . / --- . \\ | . / | . | | . o------------+----+---o . | . ----- . --- . -

The filter's time constant is Rp*C, where Rp is the parallel combination of R1 & R2. R1 & R2 form a simple voltage divider for the DC level of the signal. R2 could be a trimpot; adjust it on one sensor to match the response of the other.

A low-pass filtering op-amp is better, if you can hook one up. Then you could use a trimpot to adjust the DC gains of the two sensors to match each other, in addition to having filter caps.

Question: is your ripple: ... at line frequency? ... twice the line frequency? ... some other frequency? ... just randomly fluctuating, not at any particular frequency?

Mark

I added a capacitor and a resistor in parallel at the output like you said and it solved the problem. I already had a a resistor their. The max current these chip can take is about 20ma for input anyway. I also has some back ac lighting,weak but still affected the circuit, but when in total darkness and just the white led, it works good. thanks

ken

You're welcome ... and good luck.

Mark