I have 4 x MCP9700 temperature sensors connected A2 to A5 on an Uno. I also have 6 LEDs pretending to be relays on digital channels 3 to 8. When I run the code below the result from the sensors goes up in proportion to the number of LEDs switched on. Even just switching one on effects the results.
void setup() {
// put your setup code here, to run once:
Serial.begin(9600);
delay(500);
analogReference(INTERNAL);
Serial.println("Hello");
for (int i = 0; i < 6; i++) pinMode(i + 3, OUTPUT);
}
void loop() {
// put your main code here, to run repeatedly:
int input, volts;
float reading;
I have tried using the 1.1V internal ref but the problem is not the digitisation. Even small changes in the voltage caused by turning even a single relay on seems to have a significant effect on the sensor output.
In temperature control it's common to use hysteresis... i.e. If you have a heater with a target of 70 degrees you turn it on at 69 (or 69.9) degrees and off at 71 (or 70.1) degrees. Or any time you have a relay you might want to do that, or add delay so the relay doesn't "chatter" on & off several times per second.
Good, but we meant you'd edit your first post and enclose that code, not making another post while keeping the first bad/unformatted code.
So far, please, to make it cleaner I think you better edit post #1 and remove the whole code from there, we'll get it from post #3.
It does not matter much, but I prefer the code with code tags in a new reply. That way the topic retains its chronology and comments about the original post make sense
I have tried to a separate lab type power supply and it has no effect. Its currently running off the 9V jack. I have also tried driving the seniors with higher voltages (up to 9V from the lab supply) but it make no difference.
Its not a ripple or noise effect. I get a stepped response. I have turn on up to 6 LEDs, and the reported temperature from the sensors (all of them) will go up in (roughly) even steps for each on I turn on, and then come down again as they are turned off. It is repeatable.
This is actually a test program to separate out the issue with minimum code, which it does. The main program is much larger and does have hysteresis is also range limiting etc.
This is the wiring as it stands. Yes, it does have resisters on the LEDs. I agree there must be a some kind of cross feed. I have seen some non-zero voltages on the 0V lines which does not make any sense at all to me.
I would use a star ground configuration...
Your rather large ground current from the leds, combined with some small resistance in the ground line may cause a non zero ground voltage at the sensors. So give the leds a separate gnd wire and see if that helps.
Hi, thanks for this. I changed to a star format for the ground and made use of all three GRD points on the UNO board. This does make a very big difference. So much so that I can now convert to temperatures and add in the other sensors. I also made use of the "double read" trick that I read about in a different feed which also seems to help.
So some sensors are more effected than others. If I move the ground connections around the most effected sensors also changes.
The new test skit now looks like the below, but I am pretty sure I still have work to do on the electrical side. Do I need to add some caps? If so were do they go?
void setup() {
// put your setup code here, to run once:
Serial.begin(9600);
delay(500);
analogReference(INTERNAL);
Serial.println("Hello");
for (int i = 0; i < 6; i++) pinMode(i + 3, OUTPUT);
}
float readit(int ch){
int input;
float reading;
input = analogRead(ch); //dummy read
input = analogRead(ch); //actual read
input *=(1100/1024); //convert to mV
input -= 450; //correct for zero offset
reading = input/10.0;
return reading;
}
void loop() {
// put your main code here, to run repeatedly:
Serial.print("0, ");
Serial.print(readit(A2));
Serial.print(", ");
Serial.print(readit(A3));
Serial.print(", ");
Serial.print(readit(A4));
Serial.print(", ");
Serial.println(readit(A5));
for (int i = 0; i < 6; i++) digitalWrite(i + 3, true);
delay(2000);
Serial.print("1, ");
Serial.print(readit(A2));
Serial.print(", ");
Serial.print(readit(A3));
Serial.print(", ");
Serial.print(readit(A4));
Serial.print(", ");
Serial.println(readit(A5));
for (int i = 0; i < 6; i++) digitalWrite(i + 3, false);
delay(2000);
}
Thanks for this idea. I did have a look at it but it just inverts the problem. I am reading too high when the LEDs are off rather than on and since both states are controlled by the sensors its not the answer.
What unit is the timescale of your graph?
A 100nF capacitor from 5V to ground near each sensor might help.
Also: keep all wires to the srnsors (5V GND and signal) as short as possible and close together.
Are you in an industrial environment?
Please make a picture of your setup. The drops in value 5 are pretty big. The wobble in the others might just be the size of 1 step of the analog input (you can do some calcs. To check that ir you can print the raw values instead of the calculated values...
You are still getting two distinct voltages that depend on whether the LEDs are on or off, on three out of the four traces.
The 'value 2' trace looks good. Did you do something different on that channel?
That might be a clue as to what you need to do for the other three channels.
I'd connect the 4 sensors to one Arduino GND pin, and then connect the switches and LEDs to a differant arduino pin.
