Needing help determining why this circuit did not work

Thread Starter

gte

Joined Sep 18, 2009
357
Ok, I'm a bit confused. I thought a 15k resistor to the U3 non inverting input would eliminate the cross talk?

I need to eliminate the crosstalk because I need to measure the analog voltage going in to the U3 inverting input to bench test it accurately. Right now I get large voltage fluctuations and I am not able to measure the voltage going in because of that on my multimeter. I was able to watch the ~2.25hz square wave change its pulse width by varying the output from the potentiometer to the U3 inverting input, but I had no idea (per my multimeter) how much voltage I was actually giving the U3 inverting input, as it oscillated all over the place. I need to get rid of that crosstalk oscillation so that I can measure that accurately. If this resistor won't do it, can I isolate it somehow or how can I eliminate the crosstalk entirely? The lm324 did not seem to suffer from this, but it was not rail to rail like this IC is and I need rail to rail.

Thank you for helping.
 

crutschow

Joined Mar 14, 2008
34,285
Ok, I'm a bit confused. I thought a 15k resistor to the U3 non inverting input would eliminate the cross talk?

I need to eliminate the crosstalk because I need to measure the analog voltage going in to the U3 inverting input to bench test it accurately. Right now I get large voltage fluctuations and I am not able to measure the voltage going in because of that on my multimeter. I was able to watch the ~2.25hz square wave change its pulse width by varying the output from the potentiometer to the U3 inverting input, but I had no idea (per my multimeter) how much voltage I was actually giving the U3 inverting input, as it oscillated all over the place. I need to get rid of that crosstalk oscillation so that I can measure that accurately. If this resistor won't do it, can I isolate it somehow or how can I eliminate the crosstalk entirely? The lm324 did not seem to suffer from this, but it was not rail to rail like this IC is and I need rail to rail.
No, the resistor just reduces the current through the input protection diodes so you will still see the cross-talk voltage.

To totally eliminate it, two choices come to mind:

1. Use a rail-rail op amp that doesn't have protection diodes across the inputs.

2. Add a large capacitor (100uF or more) to ground between the pot output and a 20k ohm input resistor to the op amp input, to filter out the fluctuations and giving smooth DC that the meter can measure across the capacitor. Also add a 20k ohm resistor in series with the other op amp input.
 

Thread Starter

gte

Joined Sep 18, 2009
357
Ok, I put the RC network in and the extra resistor to the inverting input for U3. I can measure the analog voltage now, thank you for that.

It was hitting 50% duty cycle at 5 or 6 volts. Then I change R1 for a 22k resistor and it is now hitting 50% duty cycle at about 2.5v, I need it to hit 50% duty cycle at .55v to match the way the adapter box I am replacing worked and for the display to read properly. I tried 27k for R1 and it flatlined, so I'm at a limit there with the voltage divider. Right now it is .65v read before the RC network (out of the pot) and 1.25v after the RC network from the cross over interference I guess.

Can I modify the components to achieve what I need or do I need some sort of amplifier to amplify the temperature sensor voltage to make it compatible with this circuit? Or possibly bias it by 2 additional volts?

Here are the original readings:

33c or 1.71v = 91% on
40c or 1.36v = 84% on
50c or 1.08v = 75% on
55c or 0.96v = 71% on
60c or 0.85v = 67% on
65c or 0.77v = 64% on
70c or 0.66v = 59% on
75c or 0.64v = 54% on
80c or 0.52v = 49% on
85c or 0.46v = 41% on
90c or 0.43v = 39% on
95c or 0.39v = 37% on
 
Last edited:

crutschow

Joined Mar 14, 2008
34,285
Ok, I put the RC network in and the extra resistor to the inverting input for U3. I can measure the analog voltage now, thank you for that.

It was hitting 50% duty cycle at 5 or 6 volts. Then I change R1 for a 22k resistor and it is now hitting 50% duty cycle at about 2.5v, I need it to hit 50% duty cycle at .55v to match the way the adapter box I am replacing worked and for the display to read properly. I tried 27k for R1 and it flatlined, so I'm at a limit there with the voltage divider.

Can I modify the components to achieve what I need or do I need some sort of amplifier (~ x5) to amplify the temperature sensor voltage to make it compatible with this circuit?
So what variation of the circuit are we talking about?

What is the total range of input voltage that you want 0% and 100% duty-cycle.
 
Last edited:

crutschow

Joined Mar 14, 2008
34,285
You didn't answer my second question. :( I need to know what input you want for 0% duty-cycle and what input for 100% duty-cycle.
 

Thread Starter

gte

Joined Sep 18, 2009
357
I'm sorry. The only measured range I have is a few posts up. I could estimate/guess what 0% and 100% are. 2.5v looks like it would be 100% and .3v should be 27% on ... I'm not sure how to guess at the 0 value besides saying 0v should equal 0 on time. Looking at this closer it appears the scale isn't linear :(


You didn't answer my second question. :( I need to know what input you want for 0% duty-cycle and what input for 100% duty-cycle.
 

crutschow

Joined Mar 14, 2008
34,285
The simplest solution may be to buffer the temp sensor voltage with an op amp to add offset and gain to give the desired PWM duty-cycle versus input voltage.
 
Top