Detecting a change in signal

[nickexists]

I have an analog signal and I want to detect any change in it, whether it rises or drops. I am working between 0 and 3.3 volts. Currently I have the signal going through a simple high pass filter that rests at 1.65 volts. Let's say I want to detect when that voltage deviates by 0.1 volts.

Of course, I could create two reference voltages, one at 1.75 volts and one at 1.55 volts and then feed each of these into a comparator. But is there a more clever way that I could do this with only one comparator? I should mention that this project will be battery powered so power consumption is a factor.

Thanks!

 

[Old Steve]

I have an analog signal and I want to detect any change in it, whether it rises or drops. I am working between 0 and 3.3 volts. Currently I have the signal going through a simple high pass filter that rests at 1.65 volts. Let's say I want to detect when that voltage deviates by 0.1 volts.

Of course, I could create two reference voltages, one at 1.75 volts and one at 1.55 volts and then feed each of these into a comparator. But is there a more clever way that I could do this with only one comparator? I should mention that this project will be battery powered so power consumption is a factor.

Thanks!

I think you might be stuck with two comparators, but I'll be interested to see if someone can come up with the solution that you'd like.....

 

[Gryd3]

If the voltage input is always know, Old Steve's idea is probably the most reliable.
Otherwise you can look into using opamps. Specifically the differentiator and integrator circuits using them.
One will produce an output based on the 'change' of the input. The other will produce an output based on voltage and duration.
Using these, you can turn the device on. Wait for the voltage to settle, then use the opamps to watch for the voltage changing. This would be more difficult to tune though, and there is no guarantee that the output would return to normal when the input does if they drift too far.

 

[AnalogKid]

The window (dual) comparator method is reliable, and independent of the signal rate of change, overall bandwidth, etc. And with open-collector comparators you get free output combining. But it does use up a few parts. When you say "deviate", how fast? Are you trying to detect a relatively slow kind of change like baseline drift, or a faster change that is a part of what the signal is signalling?

ak

 

[nickexists]

The signal that I am looking at is quite slow. The deviations from baseline that I expect to see last for about 1 second. I had initially considered a single comparator that was set up to have a hysteresis of about 0.2V so that the baseline voltage would rest right in the middle. But I realized that if the comparator was triggered by a rising signal then it wouldn't be able to be triggered again until there was a falling signal.

I should mention that I don't need to be able to distinguish between a rising and falling signal, I just need to be able to detect them both and get a single digital signal out.

 

[Arouse1973]

I was thinking the same as ak. This might be of interest to you?http://cds.linear.com/docs/en/datasheet/1042fa.pdf
Thanks
Adam

 

[nickexists]

That's exactly what I was looking for! Thanks.

 

[davenn]

nice find, Adam

 

[Old Steve]

What he ^ said.