Hey guys

This question is a spin-off of another question, but the scope has changed.

https://forum.allaboutcircuits.com/threads/detect-analog-pulse-using-ne558.151437/#post-1298250

How do I detect any fluctuation of a constant voltage in positive or negative direction when it move outside +/- 20mV from zero volts? I am receiving an analog pulse in 1 - 5 second intervals and I need to detect the starting point of the pulse with a 10- 20 micro seconds accuracy. The catch is that I don't know if the pulse is going to go down or up first. I had a look at Schmitt triggers, but I don't think that is going to work for this application?

Any ideas is appreciated.

 

Amplify the signal (say by 10), feed the signal into a precision rectifier, then use a comparator circuit to detect that the input has changed by +/-20mV.

Or even, don’t bother with the rectification, and use a window detector.

 

The response time of a window comparator built with a dual-comparator IC, such as an LM393,would be faster than a precision rectifier.

 

Using a DelSig, running at ~ 10 bits 4 uS, with a CM range that extends
below the ground rail, and using its differential input, looks like 6-8 US
for HW + code execution to detect.

The PSOC 5LP has another 95+% resources beyond just this small part
of the internal resoureces. Component list attached. A component is an
internal resource of PSOC chip..

This could even be done as background task while processor is doing
other stuff by using custom internal component creation with schematic
capture and DMA. Use digital comparator component, DMA the A/D
result to that. Or Verilog if you prefer.



Regards, Dana

 

Thanks Hymie and Crutschow, I will give that window comparator a try, I have some lm393's here.

 

Why not to apply the KISS principle? Keep It Stupidly Simple. Just small capacitor for separating, RC integrator for filtering and opamp for amplification. Monovibrator for fixing it long enough to be seen.

 

Why not to apply the KISS principle? Keep It Stupidly Simple. Just small capacitor for separating, RC integrator for filtering and opamp for amplification. Monovibrator for fixing it long enough to be seen.

Thanks Janis, but would that work positive and negative?

 

A simple ADC will do here. Just pick an MCU with an ADC (preferably 10bit) and measure. The ADC has about 10mV per division, so if your voltage moves up or down, with 2-3 divisions you will know.

The problem is that you detect below "0V" and the ADC is normally from "0V" to "5V", you will need to set the external pins of the MCU with the range of "-2.5V to +2.5V".

 

Hey guys

This question is a spin-off of another question, but the scope has changed.

https://forum.allaboutcircuits.com/threads/detect-analog-pulse-using-ne558.151437/#post-1298250

How do I detect any fluctuation of a constant voltage in positive or negative direction when it move outside +/- 20mV from zero volts? I am receiving an analog pulse in 1 - 5 second intervals and I need to detect the starting point of the pulse with a 10- 20 micro seconds accuracy. The catch is that I don't know if the pulse is going to go down or up first. I had a look at Schmitt triggers, but I don't think that is going to work for this application?

Any ideas is appreciated.

Use opamp circuit to multiply it signal by itself (input^2) result positive output.
Delay defined by slew rate opamp.

Picbuster