Is there a way to detect 1ms pulse with voltage of around 300mV?

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

ovrebekk — Fri, 06 Jan 2017 16:01:34 GMT

I see you already got a lot of good suggestions, but I would like to know what reference you used for the LPCOMP when you tried it?

Probably none of the internal references will cut is, as they could all be above 300mV (depending on your supply voltage).

You can still use an external reference though. You could set up a voltage divider to generate any voltage between 0 and VDD, and use that as reference on either the AREF0 or AREF1 pins.

I haven't tested this myself, but in theory it should allow you to wake up on the 300mV pulse.

You will have to disable the LPCOMP whenever you use the ADC though, which could be a problem depending on how often you get pulses and how critical it is that you detect them all.

Best regards
Torbjørn

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

Roger Clark — Fri, 06 Jan 2017 00:51:16 GMT

@kernels

No worries..

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

Roger Clark — Fri, 06 Jan 2017 00:50:26 GMT

I'm not sure which response you replied to, as you posted your response and an answer

Using the ADC is not ideal for pulse detection, because of the acquisition time

I don't know the acquisition time of the ADC on the nRF51822 but I'm fairly sure if you ran it constantly, you would be able to detect a 1mS wide pulse.

However normally pulse detection is done using GPIO and possibly via interrupt GPIOTE

Do you require a low power solution, as I don't think an OpAmp is likely to be low power.

What OpAmp did you try using, is it "rail to rail" i.e will it drive the output to the supply voltage ?

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

Hein — Thu, 05 Jan 2017 23:19:32 GMT

Ahh yes of course, should have paid more attention when reading your initial answer. That is a good solution.

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

Roger Clark — Thu, 05 Jan 2017 23:14:08 GMT

You forward bias the base / gate voltage on the transistor to just below its switch on point, and the capacitor provides DC isolation, so that the bias voltage is maintained until the actual pulse arrives.

I've done a quick model of this in LTSpice and I can get it to work with a normal NPN transistor, but at the moment I get the GPIO triggerring on the trailing edge of the pulse.

I'm sure somebody who is skilled in analog design could easily design a single active component circuit to do better.

Of course the problem with this, is that you need to maintain the forward bias voltage and there would be leakage into the base of the bipolar transistor. Even so, my quick hack of a circuit was only taking 12uA

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

KS — Thu, 05 Jan 2017 23:12:56 GMT

thank you for your reply. Yes, we are using an op amp and the gain is currently increased, our pulse now shoots to 1.5V still we are unable to detect such pulse in ADC or LPCOM. some reference will be appreciated thank you

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

Hein — Thu, 05 Jan 2017 23:05:59 GMT

Wouldn't 300mV be too low to activate a transistor or fet ? You would need a comparator or opamp ?

RE: Is there a way to detect 1ms pulse with voltage of around 300mV?

Roger Clark — Thu, 05 Jan 2017 20:24:11 GMT

I think the best solution is to use a transistor as an amplifier for this signal, and then use a normal GPIO input.

Assuming you simply want to detect that there is a pulse, and you dont want to measure its length, you can feed the signal via a capacitor to a forward biased transistor

Note. This will take some current all the time, but you can probably limit this by careful choice of bias resistor values, and perhaps use a FET rather than a bipolar transistor.

Edit. Note. This method would also have a small delay between the pulse arriving, and it being transferred though the capacitor.