noise discriminator

[throbscottle]

I found this http://www.scarpaz.com/documents/an-366.pdf - on page 23 there is a "noise discriminator" circuit. It looks like just the thing to eliminate (or at least, improve) noise from mechanical vibration going into the photo-interrupter on The Ridiculously Complicated PCB Drill. It has a very vibratey spindle motor, I'm thinking of getting a better one but even smooth motors still cause issues.

I was wondering if there's an easy way to avoid having to use a D type flip-flop in there?

I also found http://www.ti.com/lit/an/scla014/scla014.pdf?ts=1597861624556 on page 17, which might be the better option - the Fairchild document has a different one on page 26.

The TI circuit would need changing to PNP if I used it.

I'm already using a 74HC123 monostable to solve some issues so it's part-way there.

What I actually have is potentially some spare xor gates and potentially a spare monostable. Though I wonder if there's a better way using our old friend the 555?

They all seem to do pretty much the same thing anyway. I just want to get rid of the short pulses caused by vibration. I don't really want to have to rename the project as The Very Ridiculously Complicated PCB Drill... Though it might be fun

(edit - wouldn't need to change the TI version, I'm just a dummy...)

 

[throbscottle]

Looks like it might be better if I make the mono non-retriggerable.

 

[gophert]

The Fairchild version looks super easy and perfect for short-pulse noise elimination.

 

[dougy83]

Can you just use the 555 as a schmidt trigger (pins 2/6 together as input, 3 as output), with an RC LPF on the input signal? You can put a diode across the resistor if you need the output to change quickly for a low input (like in the app note). Note that the 555 will obviously invert the output level.

 

[throbscottle]

I tried the TI circuit, works well enough, flakey breadboard connections notwithstanding (add a ton of vibration to that too...). Had to add the other half of the '123 onto it because I need the output to always default to low (and the input source can rest low or high).

I didn't want to add another chip for the Fairchild circuit but yes it does look just right. TBH it might come down to pin counts - 2 transistors + resistors, cap and another RC net adds more legs than a 14 pin dual d type. Sad but true.

I must investigate the 555 route. My first line of attack was a slow LPF but in this case it doesn't help. Slow it down enough to filter out the extra pulses and I get other problems. But I would not be surprised if in the 100's of 555 circuits out there, there is one that does what I want.

 

[throbscottle]

Well well well.. http://www.next.gr/other-circuits/5...pulse-width-detection-circuit-555-l59157.html
Look familiar?

 

[throbscottle]

Yay, got it working quite well on a little perf-board. Needs a little bit of tuning. Put an xor gate to provide an invert-or-don't function between the discriminator and the next monostable, which was convenient to add a resistor and small cap to the output of to get rid of some spikes that were appearing because the input isn't a very good square wave. Almost completely ignoring vibration now

So now I'm going to pull it apart and try the one from the Fairchild app note, since it's simpler and should handle the less-than-perfect input a lot more graciously! Only trouble is I would have to get some single D types. grrrr....