Switch re-triggering issue

[Western]

I asked a question in a thread a couple of weeks ago … https://www.electro-tech-online.com/threads/piezo-switch-problems.156394/

We have made some progress, albeit very slowly.

The techs had virtually given up, so has been difficult to get them to try any suggestions … but finally got a phone call this morning with some promising news.

I had been asking them to start the process … and then unplug the switch leads from the board altogether … to see if it would still falsely re-trigger when the automatic process stopped (which has been the fault all along)

In the call this morning they tell me that doing this prevents the fault from occurring. No real surprise there … but it’s progress.

(I had previously got them to remove the lead from the remote switch … which I really expected to be the issue … but that did not help)

I'm after any suggestions as to what could cause these switch circuits to falsely trigger … and what should I try to prevent it.

My suggestion was to get them to twist the (six inch) leads from the switch to the board … and also keep the remote switch leads disconnected until the issue is solved.

Should we try a signal diode across the switch terminals on the board … or any other suppression component suggestions please.

Thanks for any help.

Edit: There's also a 50nF cap to ground on the input of the schmitt trigger.

 

[AnalogKid]

There is no debouncing and no noise/transient/EMI control on your input. 200K is a high enough input impedance to be susceptable to ESD/EMI in the area. Also, I don't see any need for the diode (in that location) nor the 200 K resistor. If the intent is for the 200 K to limit the transient current into the HC14 input protection diodes, it is way better to add that protection externally.

If the switch and the micro run on the same 5 V supply, I also don't see any need for the 2/3 voltage divider. Am I missing something, or many things, or what?

If the circuit really is nothing more than two mechanical switches in parallel (whatever the wire lengths) and a schmitt trigger buffer, then I suggest:

1. Keep the 10K

2. Increase the 22.1 K to 100 K.

3. Relocate the diode in parallel with the 100 K, anode to GND (reverse voltage / negative interference spike protection).

4. Add a diode in parallel with the 10 K, cathode to +5 V.

5. Add a 0.1 uF capacitor in parallel with the 100 K (switch debounce / noise filter)

After all of that, here is another thought. Before you do any of that, as a quick way to narrow down the source of the problem, add a 0.1 uF capacitor after the 200 K resistor, from the C14 input to GND. This will act as a noise / transient / ESD filter *and* switch debouncer.

ak

 

[Western]

Thanks for your detailed reply ak.

I've certainly been questioning the need for the diode and the 200k resistor as well ... but before we go too far down that track ...

... there are 50 of these boards in this plant ... with only about 20 that never play up ... the rest are all over the place apparently ... sometimes good, but often re-triggering.

There is a plant nearby with identical boards and set-up which have never shown these symptoms ... and there are 100's of thousand of the boards throughout the world.

A redesigned board may solve the problem in this instance ... but what could be happening here that doesn't happen at least somewhere else in the world?

 

[AnalogKid]

Without a scope to capture the actual transient waveform, it's hard to say. Variation from one plant to the next is easy to explain; it could be nothing more than a noisy lamp ballast. If yours is the only plant in the world with this problem, that is a significant clue.

That's a lot of boards. What does thig thing do?

ak

 

[Western]

Without a scope to capture the actual transient waveform, it's hard to say.

Yes, life would be so much easier if I was there, in fact I was hoping I might get a free trip out of it all ... but I should be able to talk then through it if I can get some enthusiasm at the other end.

Variation from one plant to the next is easy to explain; it could be nothing more than a noisy lamp ballast. If yours is the only plant in the world with this problem, that is a significant clue.

Yes, good reminder. I have to keep bringing this up myself when I speak to the techs working on it.

There's even more of a clue in that this only happens as the device is going through its shut-off stage and a vacuum ram is operating. As far as I've been able to find out ... it never happens at any other stage.

I keep bringing them back to this point and asking why some do it consistently and others alongside (several feet) never falter ... and trying to get them to look at, think about ... the ram ... its action, function, static generation ... anything!

That's a lot of boards. What does thig thing do?

It's a automatic cup remover in a dairy ... monitors the milk flow until it peters out ... then turns off the vacuum and withdraws the cups. This place has 50, though some have from 10 or 20 up to 120 of them.

 

[dr pepper]

Try reducing the resistor values by a factor of 10.

 

[Western]

Try reducing the resistor values by a factor of 10.

If this was just the one board on my bench, I'd go right ahead and attempt to fine tune the circuit ... but as above ... this particular board works perfectly in thousands of other locations around the world ... there must be something at this particular installation causing the grief with ~30 out of its 50 boards ...

I need to work out what it is and somehow direct a solution. I'm in the middle of Australia ... and the problem is over in New Zealand.

 

[AnalogKid]

That kind of coincidental, repeatable occurance is no coincidence.

Do you have access to whomever put the 200K resistor in there, to find out why?

If the problem is (almost) perfectly repeatable, can you modify two replacement boards and ship them out for a trial? How large is the pcb assembly that includes the circuit(s), and how hard is it to replace in the field? Photos?

ak

 

[Western]

Do you have access to whomever put the 200K resistor in there, to find out why?

No, I have no connection with the designer/manufacturer ... and only have parts of the circuit because I've drawn it out.

The only contact I had, left some years ago ... and he was only a repairer of said equipment ... not the designer.

Interestingly, the previous model/iteration of the board has a more conventional design in that area ... with a 10K pull-up resistor and a 100K in place of the 200K ... and no voltage divider ... providing the full 5 volts across the switch.

If the problem is (almost) perfectly repeatable,

That's the problem ... they shipped me 3 units which have been powered up and running on the bench for weeks ... with no issues at all ... despite my attempts to fry them.

As above ... they have about 20 boards that never give any issue ... the other 30 are all over the place ... some playing up 3 or 4 or 5 times a milking ... and some only rarely. Definitely no repeatability to it at all.

How large is the pcb assembly that includes the circuit(s), and how hard is it to replace in the field? Photos?

PCB is 88 x 55mm. Easy enough to replace in the field ... handful of screws and a few plugs.

I'll send some photos shortly.

 

[AnalogKid]

Another clue - geography. Is there some generalized location detail that separates the good from the bad? Could be adjacecy to other equipment, different power sources (powered from different phase legs of three-phase AC power), etc. The problem could be power-related rather than interference on the switch wires.

If you have one group of boards that never fail, and one group that always fails sometimes, let's go to Intermittent Diagnostics 101 - swap boards. Pick one board that is rock solid, and one board that is among the worst offenders, and swap them. A few screws, come connectors - 10 minutes? Be shure to have them mark each board with a unique identifier.

ak

 

[tvtech]

Tac Switches can play havok. Sometimes (depending on the Weather etc).
I would start there. They cause probems that sometimes make it look like the Microprocessor has lost its marbles. I see it here too often on TV's. You cant test them with a Meter. Older Sansui and Tedelex sets have that problem a LOT....sets starting Auto Tuning on their own, than Menu flashes on and off, than maybe it shows than volume and than jumps to Video from TV etc....disconnect one switch at a time until the problem stops. Replace that switch with a new one.

Normally its only 1 switch playing up that confuses the Micro.
tv

 

[alec_t]

Have you checked that your problem machine is properly grounded? Rogue grounds can cause all sorts of issues.

 

[Western]

Sorry, been a big weekend and it's already late Sunday night ... actually early Mon morning already.

The problem could be power-related rather than interference on the switch wires.

They started off with the recommended power supply for the whole 50 boards ... then halved the loads between 2 supplies ... and then between 4 supplies !!!

Please keep in mind they are using identical equipment on a neighbouring property with absolutely no issues.

Having said that, until we find the actual fault, I'm not discounting anything.

let's go to Intermittent Diagnostics 101 - swap boards. Pick one board that is rock solid, and one board that is among the worst offenders, and swap them.

Yes, to their credit they had done exactly that, well before I got called in ... and were able to explain the process and reasoning behind their actions clearly ... and the results they got as well.


Thanks for your input too, tvtech

Tac Switches can play havok. Sometimes (depending on the Weather etc).

My first 30 years at work were as a tv tech ... though obviously tact switches weren't around early in the piece. Having said that ... I don't reckon I've seen all that much trouble from them myself ... so may be more an environment issue as much as anything else.

These units are using piezo switches ... in an effort to avoid moisture killing normal press button switches. Again ... the neighbouring facility is using piezos as well.

Normally its only 1 switch playing up that confuses the Micro.

Yes, makes sense. With these boards, if you upset the micro and cause a 'reset' ... it goes through an obvious startup routine ... quite different from what it does when this fault occurs ... which is identical to what happens when you press the start button.

Have you checked that your problem machine is properly grounded? Rogue grounds can cause all sorts of issues.

Actually, this is something I have asked lots of questions about.

The boards and the power supplies are not earthed at all (other than the primary of the power supplies). They run on 24V DC and their outputs switch a couple of tiny vacuum solenoids ... which again are not earthed.

I did ask them to ground the negative of the supply (or supplies) ... but I don't think they have tried that yet. I must make sure they at least try it on one or more sections of the plant.


Thanks for all the input guys ... you're keeping me honest and making sure I'm considering all the things that could possibly be the cause. Thank you.

 

[AnalogKid]

Pick one board that is rock solid, and one board that is among the worst offenders, and swap them.

Yes, to their credit they had done exactly that, well before I got called in ... and were able to explain the process and reasoning behind their actions clearly ... and the results they got as well.

You are such a tease. AND THE RESULTS WERE ... ? ? ?

ak

 

[eTech]

hi

Apparently, a piezoelectric pushbuton switch is a solid state switch that has no bounce. Its internal circuitry is driven by the "piezoelectric effect" and drives an internal FET that acts like a switch contact (except with no bounce). When pressed it is active for a specific duration.

eT

 

[Western]

You are such a tease. AND THE RESULTS WERE ... ? ? ?

Unequivocally that the boards are NOT the problem.

When they swapped boards between different locations ... the fault stayed with the location ... not the board.

And when they fitted brand new boards, the fault remained also.

I'm not too good a communicator by the looks ... sorry about that.

 

[Western]

Thanks for your input eTech ...

Apparently, a piezoelectric pushbuton switch is a solid state switch that has no bounce. Its internal circuitry is driven by the "piezoelectric effect" and drives an internal FET that acts like a switch contact (except with no bounce).

As above, the reason they use these in this situation is because they are completely sealed ... so there is not the issue they have elsewhere with water getting into the switch contacts.

These particular switches are used with several different models of boards from this manufacturer ... and seen as the top of the line answer for reliability.

They are something like $90 each !!! ... so they're not cheap junk ones.


Something else I need to point out.

Each board has two switches attached ... one to set it into 'manual' ... and a 'start' button.

When the fault occurs ... it is only ever the function normally initiated by the 'start' button that occurs ... and never the manual function.

If they were a dud batch of switches ... or interference was somehow triggering the switches ... statistically we should be getting 50-50 strikes with the manual and the start buttons ... yet we never get any where it randomly goes into manual ... zero.

I am quite convinced it is not the switches that are causing the fault.

 

[AnalogKid]

Unequivocally that the boards are NOT the problem. When they swapped boards between different locations ... the fault stayed with the location ... not the board. And when they fitted brand new boards, the fault remained also

Thanks, that clears up a lot (no sarcasm; it really does).

So, going forward, is your main task to determine the root cause of the problem in the machine, or determine how to compensate for it with changes to the board? Or, is there another option?


ak

 

[eTech]

Thanks for your input eTech ...

When the fault occurs ... it is only ever the function normally initiated by the 'start' button that occurs ... and never the manual function.
I am quite convinced it is not the switches that are causing the fault.

Hi

Which one is the "start" button on your diagram?
So you were able to isolate one switch from the other to determine which "input" is causing the fault?

If so, and the switches are not bad, then it has to be the wiring or something in close proximity to the wires inducing a glitch. The input circuit on your diagram looks a little strange...seems like it should only have transient protection since there is no switch bounce to deal with.

The one thing that is never the same at different facilities is power. Maybe a large motor or something is inducing noise into the wires.

Since the switches actually have semiconductors in them, is there special circuiting required when wiring them in parallel?
Is the same terminal on each button connected to the input and ground?
Basically, its like two FETS with source and drain connected to each other.

eT

 

[Western]

So, going forward, is your main task to determine the root cause of the problem in the machine, or determine how to compensate for it with changes to the board? Or, is there another option?

Absolutely to determine the root cause of the problem.

Am not interested in masking the fault in any way. Should be no need to, as the numbers of boards out there in the rest of the world that work without issue, proves to me there's nothing wrong with the design ... as weird as it is.

Plus, modifying all the boards in that environment would not be an easy task.