Kick-starting and LED ??

[mpm]

I've not run into this before...
Any ideas?

Situation:
We purchased some membrane keypads with embedded LED's from JRPANEL.COM
(BTW: Nice-looking product, and at reasonable prototype price & delivery.)

And before anyone asks, we're going to get away from Prototyping these now that orders are coming in at a fast enough pace to justify the expense of a "real" keypad.

Anyway - on about 4% of the order, the green LED would not illuminate, even at 8 mA on a 5-volt supply. The other 96% were fine, and would reliably illuminate even at 3 to 4 mA. (The keypad also has lots of red LED's, but no problems with any of them.)

Increasing the current to 10 mA caused all of those in the 4% fail pile to illuminate. Whereupon, removing power and reconfiguring for 5 mA, ALL worked fine, and no matter what we did, we could not get them to fail. (We'd ruled out anything mechanical prior.)

So, my question is: WTF??

Our Senior RF guy suggested the problem was with the "printed silver ink and mechanical crimp-style connector" on the plastic ribbon tail. That somehow the one-time jump to 10 mA was enough to permanent "weld" a reliably conductive trace of lower resistance, and now those keypads are "fixed".

Does that sound even remotely like the culprit to anyone?
Or something else going on...

All affected keypad behaved in exactly the same way.
A higher current to get them going - then could never get them to fail.

I've heard all the stories about cheap Chinese electronics, but is this normal?

 

[Rick C]

On Wednesday, January 15, 2020 at 3:05:40 PM UTC-5, mpm wrote:

I've not run into this before...
Any ideas?

Situation:
We purchased some membrane keypads with embedded LED's from JRPANEL.COM
(BTW: Nice-looking product, and at reasonable prototype price & delivery.)

And before anyone asks, we're going to get away from Prototyping these now that orders are coming in at a fast enough pace to justify the expense of a "real" keypad.

Anyway - on about 4% of the order, the green LED would not illuminate, even at 8 mA on a 5-volt supply. The other 96% were fine, and would reliably illuminate even at 3 to 4 mA. (The keypad also has lots of red LED's, but no problems with any of them.)

Increasing the current to 10 mA caused all of those in the 4% fail pile to illuminate. Whereupon, removing power and reconfiguring for 5 mA, ALL worked fine, and no matter what we did, we could not get them to fail. (We'd ruled out anything mechanical prior.)

So, my question is: WTF??

Our Senior RF guy suggested the problem was with the "printed silver ink and mechanical crimp-style connector" on the plastic ribbon tail. That somehow the one-time jump to 10 mA was enough to permanent "weld" a reliably conductive trace of lower resistance, and now those keypads are "fixed".

Does that sound even remotely like the culprit to anyone?
Or something else going on...

All affected keypad behaved in exactly the same way.
A higher current to get them going - then could never get them to fail.

I've heard all the stories about cheap Chinese electronics, but is this normal?

The "welding" sounds a bit sketchy to me. First, if there was no connection, how was it passing current? If there was a connection, but poor, why didn't the LED light when 8 mA was being passed? But I'm unclear on what the circuit is exactly.

Is this a simple row/column arrangement of contacts that have no electronics other than the LED? Is there an alternate path for the current other than through the LED? How did you pass more or less current given a constant 5 volts on the supply? Did you run the voltage above 5 volts?

If there is no alternate path around the LED and you were able to pass current, the LED should light unless it is defective. I would not trust a defective part that "healed" itself, but how critical are these devices? I'd at least burn them in for a few days just to prevent a nuisance return even if the devices are not critical.

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

 

[mpm]

On Wednesday, January 15, 2020 at 4:03:51 PM UTC-5, Phil Hobbs wrote:

> Maybe a bit of stray carbon ink that burned off at 10 mA?

We did put one of the failing parts under the microscope.
Didn't see anything obvious (of course, it's all translucent white, so a bit hard to tell). Also, that visual inspection test is possibly compromised because we ordered these keypads with the 3M300LSE adhesive, and it was difficult to remove it from the enclosure door.

This JRPANEL outfit is a little squirrely with the file requirements.
Basically, they want a life-size PDF.

That's right: PDF

Example at link below.
And having just looked at it, the web page is MUCH BETTER than it was back when we placed the order!

Link: https://www.jrpanel.com/img/images/JRPanel_Membrane_Switch_Sample.pdf

 

[mpm]

On Wednesday, January 15, 2020 at 3:33:51 PM UTC-5, Rick C wrote:

> The "welding" sounds a bit sketchy to me. First, if there was no connection, how was it passing current? If there was a connection, but poor, why didn't the LED light when 8 mA was being passed? But I'm unclear on what the circuit is exactly.

I'll try to describe the circuit:

5-Volt rail (regulated, never changes)
This goes to a resistor on the PCB.
Then to a header connector (.100 pins)
Then to the ribbon on the keypad tail.
Then, to whatever is inside the keypad (just traces and the LED)
Then, a shared trace on the return back to ground (shared with the red LED's)

We adjusted the current by swapping out the resistors on the PCB.

I should also note that upon further investigation, some in the 96% pass pile would actually illuminate down to 3 mA, but we don't drive them that low..

It does seem that the Chinese Green LED spec is all over the place.
But what is curious is that once you get them lit (via more current) they start to behave normally. And by "normally", I mean here that a 300-ohm resistor on a 5-volt rail seems to be a happy safe medium for them.

We calculated the current as 5V minus the drop across the dedicated resistor, divided by the resistor value. (The board was originally designed with 470-ohm resistor, which does work, 96% of the time.)

About now I should state: We were forced to select the dropping resistor value "in the dark", so to speak, as the Chinese vendor doesn't give the spec for the green LED's they're using. (It's possible they're not even all the same LED's??!)

I'm not so much worried about the forward current through the LED, so long as it is in the range of what's might be considered reasonable (i.e., < 10 mA), but why the "kickstart" thing. That's a mystery right now....

And - we can't really generate a table of R-values vs. illumination because we only had a handful fail and by the time we really got prepped for a deep dive on the problem, all of them were working! I can say that before the higher current was applied, the green LED's WERE CONDUCTING (evidenced by the R-drop), but there was no visible indication whatsoever of the LED being "on". Under normal lighting at that point, of course. But I don't think they were even barely on. It's almost like a hysteresis behavior. But in an LED?

I had them affected units quarantined, and they are burning in now.
Three days so far. No problems.

I got a feeling these membrane keypad LED's are assembled like a sandwich.
Maybe we burned some "intra-layer glue off", or something??

 

[mpm]

On Wednesday, January 15, 2020 at 4:08:47 PM UTC-5, mpm wrote:

I should also state my Senior Engineer on this project is dead set against membrane keypads with that "silver printed ink" stuff.

He's probably right. (He usually is.)

But margins are thin on this one, and unless ordered in high-enough quantities, can't make the BOM work. Although now, we're getting past that point and I'm likely to start shopping around for better keypads.

And I don't want to knock the JRPANEL stuff.
For what it is, (and overall, it ain't bad!), it does fulfill a price-point.

 

[George Herold]

On Wednesday, January 15, 2020 at 3:05:40 PM UTC-5, mpm wrote:

I've not run into this before...
Any ideas?

Situation:
We purchased some membrane keypads with embedded LED's from JRPANEL.COM
(BTW: Nice-looking product, and at reasonable prototype price & delivery.)

And before anyone asks, we're going to get away from Prototyping these now that orders are coming in at a fast enough pace to justify the expense of a "real" keypad.

Anyway - on about 4% of the order, the green LED would not illuminate, even at 8 mA on a 5-volt supply. The other 96% were fine, and would reliably illuminate even at 3 to 4 mA. (The keypad also has lots of red LED's, but no problems with any of them.)

Increasing the current to 10 mA caused all of those in the 4% fail pile to illuminate. Whereupon, removing power and reconfiguring for 5 mA, ALL worked fine, and no matter what we did, we could not get them to fail. (We'd ruled out anything mechanical prior.)

So, my question is: WTF??

Our Senior RF guy suggested the problem was with the "printed silver ink and mechanical crimp-style connector" on the plastic ribbon tail. That somehow the one-time jump to 10 mA was enough to permanent "weld" a reliably conductive trace of lower resistance, and now those keypads are "fixed".

Does that sound even remotely like the culprit to anyone?
Or something else going on...

All affected keypad behaved in exactly the same way.
A higher current to get them going - then could never get them to fail.

I've heard all the stories about cheap Chinese electronics, but is this normal?

Hmm well as a crazy idea maybe tin whiskers are shorting the green led's?
And 10 mA is enough to blow it out? This would be easy enough to check.
Measure voltage across failed led.

George h.

 

[Phil Hobbs]

On 2020-01-15 15:05, mpm wrote:

I've not run into this before...
Any ideas?

Situation:
We purchased some membrane keypads with embedded LED's from JRPANEL.COM
(BTW: Nice-looking product, and at reasonable prototype price & delivery.)

And before anyone asks, we're going to get away from Prototyping these now that orders are coming in at a fast enough pace to justify the expense of a "real" keypad.

Anyway - on about 4% of the order, the green LED would not illuminate, even at 8 mA on a 5-volt supply. The other 96% were fine, and would reliably illuminate even at 3 to 4 mA. (The keypad also has lots of red LED's, but no problems with any of them.)

Increasing the current to 10 mA caused all of those in the 4% fail pile to illuminate. Whereupon, removing power and reconfiguring for 5 mA, ALL worked fine, and no matter what we did, we could not get them to fail. (We'd ruled out anything mechanical prior.)

So, my question is: WTF??

Our Senior RF guy suggested the problem was with the "printed silver ink and mechanical crimp-style connector" on the plastic ribbon tail. That somehow the one-time jump to 10 mA was enough to permanent "weld" a reliably conductive trace of lower resistance, and now those keypads are "fixed".

Does that sound even remotely like the culprit to anyone?
Or something else going on...

All affected keypad behaved in exactly the same way.
A higher current to get them going - then could never get them to fail.

I've heard all the stories about cheap Chinese electronics, but is this normal?


Maybe a bit of stray carbon ink that burned off at 10 mA?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Rick C]

On Wednesday, January 15, 2020 at 4:26:46 PM UTC-5, George Herold wrote:

On Wednesday, January 15, 2020 at 3:05:40 PM UTC-5, mpm wrote:
I've not run into this before...
Any ideas?

Situation:
We purchased some membrane keypads with embedded LED's from JRPANEL.COM
(BTW: Nice-looking product, and at reasonable prototype price & delivery.)

And before anyone asks, we're going to get away from Prototyping these now that orders are coming in at a fast enough pace to justify the expense of a "real" keypad.

Anyway - on about 4% of the order, the green LED would not illuminate, even at 8 mA on a 5-volt supply. The other 96% were fine, and would reliably illuminate even at 3 to 4 mA. (The keypad also has lots of red LED's, but no problems with any of them.)

Increasing the current to 10 mA caused all of those in the 4% fail pile to illuminate. Whereupon, removing power and reconfiguring for 5 mA, ALL worked fine, and no matter what we did, we could not get them to fail. (We'd ruled out anything mechanical prior.)

So, my question is: WTF??

Our Senior RF guy suggested the problem was with the "printed silver ink and mechanical crimp-style connector" on the plastic ribbon tail. That somehow the one-time jump to 10 mA was enough to permanent "weld" a reliably conductive trace of lower resistance, and now those keypads are "fixed".

Does that sound even remotely like the culprit to anyone?
Or something else going on...

All affected keypad behaved in exactly the same way.
A higher current to get them going - then could never get them to fail.

I've heard all the stories about cheap Chinese electronics, but is this normal?

Hmm well as a crazy idea maybe tin whiskers are shorting the green led's?
And 10 mA is enough to blow it out? This would be easy enough to check.
Measure voltage across failed led.

George h.

My understanding is tin whiskers won't even take 1 mA to blow out. They are so tiny you can't see them without a microscope, so they blow out on very low currents. Not likely here though as they usually take time to form.

--

Rick C.

-- Get 1,000 miles of free Supercharging
-- Tesla referral code - https://ts.la/richard11209

 

[Rick C]

On Wednesday, January 15, 2020 at 4:21:41 PM UTC-5, mpm wrote:

On Wednesday, January 15, 2020 at 4:03:51 PM UTC-5, Phil Hobbs wrote:

Maybe a bit of stray carbon ink that burned off at 10 mA?

We did put one of the failing parts under the microscope.
Didn't see anything obvious (of course, it's all translucent white, so a bit hard to tell). Also, that visual inspection test is possibly compromised because we ordered these keypads with the 3M300LSE adhesive, and it was difficult to remove it from the enclosure door.

This JRPANEL outfit is a little squirrely with the file requirements.
Basically, they want a life-size PDF.

That's right: PDF

Example at link below.
And having just looked at it, the web page is MUCH BETTER than it was back when we placed the order!

Link: https://www.jrpanel.com/img/images/JRPanel_Membrane_Switch_Sample.pdf

Ok, so each LED has a separate lead and wired to a common. You still don't explain where you measure the 5 volts and what you are doing to change the current. I would assume you use a resistor from a 5 volt supply and change the resistor to get different currents?

If that is the case, what is the voltage on the LED when it didn't work? That will tell you what is wrong.

There may well be a short around the LED. It could be some of the ink making the traces if they are printed. You won't know if you don't have both the voltage and the current.

--

Rick C.

+ Get 1,000 miles of free Supercharging
+ Tesla referral code - https://ts.la/richard11209

 

Rick C <gnuarm.deletethisbit@gmail.com> wrote in
news:2e6e5c78-a375-4081-8c22-218af51db760@googlegroups.com:

My understanding is tin whiskers won't even take 1 mA to blow out.
They are so tiny you can't see them without a microscope, so they
blow out on very low currents. Not likely here though as they
usually take time to form.

"blow out" is ambiguous. Now, you have TWO whiskers, one on each
node, pointing toward each other. I am pretty sure that can lead to
some signal issues on some circuits... enough so that NASA and the
gov boys wanted no part of having to characterize that realm. We
remain exempt for that reason. No COTS in space rated for external
excursions.

63/37 should STILL be the rule. Metallic form lead does NOT glean
into the environment.

If it did, the lead reading at water tables near gun shooting
ranges would show higher, and they do not!

RoHS was the most stupid thing the euro boys ever did. It cost the
entire industry billions and it is ongoing.

They should have gone after the mercury and Cadmium and such, but
NOT metallic form lead.

 

[Jasen Betts]

On 2020-01-15, mpm <mpmillard@aol.com> wrote:

I've not run into this before...
Any ideas?

Situation:
We purchased some membrane keypads with embedded LED's from JRPANEL.COM
(BTW: Nice-looking product, and at reasonable prototype price & delivery.)

And before anyone asks, we're going to get away from Prototyping these now that orders are coming in at a fast enough pace to justify the expense of a "real" keypad.

Anyway - on about 4% of the order, the green LED would not illuminate, even at 8 mA on a 5-volt supply. The other 96% were fine, and would reliably illuminate even at 3 to 4 mA. (The keypad also has lots of red LED's, but no problems with any of them.)

Increasing the current to 10 mA caused all of those in the 4% fail pile to illuminate. Whereupon, removing power and reconfiguring for 5 mA, ALL worked fine, and no matter what we did, we could not get them to fail. (We'd ruled out anything mechanical prior.)

So, my question is: WTF??

Our Senior RF guy suggested the problem was with the "printed silver ink and mechanical crimp-style connector" on the plastic ribbon tail. That somehow the one-time jump to 10 mA was enough to permanent "weld" a reliably conductive trace of lower resistance, and now those keypads are "fixed".

it sounds more like some sort of whisker (or other fine concentrated
current path) that was destoyed by the 10mA current

How did the voltage drop of the bad LEDs look at 1mA?

--
Jasen.

 

On Thursday, 16 January 2020 03:04:00 UTC, DecadentLinux...@decadence.org wrote:

Rick C <gnuarm.deletethisbit@gmail.com> wrote in
news:2e6e5c78-a375-4081-8c22-218af51db760@googlegroups.com:

My understanding is tin whiskers won't even take 1 mA to blow out.
They are so tiny you can't see them without a microscope, so they
blow out on very low currents. Not likely here though as they
usually take time to form.


"blow out" is ambiguous. Now, you have TWO whiskers, one on each
node, pointing toward each other. I am pretty sure that can lead to
some signal issues on some circuits... enough so that NASA and the
gov boys wanted no part of having to characterize that realm. We
remain exempt for that reason. No COTS in space rated for external
excursions.

63/37 should STILL be the rule. Metallic form lead does NOT glean
into the environment.

If it did, the lead reading at water tables near gun shooting
ranges would show higher, and they do not!

RoHS was the most stupid thing the euro boys ever did. It cost the
entire industry billions and it is ongoing.

They should have gone after the mercury and Cadmium and such, but
NOT metallic form lead.

Oh they have far more stupid things, both done & planned. This continent doesn't know what it's doing half the time.

 

[Jasen Betts]

On 2020-01-15, mpm <mpmillard@aol.com> wrote:

On Wednesday, January 15, 2020 at 4:08:47 PM UTC-5, mpm wrote:

I should also state my Senior Engineer on this project is dead set against membrane keypads with that "silver printed ink" stuff.

it's a mature technology. my first two computers had that for their
keyboard, they worked ok but eventually the plastic failed with stress
cracks, then several PCs with mechanical switches, then several with
that silver ink stuff deep inside. but I'm back with mechanical
switches now.


--
Jasen.

 

[Rick C]

On Thursday, January 16, 2020 at 1:00:53 AM UTC-5, Jasen Betts wrote:

On 2020-01-15, mpm <mpmillard@aol.com> wrote:
On Wednesday, January 15, 2020 at 4:08:47 PM UTC-5, mpm wrote:

I should also state my Senior Engineer on this project is dead set against membrane keypads with that "silver printed ink" stuff.

it's a mature technology. my first two computers had that for their
keyboard, they worked ok but eventually the plastic failed with stress
cracks, then several PCs with mechanical switches, then several with
that silver ink stuff deep inside. but I'm back with mechanical
switches now.

I had a Microsoft keyboard with what appeared to be the silver printed ink. One of the traces corroded and it quit working. lol It took some four years or so, but still, the trace corroded???

--

Rick C.

-+ Get 1,000 miles of free Supercharging
-+ Tesla referral code - https://ts.la/richard11209

 

[whit3rd]

On Wednesday, January 15, 2020 at 10:30:16 PM UTC-8, Rick C wrote:
now.

I had a Microsoft keyboard with what appeared to be the silver printed ink. One of the traces corroded and it quit working. lol It took some four years or so, but still, the trace corroded???

Or, maybe an impurity in the plastic film didn't wet properly, and the trace wasn't
quite intact to begin with... but yes, any metal with current can corrode, if you
have a moisture film a trickle of current will rust stainless steel. Silver or nickel
aren't immune either (aIthough if you WANT to etch silver, a little chlorate ion helps).

 

[mpm]

On Thursday, January 16, 2020 at 1:00:53 AM UTC-5, Jasen Betts wrote:
> How did the voltage drop of the bad LEDs look at 1mA?

I don't know if we had a chance to measure that.
I've been out sick the past two days with a cold, and my tomorrow is out of the office. So, it'll be Monday before I can post the readings we did take.

From memory, the currents looked about right, but no illumination.
This would have been in the neighborhood of 3.5 mA upwards to 6 mA.
But, after hitting them with higher current, they illuminated and would thereafter work at 4 to 5 mA with no problems.

My money is on something strange going on with the physical construction of the layers in the membrane keypad. The traces themselves are far enough apart I don't see how that could be an issue. And as you may know, keypads are layed out like single-side PCB's, with no options for jumpers. (At least, not through this vendor.)

 

[Michael Terrell]

On Thursday, January 16, 2020 at 7:32:35 PM UTC-5, mpm wrote:

On Thursday, January 16, 2020 at 1:00:53 AM UTC-5, Jasen Betts wrote:
How did the voltage drop of the bad LEDs look at 1mA?

I don't know if we had a chance to measure that.
I've been out sick the past two days with a cold, and my tomorrow is out of the office. So, it'll be Monday before I can post the readings we did take.

From memory, the currents looked about right, but no illumination.
This would have been in the neighborhood of 3.5 mA upwards to 6 mA.
But, after hitting them with higher current, they illuminated and would thereafter work at 4 to 5 mA with no problems.

My money is on something strange going on with the physical construction of the layers in the membrane keypad. The traces themselves are far enough apart I don't see how that could be an issue. And as you may know, keypads are layed out like single-side PCB's, with no options for jumpers. (At least, not through this vendor.)

This reminds me of a batch of boards at Microdynre. The product has a ten year old design that had been rock stable. All of a sudden the Intersil voltmeter ICs weren't working properly. Someone had scratched one of the films to make the blank boards You couldn't see the copper whisker without a microscope. The was an even smaller gap in the whisker which didn't show up with an ohm meter.VCC was 5VDC, and it would arc over the gap, starting around 3.65VDC One of the traces was the input pin to the meter. The RF caused it to display random numbers. Purchasing wanted to scrap the whole run and wait for new blanks, but a quick pass with a new Exacto blade would remove the entire whisker and allowed us to ship the radios on time. A new set of films for the board house was the permanent fix.

 

[Rick C]

On Friday, January 17, 2020 at 12:46:34 PM UTC-5, George Herold wrote:

On Thursday, January 16, 2020 at 7:32:35 PM UTC-5, mpm wrote:
On Thursday, January 16, 2020 at 1:00:53 AM UTC-5, Jasen Betts wrote:
How did the voltage drop of the bad LEDs look at 1mA?

I don't know if we had a chance to measure that.
I've been out sick the past two days with a cold, and my tomorrow is out of the office. So, it'll be Monday before I can post the readings we did take.

From memory, the currents looked about right, but no illumination.
This would have been in the neighborhood of 3.5 mA upwards to 6 mA.
But, after hitting them with higher current, they illuminated and would thereafter work at 4 to 5 mA with no problems.

My money is on something strange going on with the physical construction of the layers in the membrane keypad. The traces themselves are far enough apart I don't see how that could be an issue. And as you may know, keypads are layed out like single-side PCB's, with no options for jumpers. (At least, not through this vendor.)

Well that does sound like a short somewhere. I'm still guessing some
sort of metalic whisker.

If this is a printed ink membrane my money is on stray ink. Not sure how a metallic whisker would be formed. Tin whiskers grow and take time to form..

--

Rick C.

+- Get 1,000 miles of free Supercharging
+- Tesla referral code - https://ts.la/richard11209

 

[George Herold]

On Thursday, January 16, 2020 at 7:32:35 PM UTC-5, mpm wrote:

On Thursday, January 16, 2020 at 1:00:53 AM UTC-5, Jasen Betts wrote:
How did the voltage drop of the bad LEDs look at 1mA?

I don't know if we had a chance to measure that.
I've been out sick the past two days with a cold, and my tomorrow is out of the office. So, it'll be Monday before I can post the readings we did take.

From memory, the currents looked about right, but no illumination.
This would have been in the neighborhood of 3.5 mA upwards to 6 mA.
But, after hitting them with higher current, they illuminated and would thereafter work at 4 to 5 mA with no problems.

My money is on something strange going on with the physical construction of the layers in the membrane keypad. The traces themselves are far enough apart I don't see how that could be an issue. And as you may know, keypads are layed out like single-side PCB's, with no options for jumpers. (At least, not through this vendor.)

Well that does sound like a short somewhere. I'm still guessing some
sort of metalic whisker.

George H.

 

[whit3rd]

On Friday, January 17, 2020 at 10:04:20 AM UTC-8, Rick C wrote:

> If this is a printed ink membrane my money is on stray ink.

Yep, all it takes is a bit of fiber dangling from the (tip/brush/nozzle). I'm guessing
that it didn't show up in visual inspection because it was very thin. Had that
happen on printed circuits, too (etched not-quite-all of the way through).

Wet items are complex at all scales (from wet chemistry up to hillsides about to avalanche).