Using Failure Modes of Components as Indicator
Using Failure Modes of Components as Indicator
(OP)
I am building a couple of simple detector circuits to determine if our board has seen an overvoltage or ESD event. The limiting reagent is that it must be real cheap because it will go onto several thousand boards. And when I say cheap, my boss doesn't want to dole out $1.10 for 15mA fuses. He wants to use cheap 5cent components that would fail rather quickly, and our technician can simply measure with a voltmeter.
To detect an electro static field, I'm using a circuit like this:
Ridiculously Sensitive Charge Detector
http://amasci.com/emotor/chargdet.html
but with a zero ohm resistor or a 2milliamp inductor or a 2.5V Tantalum capacitor instead of the LED. But the problem is that these devices take far more abuse than they are rated for. I can run 1AMP through the 2milliamp inductor and it still shows no measurable in-circuit difference. On larger voltage Tantalum capacitors I can kill one with reverse voltage ~equal to its full voltage rating, but the 2.5V tantalum caps can take more than 5V of punishment. I'm thinking of using a simple signal diode with minimal Power Dissipation capability, Ifsm=Current Forward Surge Maximum less than 30mAmps but haven't gotten those diodes yet. A cheap voltmeter looking at Hfe sees the difference on the diodes that I do have, but they are 4.3V Zeners. Are Zeners more susceptible to failure than signal diodes?
My question would be, in this circuit, is there a cheap device that anyone has seen fail far too easily that would be usable? Perhaps a cheap Op-Amp? I've tried to blow the gates on our JFET but they are remarkably robust.
Hopefully, the circuit fails in the open circuit mode, but that is not all that necessary. I tried Aluminum Electrolytic caps that are advertised to fail in the open mode but they fail in short circuit, (and they fail far above their rating) so that didn't work.
Also, to monitor a 5V power rail, a 5.5V Zener would pull any extra power above 5v, and hopefully that current would kill the next item in series. What would be the best item?
And the last circuit is simply a long wire to pick up ESD and funnel it to Ground rather than through our chips on board, with an indicator device between the wire and Ground. I'm pretty sure the best candidate for this is the Tantalum Capacitor but I'm hoping to hear of some cheap current-sensitive device that drives engineers crazy because they fail so often. What would be the best candidate?
To detect an electro static field, I'm using a circuit like this:
Ridiculously Sensitive Charge Detector
http://amasci.com/emotor/chargdet.html
but with a zero ohm resistor or a 2milliamp inductor or a 2.5V Tantalum capacitor instead of the LED. But the problem is that these devices take far more abuse than they are rated for. I can run 1AMP through the 2milliamp inductor and it still shows no measurable in-circuit difference. On larger voltage Tantalum capacitors I can kill one with reverse voltage ~equal to its full voltage rating, but the 2.5V tantalum caps can take more than 5V of punishment. I'm thinking of using a simple signal diode with minimal Power Dissipation capability, Ifsm=Current Forward Surge Maximum less than 30mAmps but haven't gotten those diodes yet. A cheap voltmeter looking at Hfe sees the difference on the diodes that I do have, but they are 4.3V Zeners. Are Zeners more susceptible to failure than signal diodes?
My question would be, in this circuit, is there a cheap device that anyone has seen fail far too easily that would be usable? Perhaps a cheap Op-Amp? I've tried to blow the gates on our JFET but they are remarkably robust.
Hopefully, the circuit fails in the open circuit mode, but that is not all that necessary. I tried Aluminum Electrolytic caps that are advertised to fail in the open mode but they fail in short circuit, (and they fail far above their rating) so that didn't work.
Also, to monitor a 5V power rail, a 5.5V Zener would pull any extra power above 5v, and hopefully that current would kill the next item in series. What would be the best item?
And the last circuit is simply a long wire to pick up ESD and funnel it to Ground rather than through our chips on board, with an indicator device between the wire and Ground. I'm pretty sure the best candidate for this is the Tantalum Capacitor but I'm hoping to hear of some cheap current-sensitive device that drives engineers crazy because they fail so often. What would be the best candidate?





RE: Using Failure Modes of Components as Indicator
If a component fails repeatedly - and that is what it seems to be about - then concentrate on finding out if it fails from bad design, running close to margins or external influence like ESD.
Spending resources on an ESD indicator will not get you anywhere. ESD is everywhere and something you need to design for. So, no need to detect it.
Gunnar Englund
www.gke.org
--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.
RE: Using Failure Modes of Components as Indicator
Another issue is that ESD is normally far more of a problem during servicing, when a technician is actually touching the exposed circuit card (perhaps even powered off), than when it is installed in its chassis. An active charge detector is obviously not going to work when the card has been powered off.
As Gunnar suggests, put your efforts towards ESD design. Plus warning labels, packaging, manual updates and training.
RE: Using Failure Modes of Components as Indicator
Your attempt to use ordinary components as the equivalent of fuses to show a 'tripped' condition is largely futile, as although they obviously have failure modes these will not be guaranteed to occur at the sort of currents your detector circuit can provide.
If you are experiencing large numbers of failures in the field it may well be time to re-engineer the board or at least add protection circuitry - not detection circuitry. Without more details of the board it is difficult to provide advice, but it could well involve components like 'transorbs' on the inputs which are not cheap. From your previous comments your boss probably won't wan't to go for that, but there probably isn't much alternative: reliability has to be designed in from the start otherwise trying to add it afterwards will always be more expensive and difficult.
RE: Using Failure Modes of Components as Indicator
The OP's post implied redesign.
RE: Using Failure Modes of Components as Indicator
What's more problematic is that people obviously don't intentionally design their parts to fail at a particular threshold, so part-to-part variations in behavior must be expected, and any data collected therefrom would be quite worthless.
TTFN
FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize
RE: Using Failure Modes of Components as Indicator
In answer to various posts, yes, this is a design already in the field. In fact it is multiple designs but they all have 5V rails and technicians have a tendency to pick up the boards by their edges, which is coincidentally where most of our ESD damage is taking place. We have already spent resources towards warning, packaging, training etc. but we continue to get returns that are obviously due to EOS. That means our customers are not using proper ESD protection but claim they do, and we lose margin on the returns. So if we can show that the ESD circuit is tripped, the warranty is voided and our repairs are billable rather than a loss.
Adding transorbs to each communication line is prohibitively expensive (not only in components but in engineering redesign time ,characterization & manufacturing) and could affect each circuit's capabilities.
An active charge detector is obviously not going to work when the card has been powered off -- very true, which is why I'm working on the third circuit which is a power-off indicator.
Since our team inherited these designs out in the field, the effort is towards a post-design piggyback circuit that, if it works, would be implemented in future designs.
Thanks again for the feedback.
RE: Using Failure Modes of Components as Indicator
A self-powered voltage monitor, however, will most likely cost more than $1.10.
TTFN
FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize
RE: Using Failure Modes of Components as Indicator
Design your products to handle expected abuse in the field, not more ways to monitor said abuse. If $1 is considered too high of a price to pay for such a retrofit, your company has no chance in hell of resolving this issue and will continue to bleed money with repairs and field calls.
Do you want your bill up front or spread out over 10 years?
Dan - Owner

http://www.Hi-TecDesigns.com
RE: Using Failure Modes of Components as Indicator
If your boards are having this problem fix it. A decent board will survive shipping in loose white peanuts and handling by the worst Scotch-Tape packing noob. It's not hard to protect a board. It's not expensive either. Generally a few added resistors and in some cases a few capacitors. Perhaps in a really bad case - a better layout job.
As for the fuse, there are many papers on using circuit traces as a reliable fuse with all the equations to set the trip point. You can't get cheaper than a piece of trace. You put said trace between points you can later place a fuse across, perhaps a Pico fuse, to resurrect the board.
Keith Cress
kcress - http://www.flaminsystems.com
RE: Using Failure Modes of Components as Indicator
***Indeed, the plan is to test the circuits at the factory just before shipping.
The problem right now is, the "weak" parts aren't that weak and do not fail at the currents/Voltages the ESD detector circuit provides.
RE: Using Failure Modes of Components as Indicator
If the board is digital, then it can tolerate having some leakage currents running around, so a slightly conductive overcoat might be also an option, which would dissipate the ESD charge.
TTFN
FAQ731-376: Eng-Tips.com Forum Policies
Chinese prisoner wins Nobel Peace Prize
RE: Using Failure Modes of Components as Indicator
***I thought so as well until I had called my cell phone company when my phone died. They basically checked a circuit indicator that showed whether it had been in contact with water before issuing an RMA. I happened to have the exact same phone model that had gone into the drink for a few seconds and, sure enough, the indicator was different. If such an approach works with $5Billion companies, it might work with our puny $14M company.
Design your products to handle expected abuse in the field, not more ways to monitor said abuse.
***The products are already designed, we inherited the design.
If $1 is considered too high of a price to pay for such a retrofit, your company has no chance in hell of resolving this issue and will continue to bleed money with repairs and field calls.
***Very well said. I'm not sure if I can accomplish this task before me. These sort of things come down to the pennies when it comes for time for the CEO to make a decision, and sometimes at that point it is difficult for him to see the forest for the trees. I would hope that he would listen to me when the time came but it's more likely that he would give credence to someone like you rather than me when we're saying the same thing... that's just the way his mind operates. Thanks for the feedback.
RE: Using Failure Modes of Components as Indicator
Anyone care to guess what started happening? Lawsuits. The stickers weren't foolproof, and high humidity conditions were setting them off. Warranties were denied, rumblings of class-action lawsuits started to form like thunderheads on the horizon.
It never ends well for the company...
Dan - Owner

http://www.Hi-TecDesigns.com