Short Circuit Proof Lamp Driver Circuit

It depends on the characteristics of the lamp....

I'd have thaught a simple resistor would help considerably. Or use an inherently SC proof transformer...?

We added series resistors but the value had to be kept very low to prevent excessive voltage drop under normal load and a short circuit still eventually takes out the driver (although they no longer burst into flames). There are over 600 lamps in the system, so individual fuses aren't practical. Sometimes when the lamps fail, they fail as a short-circuit, which is one source of the problem. What I was hoping for was a simple FET design with short-circuit detection and feedback to automatically foldback or shut-down the driver under a short circuit condition. I know it's possible. I'm just not a very accomplished analog designer and I thought someone might know of an integrated solution to avoid using discrete voltage comparitors.

Why not use a constant-current driver circuit? Unless, of course, you don't want 200 mA all the time. There are ways to get around that, too, if the source voltage is constant for a given application.

Think about the above, and reply if you think it's useful.

For the most part, I'm a digital designer and only 'dabble' in analog circuitry from cookbooks and reading data sheets. I'm vaguly familiar with the concepts behind a constant-current driver, but wasn't under the impression that this would necessarily make the circuit 'short-circuit proof'. Thanks for this reply. I'll go back and research this option.

It sounds like you may be switching them at some speed??

With 600 lamps you're going to get a high number of failures...

Maybe filament lamps are not a good idea in this project... Could leds be used or is it out of your control?

John.

What I've done in the past to improve reliability for medical applications etc... is to introduce a 'keep alive' resistor across the switch, so that a small current is always flowing through the lamp filament, this just keeps the filament at red heat and reduces the switch on surge current to only two times the steady state current, but at the same time it looks like its off...

It does, of course, mean a much higher power consumption, but if you're switching the filaments off and on fast then you will get a much extended filament life...

John.

I like this idea. If we replace the ULN2076's with discrete drivers using your idea of a weak 'keep alive' resistor, we can sense the voltage across this 'keep alive' resistor, to detect a short circuit condition before turning on the high-current drivers. Thanks! The funny thing is that I could have sworn I saw a news announcement sometime over the past few months in one of the electronic rag's announcing a low to medium current FET driver with built in short circuit protection. But of course, now that I have an opportunity to actually redesign the circuit, I can't find it anywhere ... typical! ... but I did find a half eaten Twinkie and a rusty slide rule, so it wasn't a total loss = ) Thanks again.

This sounds more complex than using constant current drivers . . . An ordinary bipolar transistor can be made to conduct only 200 mA in the nearly fully ON condition (thereby minimizing the power dissipation when conducting) by limiting the base current with a single resistor.

My advice is to look at some application notes for semiconductors and linear IC's to get some ideas.

Thanks Bill, your suggestion sounds interesting and definately less elaborate than the scheme I was envisioning, which is important considering we have over 600 light bulbs. I appreciate your help. - Steve

LM195 has internal thermal, power, and current limiting, "making them virtually impossible to destroy from any type of overload," as stated in National Semiconductor's datasheet (http://cache.national.com/ds/LM/LM195.pdf - sorry, link no longer available)

Thanks Bill. This looks like an ideal solution. I appreciate your help. - Steve

I didn't bother listing those as an option...

They are horrendously expensive - £5 each !!!

They are big - a power package!

If you're going to use 600 of them you had better warn National Semicon to ramp up production!!!

John.

LOL - oops ... I guess we'd make National's day. But I like the general idea. Maybe they've got a lower current version that's also a little easier on the bank.

Thanks guys ... Steve

Thanks for the info, and the effort on your point to look up the tech note and send me the link. I've used LM317's for decades as a voltage regulator but I'd overlooked it as a constant current generator. This is a great tip and meets both my cost and performance goals. We already use LM317's in this application, so now we'll just be buying a whole boatload more! --- which should make the price drop even further.

Thanks again to both of you for you help and your suggestions. Have a great day!

We designed this system in 1982 and it's been deployed on hundreds of airplanes, helicopters, boats, RV's and blimps around the world and is still being produced today. For the most part if you look up in the sky and see twinkling lights, it's either us, a UFO or Goodyear (and we have both of them greatly outnumbered ... = ) The system has been surprisingly reliable and we really have very few complaints. But once in a great while a bulb fails 'short' rather than 'open' and that's when we lose the occational driver. I have an opportunity to redesign this system using modern technology and I'd like to replace the ULN2076 quad 1.5 amp lamp drivers with an equivilant driver or discrete circuit with short-circuit detection / protection. We do offer LED's as an option, but they still cost a lot more than bulbs, so most people don't go for that option as it would greatly increase the cost of the lighted message center.

I use similar 12 or 24V lamp drivers for studio cue lights with self-resetting fuses. Polycom are one of the leading manufacturers of self resetting fuses. The fuses are slow to trip so cope well with cold lamps, but the driver be able to handle the surge before the fuse trips. a slightly oversized driver is usually adequate.

As I was reading the replies to this, I was thinking "a polyfuse would work", Chankley beat me to it. Polyfuse is a trademark name, I think, and there are now a few different manufactures of these devices. They are marketed as "resettable fuses" and such. They are basically a PTC thermistor, initial resistance is low, but as current increases, they heat up, and R goes up. The will reach a thermal equilibrium in the event of a short, and will revert back to the "reset" state once the short is removed. See here for more info on one manufactures

http://www.circuitprotection.com/polyswitch.asp Looks like Tyco bought out Raychem. Another manufacturer is Bourns;

http://www.bourns.com/circuit.aspx?cmsphid=62631736|7413229|8404650

We've used these in electronics systems with success, not too sure how suited they are to lightbulbs, but worth a look. Cheap and simple compared to a constant current circuit :)

Tom

Thanks for the suggestion and link Tom. I'll look into it. - Steve

Thanks Tom for giving the correct manufacturers names, I guess I merged Polyfuse and Raychem. I should have checked and not relied on memory.

Are those poly resettable 'fuses' going to be fast enough to protect a transistor in the event of a short circuit though?

I doubt it very much... John.

Without knowing all the details of how he's planning to revamp the system, it's hard to say if the resettable fuses would work. Earlier he said that with a series-R, the drivers would last much longer under a short condition, maybe the PTC fuse along with series-R would do the trick.

Google found a couple IC's for this, but at one per lamp, it might get a bit pricey!

This thing has all he needs (except multiple outputs):

http://www.onsemi.com/pub_link/Collateral/CS1108-D.PDF

a similar one:

http://www.stmicroelectronics.com/stonline/products/literature/ds/1517.pdf

Tom

Ideally I was looking for a modern replacement for the UCN2076 which is a quad, 1.5 Amp (from memory???) source (as opposed to sink) driver. Each UCN2076 driver supplies +12 to + 28v to the bulb when turned on while the other side of the bulb is chassis ground. The UCN2076 devices would be perfect for this application, if they only had short circuit protection. With this in mind, the idea of a 'resettable fuse' sounds pretty interesting. But you've all given me a host of things to look into, so I appreciate everyone's comments. Thanks - Steve

Steve,

A brief search yielded some promising results, at the risk of spending too much of my employers time, I'll let you you weed through the results. I used the following search on Google

high-side octal driver ic

You can substitute "quad" "dual" "8" etc. to look at different channel numbers. There's a lot of stuff available today with the advancements in high-performance FET's, and built in microcontrollers/logic for the short/open monitoring. ST Microelectronics is one that came up often. Applications would be automotive/industrial, where things have to be pretty robust. Good luck, for grins, let us know what you decide on, it may help someone else when they find this thread in the future.

Tom

Great input. I'll follow your suggestion and look into these drivers. Unfortunately, this 'hot project' was just put on the back burner and I've been reassigned to a higher priority fire drill ... = ) Dilbert Rules ... So I won't be able to provide feedback as to our final direction, as I've been Re-Directed. But thank you all of your time and your great suggestions and input! - Steve

Thanks for your time and your input. Sounds like it's worth looking into. Cost wise, how do you think this would compare to simply using a lm317?In large quantities, I'm guessing the lm317 might cost less than a dollar.

Bill in #14 had one of the answers. The other one is - if you are lucky is a drop-in one. A number of manufacturers produce exact physical copies, but built with LED diodes and a series resistor in the package. If you find what you need, just pull the lamp, plug in the LED lamp, and presto, instant replacement, no inrush, less consumption, no periodic replacement. Start with Newark and Mouser catalogs toward the makers. They are advertised periodically.