fried VME modules

co> >> It would be harder to get the smart intern!

It might be difficult to ask them to swap out the power supplies, but easy to swap out the connectors. The 5v connectors are molex type 9 pins. Simple enough to provide a male to female tap with over-voltage cutoff.

Give them free cables with time limited firmware. Charge for firmware upgrades.

then it's not your problem to worry about.

You obviously don't have customers. Or relatives. ;)

Cheers

Phil Hobbs

Are these cards running off a welding power supply? The caps in those over-shooting pieces of junk will probably dry up before anybody musters up the courage to install new cards in the chassis. Change is real hard, too hard in fact.

relatives with computer problems get shipped a new computer. I don't screw around with old trash. It's not worth anybody's time.

When I had customers, they were allowed to make any decision they wanted. I billed by time. The longer and more complex they wanted a process, the happier I was to assist. I'd make suggestions that might be helpful, but if they preferred to do things the hard way, that was on them and not a hobby project for me to take care of.

Okay, you were selling your time then, rather than products. Makes a difference. There are some customers who will waste your time if you let them, but IME that's not the norm.

Helping out the good ones makes a lot of sense from many perspectives.

Cheers

Phil Hobbs

One of my customers gets one of every new VME module for free. And most get demo loaners, prototypes, free upgrades, problem solving, simulation. They seem to like that.

We sell stuff, not engineering hours.

Working for free strikes me as a sound business practice.

Customers love it. Then they buy stuff.

he also just gives stuff away too. Seems like a nice guy.

I wish but no time, three clients yanked me out of early retirement.

If you just have to muffle a wee spike you could use a TL431 that drives a HUGE FET to swallow the spike. Must have a big fat SOA. The VME module then should have a fuse in case the spike doesn't fall into the "wee and short" class.

If longer or permanent overvoltage is possible it would have to have a crowbar using a HUGE SCR. Or if you need belt plus suspender, both.

Of course they love free stuff, but whether that translates into future sales is another question. And it means the stuff that isn't free is over-priced to pay for the "free" stuff (which really isn't free).

Not really. The incremental cost to build most products is a small fraction of the selling price. Advertising is expensive and usually provides no feedback, whereas we know what happens when we loan or give away a unit.

If Pratt or Boeing want a loaner or a freebie, they get one. That's an awfully good investment.

Buy decisions are made by people, and people are emotional. If they like us, and know we want to help them, and see that we aren't penny-pinchey greedy, it can tip the decision to design our stuff into their system, or not.

Another good thing about loaners/freebies is that some of their people usually have to write code to try it in their system. And they'd prefer to not throw that away.

The transient seems to last 10 ms or so, so I guess we could stress parts and connector pins a lot for that long.

This is about it:

You're hitting that poor little 20-mA LM4040 a bit hard, no?

If you've got the adjustable version of the TLV431 in stock, you could use the J. Thompson Memorial Voltage Detector thing--voltage divider to the ADJ pin, anode to the base of the BCX71 via a resistor. That'd get you a sharper turn-on, and it's pretty fast--looks vertical on a 1 us/div plot, according to the Diodes datasheet.

It pulls down to about 1 V.

Cheers

Phil Hobbs

I thought that too, but it hits about 40 mA for a few milliseconds so is probably OK. I could use a depletion fet instead of the resistor.

I could do that, but it does add the resistor tolerances so I'd have to push the clamp start up a tad. I don't have the TLV, but we do have the LM4041-adj, +-1% plus resistors.

The voltage detector thing would have a much sharper turn-on than the BJT plus diode, which would make up for some of the resistor slop.

The LM4041-adj works like a PNP, as I recall, whereas the 431 looks like an NPN. That would be a bit less convenient for this circuit, though it's a lifesaver for negative supplies.

I try to use lots of 4041s and the occasional LM385, just to help keep them around. You could use it in place of the BCX71, but you'd lose a volt of gate drive in the process.

Cheers

Phil Hobbs

One other thing: That 40 mA is twice the abs max of the TI version, at least. Even if it survives, counting on it to stay in regulation in those conditions seems a bit sporty to me.

Cheers

Phil Hobbs

I can't see John's schematic well enough and don't want to do a Dropbox sign-up but personally I'd also go TL431 here. 0.1% resistors are easily available and cheap these days. Flipping the control polarity around only takes another transistor and a couple resistors. Or use a pre-biased BJT to save space and parts. I've been using those a lot lately.

The design currently in the works here needs to fit in the area of half a postage stamp and has dozens of parts. Most are the size of a salt grain. Like John used to say, 50% of an EEs job is usually packaging. Here it's more like 70%.