2N3904 maximum ratings Vceo Vcbo

Why not just use a snubber diode across the inductor and be sure?

As long as the base is not floating (the Vceo case) you'll be OK.

The base-grounded case is sometimes denoted Vcer (b-e resistor) or Vcex (shorted), and is about equal to Vcbo.

Keep the base resistor low, like 1K or something.

BVcbo = Collector-to-base breakdown, emitter open BVceo = Collector-to-emitter breakdown, base open BVcer = Collector-to-emitter breakdown, base-to-emitter resistor BVces = Collector-to-emitter breakdown, base-to-emitter shorted BVcex = Collector-to-emitter breakdown, base-to-emitter reverse biased

etc

See..

...Jim Thompson

On Sat, 2 May 2015 19:45:51 -0700, "Bill Bowden" Gave us:

Absolutely presents a problem...

This will let the smoke out. No question.

That's only 8 volts above the Vceo spec. No problem.

Depends on the load-line... a snubber of some sort might be necessary. ...Jim Thompson

Not the classic load line, but if there's an unclamped inductive load, the flyback could be hundreds of volts.

I thought the specs indicated "guaranteed" performance? Vceo of 40 would imply every device would meet that spec plus a little more? It's a resistive load, so there aren't any inductive considerations. And the base will be grounded to the emitter through a 3K resistor when the collector voltage is

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

I would expect the answer to this question would depend on the date code. A part from the 1980s made by diffusion would probably have a very wide ran ge of propeerties, both within a lot and lot-to-lot. Something made in the 21st century by ion implantation might have a very tight range, and could be manufactured with very little "safety margin".

Why do you raise the question? Is it academic curiosity, or do you havee a specific application and lots of 2N3904s in the warehouse? The 2N3904 is cheap (about 10 cents is smalll hobbyist quantities), so a part with a suff iciently higher BV is probably less than 20 cents in large manufacturing qu antities. You would have to sell a lot of widgets to recover the additiona l engineering cost of using an underspec'd part.

The collector-base leakage rises steeply with V_CB. With the base left open, it gets multiplied by beta, so I_C goes through the roof at a lower voltage. If you prevent the B-E junction from getting forward-biased, e.g. by shorting B-E, it holds up out to V_CES, which is higher. A resistor isn't quite as good, and reverse bias is a bit better, so you have a family of breakdown voltages rather than just one.

V_CEO < V_CER < V_CES < V_CEX .

Cheers

Phil Hobbs

Toss in a couple of 1N4001s (or surface-mount equal) in series with the fan connector and get on down the road.

Yes, I usually write software for a living... why are you looking at me funny?

Pic somewhat related:

Matt Roberds

But the original board, rev A, first try, works with no kluges. Hacking it would be a sin.

I think that about my code when all the modules are still sitting around version 0.x. Usually, by the time it ships, some modules are up to 1.5 or more. The beauty is still there, but you have to look past the cruft to see it.

Stick the diodes in the fan harness?

Matt Roberds

Couldn't you get Avnet to change their colour scheme? It clashes horribly. ;)

Cheers

Phil Hobbs

Think so? I kind of like it. But then, I have a red car and I'm typing in a red t-shirt.

Code can be, and often is, edited and recompiled scores, maybe hundreds of times, before people decide that it's good enough, or that it has to ship. Code doesn't have pads that fall off after a dozen reworks. A PC board can't be edited and re-etched in 10 minutes; iterations take more like a month, and are messy and expensive and publically embarassing. Electronic design has to be done right, and brutally checked, at the engineering level, before the Gerbers are formally released and boards ordered and assembled and tested.

In general, the easier it is to change something, the less care will go into its design, and the more it will get changed. And the more bugs will never get fixed.

We could, a 3 volt zener or a resistor in a bit of shrink tubing. Or just run a 12 volt fan at 15 volts! That looks like it will work.

Our FPGA surface temp is over 100C, and the box quits working at 80 to

That red uZed board has a couple of holes for mounting a tiny fan above the FPGA. We may do that, as opposed to trying to cool the entire box.

This rig lets us run up to 90C ambient. The guy who tested it wimped out there; I'd have cranked up the temp until something failed.

There is an on-chip temp sensor in the ZYNQ, but we'd need an FPGA compile and some ARM code to access it, which isn't in the budget just now. Maybe next iteration.