Small SCRs with 1mA or less in holding current?

Yeah, but that's another resistor and, worse yet, a zener on top of that. S'pose you could use the kind of FET with gate protection zeners (assuming they're rated for DC), but that's still a resistor, which puts it on par with the BJT version..

Tim

te:

rote:

rom 0201

cost

ors in

ler

mm

, which

uess you

t an

g quite

ten get

ecide they

is

can't just use a regulator with enable?

-Lasse

which

they

There is also a "new and improved" version with tubes. A German university is still running one, at least until the last service guy passes away I guess.

Yup. A big one.

which

they

Not really. They all have an UVLO so that could be used but there is also some other stuff connected. We want to keep this under uC control. Of course, there would be other ways as well such as non-volatile digital potmeters but that starts to eat into the available real estate.

The other issue is that switcher chips do not quite drop to zero power in UVLO shutdown and I've got half a dozen.

Now you're just being silly.

Best regards, Spehro Pefhany

I see you referenced a Panasonic resistor array. Has anyone noticed that Panasonic resistors are more susceptible than normal to the end-caps falling off when de-soldering. I never seemed to have that problem before and now it seems I can't reuse any of them.

I ask again, what are the resistors for? Please give us a schematic and approximate values and reasons for each and every resistor. Perhaps then you will discover the original SCR equivalent...no resistors!

this:

Yea, but a quick (but admittedly incomplete) survey shows all of those have 50V breakdown at best.

Given the specs on a typical transistor data sheet, predict the worst-case leakage current (it will be at high temperature where both leakage and beta are high) and whether it will turn on by itself.

If there's a series gate resistor to a CMOS gate you only need one resistor (BE on the PNP) and maybe none if you can be sure the NPN has low leakage and the PNP has lousy gain at low current.

Nice thing about SCRs made from BJTs is that you can make an NPNP one as easily as a PNPN type.

Best regards, Spehro Pefhany

Maybe this explains it:

I cannot rely on undocumented device parameters on this one. But as I mentioned, I now have a solution sans SCR. Still would be nice to know if there are very low current ones somewhere.

What I meant is that you cannot place two 0402 side by side within a 1mm by 1mm real estate. With arrays, yes, you can get there but IME then I run into the same bottleneck as usual: Contract mfg balking because this is more difficult than regular 0402 reflow soldering.

How is a 0404 any harder to place than an 0402? The placement accuracy is the same and an 0404 should be easier to pick up.

I doubt that. If they can't handle 0.65mm pitch, they're lower-end than any assembly house I've dealt with this millenium.

Seems more likely to me that they're worried about tombstoning on smaller parts. I've even seen that on 0402s (about 0.05% of units required rework in one bad production run). Arrays would be better than 0402s, not worse. 0201s would definitely be worse, hence the "rule".

Here's some good info on the issue:

Best regards, Spehro Pefhany

It's not the placement, it is the reflow process. That is more difficult because of the finer pitch in the array. It can be done, as evidenced in cell phones. But many lower qty contract mfgs don't like it or won't take such a job.

Oh, I could probably cajole them into it somehow :-)

Not on this job, but on others my experience was that domestic places are often a lot more conservative in terms of what they will take on. Versus places in Asia. Then of course the Asian places usually won't touch anything that's less than 10k boards/month.

"Stonehenging", cool. Reminds me of when the Griswolds wrere there:

Cell phones use far smaller devices than 0404s (more like 01005s). I still don't believe that anyone who can do 0402s, can't do 0404s. Nope, not buying that one.

Ahh..That means the "biased transistor" cannot be used to emulate that particular circuit, because the resistors are in series with the base. Those 1K resistors are not needed, at best they might increase the holding current; would have to be tweaked for a given transistor set (due to betas). Both certainly are not needed to achieve hold current tweaking; only the NPN E-B resistor would be needed.

Tim Williams a écrit :

I was of course thinking of the small ones with integrated gate zeners...

I once got caught with those, designing a resonant driving for a small class E amplifier. The gate couldn't go bellow ground... Oops

Some gate clamped MOSFETS exist with symmetrical clamp voltage.

D'oh!

Hmm, most of the time I see symmetrical, or none. Also weird!

Tim