OT: SOT-523 PNP Question

Reform the leads of some MMBT2907's. It will take a lot of labor.

Why did he build 5600 boards without testing a first article? He deserves a nervous breakdown.

I know. But he's a PhD, so what can I say ?>:-} He says they worked when tested, but I suspect they "worked" long enough to fry the device... C-E interchange with 9V across it :-( ...Jim Thompson

It may have worked, with modest reverse beta. Some PNPs have high b-e breakdown voltages, 12 volts maybe.

He might just find a transistor that works that way. But not by Tuesday.

Why not cut and jumper? Make him do it himself, all 5600 boards.

MMBT3906 datasheet says VEBOmax = 5V. I have the client running some breakdown tests for me as soon as he gets back to his Hong Kong offices in the morning.

It may happen, though I have an add-a-component fix that is easier than R&R... just slobber on top >:-} ...Jim Thompson

Most SOT-523 PNPs have the same pinout. Without knowing the circuit, could a P-channel mosfet, or P-channel JFET be used that may have the desired pinout?

The SOT-523 is quite a small device. I'd hate to have to do any mod work on it!

Ph.D's love to learn things. This will certainly be a great lesson.

boB

See attached drawing. Drawing is looking at bottom of 3906; red is rotated position. Note 2 leads can be soldered on board and the 3rd fiddled with added wire. Not the best "fit" in the world and highly dependent on layout and PCB spacing.

Alternate solution: make a 2-sided flex (or similar) small PCB that does the scramble. At that quantity, the scramble patch board could be pick&placed with glue and reflowed, then the 2N3906 pick&placed for 2nd reflow. Neater results, if part layout is tight, this may be the only way.

Charge extra for the solution that you recommend.

Are the boards already assembled?

Yep :-( But it's not going to be too bad... I've found several alternate solutions that eliminate the PNP... and the labor is on mainland China. ...Jim Thompson

I've solved that problem on prototype boards by straightening the part leads then mounting it upside down with blobs of solder. That's probably not a good solution for 5600 boards. I'm not sure that Robert's solution isn't better.

How big of a hole does he need to bury 5600 boards?

As far as i have seen, ALL of the "specs" for bipolars are 5 volts, and that clearly is a lie. Most have a typical breakdown in the 8V region,some are higher.

Do share the fix when it's final.

I know, that why I asked the client to use a real scope and get me some measurements. ...Jim Thompson

Replaced a switchable current mirror (overkill) with a pull-up resistor O:-) ...Jim Thompson

Who designed the original circuit to start with?

Jamie

BUT... upside down only interchanges pins 1 & 2; the callout you gave has pins 2 & 3 interchanged.

Too bad the parts are in place - would have to be removed before my flex PCB "patch" was placed..

Like i said, upside down does not interchange pins 2 & 3.

Who else... the PhD, natch!

Too many people fall in love with current mirrors and think that they can make good ones from discrete transistors. We even had an example of that here, not too long ago.