Largely impossible to say -- it depends on a number of factors. The consideration that sets one limit (such as thermal heating of the die causing dopant migration) may be completely different than the consideration that sets the other (such as fusing current for the bond wires). Even if the limits are both due to the same consideration, one might be "worse" in one package and the other might be worse in another package.

Which I why I would use a 1A diode, even if it only has to take 1A for a short period of time.

 

Which I why I would use a 1A diode, even if it only has to take 1A for a short period of time.

Agreed, though certainly other considerations can trump this, such as cost or footprint. I don't believe in pushing device specifications unless you have a damn good reason -- but I'm more than willing to acknowledge that damn-good reasons do exist.

 

and when things are pushed beyond manufacturer recommended limits, one better do own stress tests and not rely on anecdotal survival. diodes are low cost and plenty to choose from. there are plenty of products in small packages that will handle the current.

 

Late to the party, but . . .

Contrary to Bob's opinion, my opinion is that you were pretty close with your first two schematics. Other than component values (which is your original question), there is only one circuit edit. Fixes for the right-hand schematic:

0. For your sake and ours, re-draw it as in post #15: inputs to the left, outputs to the right, power up, ground down.

1. Keep left-side R6, move it to the right-side schematic, and increase it to 100K. This resistor is needed to assure a complete and rapid turn-off of Q11.

2. Keep R60 and decrease it to 470 ohms.

3. Keep R62 and decrease it to 1K.

4. Change Q11 to a transistor with a continuous collector current rating of at least 2 A.

Also, R63 can be deleted.

ak

 

and when things are pushed beyond manufacturer recommended limits, one better do own stress tests and not rely on anecdotal survival. diodes are low cost and plenty to choose from. there are plenty of products in small packages that will handle the current.

Agreed. Cost for something like this is usually only a damn-good reason when heading for serious volume production. If using the lower-rated diode saved $0.01 per thousand (not a penny each, but a penny per thousand) but they are going to produce ten million units a year, then they've added $100 to the bottom line each year for each one of these in the design. If that circuit is also used in multiple high-volume designs, that gets even better. Offsetting that, however, is the up-front cost of doing the tests to ensure that the performance is acceptable. So, for a design that's only going to have a few (or a few hundred) made, it's pretty much a no-brainer than the cost savings are a non-factor.

One professor I had used to work for RCA back in their heyday and he said that one of their highest-paid engineers had one job -- to reduce the cost of manufacture of the flyback transformer by five cents each year. As long as he did that, he was guaranteed to keep his position for at least another year because the additional profit to the bottom line more than paid for his salary.