design help

Nice. Just what my direct experience tells me as well.

I tried the same thing was I was about 10. I figured I'd out fox PolyPaks by making my own transistor from diodes, saving $$$.

Even better--'tunnel' diodes and more...

Cheers, James Arthur

Neat. Solid-state electronics could have got going 100 years ago.

John

I've seen the semi-opposite too: drive the coil with full voltage to pull the relay in, then change to a switching drive mode to keep the relay pulled in.

Saves power beaucoup, since the relay's holding current is

Not the same thing as a dual emmiter transistor at all. Oh, wait, that is not the issue here, it is just a typo.

Neither is the device, as drawn. The BAV170 is as close to the device drawn as any.

A bit more than that, but basically.

Some of the early manufactured transistors were made in a very similar (point contact) way. So yes, it is possible. That might be a fun high school science project nowdays.

my measurements suggest that it does, but it also caused the NC contacts to bounce much more

with a 1K resistor in series with the diode I see the NO contacts opening about 3 ms earlier and the NC contacts closing (very briefly) about 0.8ms later. and continuing to bounce for about 6ms

with just the diode the contacts open (3ms) later, the NC close briefly about 1ms later but only bouncees for about 2.5ms

so the time from the transistor switching off to the final closure of the NO contacts is about the same each way,

I only tested 1 relay.

given the dramatic reduction in bouncing I'll continue to use flyback diodes especially where the NC contacts are being used,

depends what you use instead of the diode.

Living in Arizona is the first sign...

Don't skimp on the shaker table testing if you are thinking about going this way.

Best regards, Spehro Pefhany

Only if you are totally incompetent at math.

The drain of a mosfet is not a "stiff voltage source." It's more like a current source.

All those guys used me as a role model? How flattering.

Folly? Get a life. Take a walk or something.

problems.

So complexity, the opposite of "a paucity of parts", leads to bugs. And makes it hard for a beginner to build.

Then quit jumping on my circuits and be helpful. Or don't. You don't matter, you know, because you aren't making sense.

That statement makes no sense at all, even if it were true. You're just flinging insults without reason. An engineer can't think straight when he's angry, which is why good engineers are calm.

But Authority? By saying out front that I don't understand something? That's some weird definition of "authority." I suppose my true sin is doubting your absolute wisdom and your incoherent, ancient Japanese appnote.

John

Sage advice.

Cheers, James Arthur

You're the one deficient at math.

Indeed. Most of Linux's ISRs are in C; Linux' use of assembler is mainly for stuff which is internal to the CPU, where you need to use specialised instructions which don't have C equivalents.

The other common case which C can't handle is multi-precision arithmetic, where you need to get at the carry flag.

Even so, it helps if you learned assembler first. If you start out with mathematical theory and high-level languages then migrate down to C, you're likely to have a hard time understanding the task at hand.

In particular, if you're writing a device driver in C, and it doesn't work for a non-obvious reason, the next step is normally to compile to assembler and see what's *really* going on.

Also, optimised code can often only be debugged at the assembler level. You can't debug it at the source level as the source code bears almost no resemblance to the object code.

On a sunny day (Sun, 01 Mar 2009 00:14:21 GMT) it happened James Arthur wrote in :

I am very impressed, beautiful stuff. Need to try this one day :-) And, it seems he has some RF power too in that second link.