wimpy diode

An LED? or is that too much voltage?

George H.

I want about 0.8 volts, to be a little above the gate turn-on voltage of the enhancement phemt. It would be cool to see it glow, but a visible LED would be 1.8 volts or so.

I have used LEDs as the voltage reference for simple current sources, where they are just right.

When I tested a bunch of mixed unmarked diodes, one dropped 1.35v at 0.86mA. Don't know what it was, perhaps something HV.

Here are the measured Vfs at around 0.86mA. This was using an uncalibrated meter.

NT

Would one of those Vbe multiplier circuit work? (I don't know why I ask you things like that... I figure you know 100 times more electronics than I do.. the chances I'll have something to add are small.)

But even the blind pig sometimes finds an acorn :^)

George H.

Aren't 2 diodes cheaper?

NT

The right LED at the right current will have a temperature coefficient that almost cancels the Vbe of a transistor.

Plus, it glows in the dark!

Some HV diodes are actually two in series ( UF4007), and that used to be a common bias component- called a 'stabistor'. Alas, those actually had SPECIFIED limits on forward voltage, not just a worst-case maximum.

An easy solution: a transistor with C-B and B-E bias resistors.

I posted the BCX70 data. It looks perfect, using the b-e junction alone.

I doubt that a transistor with lower Vbe, and a couple of Vbe-multiplier resistors, would be any more accurate.

I was wondering about that. I suppose in theory you could use 2 diodes and the emitter resistor was also a diode! :)

NT

that's misleading editing. I was talking about use in current sources, not the former application that wanted 0.8v

I think that's equivalent to a current mirror. One diode and one transistor does that, too.

The LED thing is not a mirror; the output current is fairly insensitive to supply voltage. I think one more resistor could null that out, too.

OK, but that will have a pretty bad tempco, roughly 0.3% per degree C. The LED thing is way better. 2 diodes is OK if accuracy doesn't matter too much.

without thinking too much about I think 2 diodes will have ~2x2mV/K with the right color LED I think you can find one that is closer to ~2mV/K

-Lasse

The 1N914/1N914A is quite flaccid.

Vishay makes the TZMC1V0. The datasheet is extremely short but it looks like a wimpy diode.

But, that (if I read the B-E versus (B+C)-E data correctly) depends on Rbb. It's not a well-controlled parameter. The BCX70 is made by Infineon, Fairchild, NXP, Rohm; it doesn't have to always be the same as you have measured.

Well, we have 2600 in stock, probably off a reel of 3000. I'm just trying to turn on (and off!) an Avago ATF-50189, which isn't terribly picky. I'd expect anybody's BCX70J to be pretty similar; they must all have similar geometry and doping, to meet the high/tight beta specs.

Yeah, that's the l

That 90 PPM tempco could be all in the emitter resistor. This was good enough, but could certainly be tweaked.

Bizarre. They call it a 0.75 volt zener diode, but it sure looks like a regular PN diode, used forward biased.