inverse beta

That last one ignores Re, which can really matter.

A BCX70J has minimum Vce (at zero collector current) of about 4 mV at Ib around 400 uA. At higher or lower base currents, collector voltage goes up. It is admittedly a smallish transistor.

A transistor switch operating in inverted (over-driven emitter-follower) mode can have positive or negative or zero Vce drop, depending on how you tweak the base current.

I built a bunch of 16-bit DACs once, for some military Strangelovian war-room display, with NPN-PNP complementary followers driving an R-2R network of selected wirewound resistors. Hey, I was just a kid. We trimmed the base currents on the first few MSBs to tune the switch offsets to zero. We verified the outputs with one of those old Fluke differential voltmeters, which actually worked pretty well; the meter showed the error directly.

I don't trust those equations, or Spice, so I measure stuff.

I have a few power supplies that look like that--the old HP 61xx ones with the thumbwheels. They're still pretty accurate--mostly within 0.1%.

Cheers

Phil Hobbs

Tests.JPG

chopper apps drove the emitter/base current ratio to less than one by some integer multiple. Looks like the JFETs replaced transistors for chopper ap plications and even they are being phased out now. Here's some first order modeling:

Right- you can't go adding resistances to those transcendental equations wi thout making them intractable for punching out out a quick estimate on a ha ndheld calculator. The RE is the inverted RC so its voltage drop works agai nst obtaining a 0V VEC. But you're only dealing with a few hundred uA max s o that helps.

The equations are good for preventing going on a fool's errand and they do provide some guidance on things you can do to make the circuit insensitive to parameter variations, unknowns, and with bipolar especially, temperature dependence issues, which I suppose for a first cut 1%/oC beta sensitivity will do.

Nice; also rather decent documentation. Thanks.

There's a good whiteboard just to the right of my workbench. I scribble notes as I work on things, and have a nice setup for photographing the board. If the results are relevant to a part that we stock, I put the pic into the related "Pdata" folder for that part, in our inventory database. Or if it's not a stocked part, I stick it in some project folder. That often turns out to be useful.

In that last case, I could probably have increased Ib until the saturation voltage hit zero.

Once i had the fantasy that the saturation voltage would go negative. Also that a suitably biased transistor could be used as an electron gun in a CRT.

If biased hard enough to be thermionic...? :)

piglet

Oh, it does in that "inverted" switch configuration.

There have been articles about semiconductor diodes that launch electrons into free space, the application being semiconductor-tube hybrids. Those would be dynamite for RF amps. Replace the b-c silicon drift region of a transistor with a chunk of vacuum.

Similar things were tried with field-emission cathodes, microtips and nanotubes. Obviously, none of those things have worked very well so far.

E-B avalanche. It's an old Pease(?) riddle. Try it! :)

Tim

His explanation, photoemission and photoelectrics, was conjectural. I don't know if anyone has verified that.

I think there is some sort of venturi effect in PHEMTS that can make a negative voltage.

Tests.JPG

chopper apps drove the emitter/base current ratio to less than one by some integer multiple. Looks like the JFETs replaced transistors for chopper ap plications and even they are being phased out now. Here's some first order modeling:

Is emitter follower mode popular for a shunt switch application? You can't get a negative VCE in common emitter saturated operation, inverted or forwa rd mode, the collector would have to source current into its external circu it. I can see that happening for emitter follower, common collector, but ca n't see how that configuration shorts a signal channel.

George Herold and I had fun exploring this in the thread "A very silly circuit" from october 2015 and concluded that at least for garden variety silicon bipolars the Pease conumdrum effect is photo-electric.

piglet

Yup. If you put a couple of volts D-S on an ATF38143 depletion pHEMT, it'll bias its own gate a few hundred mV negative.

Cheers

Phil Hobbs

The negative voltage can be from a photovoltaic effect. The pn junction bias emits photons and the gate absorbs the light.

A pHEMT is more like a FET, so I doubt that's it, but I suppose it's possible. Potentially useful if it happens reliably.

Cheers

Phil Hobbs

The generated gate voltage is in the turn-off direction. Don't solar cells and photodiodes make voltage in the turn-on polarity?

Yup. However there are complicated heterojunction things going on in there that I don't understand, so I wouldn't rule it out altogether. I don't have another candidate for the physics, myself.

Cheers

Phil Hobbs

I'll see your 16 and raise you 160. I mentioned the Zetex low-saturation-voltage BJTs are good candidates. ZTX618 has inverse beta of 160 at -2 volts, Ib = 1mA. The negative inverted Vce(sat) = -3.8 mV, pretty good, 7x better than the BCX70.

Other candidate parts (see our Table 8.1, which has lots of low rbb' parts, with measurements) are the ZTX851, inverse beta = 100, and Vsat = -7.6 mV. The ZTX450, inverse beta = 40, Vsat = -16 mV, not as good.

Coincidentally, vacuum tubes do the same thing, to the tune of -1 to -3V. It's not motivated by plate current though, just the equilibrium space charge potential (out on the tail of the Maxwell-Boltzmann distribution, since grid leakage is much less than cathode emission, hence |Vgrid| >> Vth).

A 6AL5 works better than any silicon diode, in the sub-1mA range or thereabouts. Partly because its zero intercept is negative. :^)

Tim