AW: [sdiy] Zener-oscillators

[Czech Martin]

This was the point here. The UJT is somehow strange,
completely different.


The Snap back NPN (and PNP) works with avalanche plus
normal bipolar current amplification.

The SCR with trigger works with combined normal current
amplification. If the trigger is left away, the device
will trigger with avalanche breakdown and then combined 
normal current amplification.

The DIAC has two physical implementations:
1) symmetric NPN (see NPN snap back above)
2) antiparallel SCRs without trigger pin
(see SCR).
Due to symmetry it will work in both directions.


The TRIAC is a six layer structure, basically
two antiparallel SCRs with a common trigger pin.
Should work in both directions (main path as
well as trigger path), but not completely
symmetrical.

Note: The SCR on single substrate, i.e. not the two
tranny version, will be not too usefull, because
it is hard to turn it off again without turning the current source
completely off.
The two tranny version has the possibility to introduce
additional resistance that makes turn off possible,
but the ON conduction will also suffer from that 
a little bit.

The UJT is (according to Sze) a rod or piece of semiconductor
with terminals on both ends. Slightly doped.
The terminals are called bases.
So there is always some conduction between the bases,
you can not turn this off (perhaps you can using field
effect with emitter ...)
In the midle there is a diffused emitter. 
Note that both bases meet in the middle under the emitter,
just with an ohmic junction. 
Base 2 is at the higher potential.
Base 1 at 0. If the emitter
is forward biased (e.g. p-type_, the base with the lower
potential (say base 1) will be floded with holes, thus conductiviy
will rise. 
This then leads to lower emitter-base1 voltage,
we see a snap back. Since there is no npn or pnp structure
there is no current amplification. So the UJT does have nothing
to do with a NPN/PNP or SCR. The UJT currents are therefore
much lower, compared to a fired SCR.
The device wich is similar to UJT would be the JFET,
but I gues that the channel doping in a JFET is so high
that the conductivity modulation effect can not really
be observed.


m.c.



-----Original Message-----
From: René Schmitz [mailto:uzs159 at uni-bonn.de]
Sent: Mittwoch, 2. April 2003 11:18
To: synth-diy at dropmix.xs4all.nl
Subject: Re: AW: [sdiy] Zener-oscillators



Hi Harry and all,

> The four layer (npnp) device family does what you are trying to do by
> design, rather than by accident (like transistor avalanche).  These
> would include the four layer diode (now very rare) and  (with a gate
> added), the SCR, Unijunction and Programmable Unijunction Transistor,
> and maybe the Diac (though I'm not sure of that one).

The Unijunction is not a four layer device. As its name says it has one
junction. I remember vaguely that the emitter is also heavily doped in
these, so it could indeed be the same avalanche effect that exists in
the transistor.


Cheers,
  René

-- 
uzs159 at uni-bonn.de
http://www.uni-bonn.de/~uzs159