[Prev][Next][Index][Thread]

[Fwd: Re: SCR's]



-- BEGIN included message





Hi all,
         A couple of additional comments might be in order:

> 1. SCRs are not a resistive device - because they are a
> semiconductor, they do have a pretty constant voltage drop across
> them (as I recall, 2 diode drops when conducting, e.g. 1.4 volts).

That characteristic approximates a one-way spark gap. That means that 
the device "resistance" is actually a function of current and not 
fixed.
    The peak current rating is important but even more important is 
the rate at which the current builds up. If that rate is too great it 
will melt the die as it doesn't conduct over the whole die area 
instantaneously. This is the dI/dt rating I spoke of.

> Another comment someone made was that driving a tank circuit
> through an SCR would back bias the electrolytics used for current
> source - the SCR will, of course, turn off as soon as the current
> goes through 0, so this will not happen. 

That poses a problem: if k in the coil is significantly less than 1, 
the primary cannot oscillate. If it can't do that, only a fraction of 
the cap energy will get to the secondary before the device switches 
off and the cap will be left sitting with a voltage reversal on it 
(if it could stand it that is). Remember, current in the primary 
can't stop flowing until the magnetic field has finished collapsing 
back in on itself. Voltage reversals in half a cycle of oscillation in 
the primary can be over 90% for losse coupling. A reversal of 70% 
still occurs when fully one half of the cap energy remains/has gone.

> I don't have any SCR reference material handy, but I seem to recall that even 
> farily small SCRs can handle several thousand amps (peak), and that the large 
> ones peak over 10,000 amps - If my memory is correct, they should be ideal to 
> simulate a spark gap.
> 
> However, at low voltages, you're really limiting the peak amperage.  I don't
> know what the on-resistance of a SCR is, but let's assume 10 ohms.

It is *current dependent*, not constant, and definitely not linear.

> At 100v,
> that's only a maximum of 10A it can conduct.  In addition, at the high
> frequencies that a TC operates at, all conductors only conduct on the
> outermost few mils of the material (the skin effect), which increases the
> resistance of all conductors several orders of magnitude over the normal DC
> resistance.
> 
> Hope this helps,
> Chris C.

Malcolm






-- END included message