|
Hi Fucian (?),
I've found that calculations, in your case,almost
never agree with actual measurements.
Hidden (i.e., real-world) variables like turns
spacing, capacitor ESR and inductance, wiring inductance (conductor spacingand
current loop area) and switch characteristics each modulate the peak
current. Most of the hidden variables, though, are inside your
capacitor.
I have personally found that physically measuring
the experiment is always more reliable (and sometimes faster) than using
math. Even a ballpark calculation can be a liability with high current
discharges, under some conditions.
To measure your setup, begin by using no SCR.
Terminate the connectors that would otherwise go to an SCR with 2 heavy
wires or sections of copper tubing that you can bang together, resulting
in a spark that discharges the capacitor through your
coil.
In series with the coil, splice
in an 0.1-ohm, 10-watt wirewound resistor. This will let you measure the
current. (You can use a noninductive resistor if you can find one, but herea
normal, inductive resistor will cause only a slight and negligible
error.)
Connect a 1-amp, standard silicon diode and a
220-470 uF, 35V capacitor (which can be from the junkbox) in series with each
other, then place this "chain" in parallel across the 0.1-ohm resistor.
Polarize the diode so it will conduct, charging the capacitor, when there
is current (a voltage drop) across the resistor.
For every 10 amps the coil carries, there will be
1 volt of drop across the 0.1-ohm resistor.
If your coil carries 100 amps peak, for instance
the capacitor will charge to about 9.4 volts.
It misses 0.6 volts since the diode "drops"that
much potential during conduction.
To test, just bang the open wires together HARD
(you can smash them together with a mallet; the faster they join the more
accurate your test will be) and quickly measure the voltage on the little cap
with a digital voltmeter. (It needs to have high input resistance, so it
doesn't drain the capacitor while trying to measure it.)
Take the voltage reading, and add 1 volt toit
(really all you need to add is 0.6 volt, but adding 1.0 volt gives you a
reasonable error margin).
(It is wise, though to measure the resistance of
the small capacitor with an ohm-meter before you use it. It should end up in the
meg-ohms or off the scale on the high end. Anything lower means leakage, and
resulting error in what you measure - like measuring a really leaky
balloon)
This voltage reading + 1 volt is your approximate
peak current, divided by ten.
So, take your measured value, add one, then
multiply by ten and you have the peak current.
The SCR you use should be able to handle atLEAST
75% of this current value.
Because you are using fast current pulses, you can
use an SCR that is rated for somewhat less current than the peak value,
if you have to. SCRs fuse, in the overcurrent condition, from hotspots in the
silicon. Fast discharges, like yours here, are done before full heating can
occur - so you can "cheat" with lesser values than what would seem like
enough.
Having said this, I don't recommend it. Avoid
downtime and frustration - spend the extra $3 on that SCR with the full rating
:)
I write this assuming you have a lab that has some
parts, and a DVM.
If you don't, just use a 100A, 600V SCR. I
can about guarantee it will not blow out - if you do one thing that most
people overlook.
Of course, when you fire an SCR through an
inductive coil (like yours here), current flows through it, causing heat and a
magnetic field around the coil.
However, by the time the capacitor is discharged
(at zero volts), there is often still a magnetic field around the coil, still in
the process of collapsing.
This fading magnetic field, because it is
changing, causes current to continue flowing through the coil. The SCRwill
not and cannot conduct this current anywhere - the polarity is wrong
and the SCR's off, anyway. Because the current is thus trapped, it will often
build up to a high voltage (like the ignition coil in a car, releasing its
spark) and blow out the SCR.
Very large, expensive SCRs can be easily destroyed
with relatively modest coils and capacitors, if you don't add a diode
across the coil to short out this trapped current that still flows
after the cap is discharged.
Just solder in a diode across the coil, somewhere
in parallel with it.
Polarize the diode so that it will not conduct
when the capacitor/SCR are "on".
It will appear to be in "backwards" in the
circuit, as if it would seemingly never have a chance to conduct.
Use a diode with about 1/8 the current rating, but
the same voltage rating, as the SCR. More current-rating is always better,
but SCR/8 (or thereabouts) is almost always enough to do the
trick.
If you use a 15-amp, 600 volt diode across your
coil and in "backwards" you can't go wrong.
This little piece of advice here can save you a
lot of frustration and money. Too many people forget this "back-diode" and learn
the hard, expensive way. (Yes, I am one of them!)
By the way, this FEMM list is primarily here for
discussions on the FEMM software and related mathematics. FEMM, however, can
not model your question (it's more of a SPICE simulator thing).
You might find that you have more luck asking
questions like this on other email groups that deal with practical experiments.
For starters, try jlnlabs@xxxxxxxxxxxxxxx - other
members there also deal with the kind of circuits that you are
using!
(If you want to try the list, go to http://groups.yahoo.com/group/jlnlabs to
subscribe. Or, you can check out the list owner's website at http://members.aol.com/jnaudin509/index.htm,
but beware the hideous MIDI music that plays when you go there ;)
Regards,
Graham G.
PS -
A good surplus retailer that carries large,high
current SCRs has a URL at:
|