[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[TCML] Feedback drive with a CT for a linear (non-DR) SSTC
Hello everyone, I'm Ivan, a russian coiler. After some research I got some
questions on the feedback driver design. I'm new here so sorry for any
stupidity involved in this post, will improve next time :)
Now to the topic itself. I have been working on a linear (eg. non-DR) SSTC
recently and got some positive results on 494-based (HSMC, not TL version
as it has a shorter deadtime) system, namely 40cm spark at up to 60bps @
50% d/c cycle 170vdc feed via a full-wave smooth doubler from 60vdc
xformer, drawing 30A max off the 60V line itself). The system itself was a
494 PWM driving a decoupling xformer (sort of a GDT, but instead of
driving the gates it was sending signal to the output UCCs that were
connected to the irfp260 mosfet gates directly via 2.3 Ohm resistors) with
a couple of 37322s, thus implementing the deadtime as well (you can watch
it on YouTube on my channel btw). However an attempt to move onto the
off-line 220vac feed failed - the system worked for some time giving off
60+cm sparks using irg4pc50w IGBTs as the bridge switches and then the
IGBTs blew along with the protecting fuse (funny or not - only the lower
pair failed, I was really surprized to see it's not a diagonal or
single-sided failure). As it was clear that unipolar drive is not that
good for IGBT drive and such a fast dv/dt, I designed an optocoupled
driver supplying +12v at the high and -8v at the low state. So, the first
question is - is the 6n137 optocoupler (Fairchild) suitable for that? As
it seems, it's CMR is enough for this voltage rise speed (the bridge
rigged up to a resistive load didn't show any spikes on the scope trace
picked up from a shunt attached between the film caps and the bridge
itself), but I'm still unsure if it will work that well in a coil system
running in (hopefully :) ) near-ZCS.
The second question is about the oscillator. I tried 494 on the optos as
well, but it gave no good results as the deadtime without the additional
mess added by the decoupling transformer and with the additional turn-on
time due to the gate going negative while low is once too high. So, having
to change once too much in the system, I decided to move onto a
feedback-based drive, which, if properly protected, will allow a better
switching mode and help to obtain even longer sparks due to the ability to
get the frequency that low that on an open-loop oscillator the coil
wouldn't even be able start to break out on this amount of undertune - and
still with no risk of sync loss. I designed a feedback driver that picks
up the signal with a hc14 gate, uses the hc14 and hc08 gates to get two
pulses with a manually set deadtime and, picking up the deadtime pulses
and using them to charge a frequency threshold capacitor, can
automatically be disabled if the system had lost the feedback signal for
one or another reason or just the operator put something to the coil so
it's Fres went dangerousely low. The lab tests went well, so it seems it's
time to connect this one to the bridge (decided to use the IXYS 30N50 fets
there btw), but I've got the one last problem - the feedback itself. I
thought of using a CT rigged up to the secondary ground, but the
calculations proved that it's not as clear as it seems. Normally the
secondary current is less then 3 Amps, so the CT ratio should be pretty
high in order to get it properly loaded the whole secondary duty cycle by
means of the driver's input zeners or diode fork. If the correct load is
not present, the voltage output fase of the CT will get displaced so will
get fased more like the input voltage, not current - eg. up to 90 deg
shift. Should I wind a 200 turn CT (I plan to load it with 15v zeners and
then send the signal into the system itself through a 1k resistor, a 100n
cap and a diode fork to normalize it to 5v for the logic to accept), or
it's not necessary and even 100 turns will be ok (the secondary is a
150kHz 37x18cm one)? Or I should use no CT at all and just attach an
antenna feedback? I'm not really happy about the antenna idea as it limits
the coil positioning and adds some risk of arcover..
Thanks in advance!
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla