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

Re: Solid-state TC - transformer design

On Thu, 24 Oct 1996, Tesla List wrote:
>    I'm going to use a 10 ohm 225w power resistor placed before the
> filter capacitors in the circuit to limit the rate of charge to the
> capacitors. In previous circuts I blown the bridge rectifier even though
> it was a high current, high voltage type! The filter caps are two 4500uf

No wonder. Assume you turn the power on at the mains peak voltage
moment. Then caps are at zero volts and the current is limited by
only wiring, bridge etc. losses. Current draw will be huge!

In SMPS people usually limit the current by a NTC resistor. (Someone
did call it a gizmo already I think). The idea is that it has high
resistance at the startup. When current flows through it (ie. unit
has been on for a while) it heats up and the resistance lowers to
a much lower value. Therefore losses are limited.

NTC has one drawback: if you turn the power off and capasitors
are discharged quickly. Then if you turn the power quickly on before
the NTC has had time to cool down it will not limit current maybe.
This is a thing everyone should keep in mind. In coiling it can
quite often be arranged so that the operator will let it cool down
before reapplying power.

Then there is always the good old solution: resistor in series. However,
it will be very lossy. Therefore you should trigger a thyristor/triac,
relay or something to bypass it after a while when caps are charged
already and it is no longer needed.

Someone did mention power factor correction. It certainly depends on
the type of apparatus you have. If you use small caps which will
discharge a lot during mains cycle you might not even need one.
In that case you will have input power of rectified sinuoidal and
output power rect. sin. modulated. You will hear the 50/60/100/120Hz
note on that one.

If you have large caps you probably want to keep the voltage about
constant and hear only a small "hssss.." In that case you may or
sometimes even must consider using chokes to improve power factor.
You may also have some nice switch mode power factor correction
circuits there. It will mean more to do. For a bigginner I suggest
using small caps so you will not need to think about those things.

In a first prototype even half-wave rectifying may be nice: it means
half of the power. It is however an advantage when you debug the
circuit. It will mean half less heat problems! :)

> in parallel. I have also recently acquired a surplus 120vac variac rated
> at 45amps to control the voltage. You can disregard the filter network

That is of course one possibility. Keep the voltage down at the startup
and let it build up gradually. On a prototype that is usually the
case anyway. Only a fool applies full voltage to a first test!

Someone mentioned also snubbing. It depends a lot on the circuit and
used switching devices if it is needed at all or how much. With
thyristors it is a must to limit di/dt and dv/dt. With bipolar
transistors it is a must to keep the device within SOA. With mosfets
that have a poor layout it is needed to prevent voltage spikes.

I'm right now working on one nice prototype which will not need snubbers
at all. Keeping the layout good and taking use of the parasitic
elements will do the job. Anyway, the all just wast power and I cannot
afford it with the low power rating I have in my appartment! :)
--
"If the troubles are a bit less sever (say, someone merely hit your keyboard
 with an axe .."                  - Linux System Administrators guide V1.0
Harri.Suomalainen-at-hut.fi - PGP key available by fingering haba-at-alpha.hut.fi