Re: Solid State coils.
Subject: Re: Solid State coils.
From: "Malcolm Watts" <MALCOLM-at-directorate.wnp.ac.nz>
Date: Tue, 23 Apr 1996 08:59:31 +1200
>Received: from rata.vuw.ac.nz (root-at-rata.vuw.ac.nz [18.104.22.168]) by uucp-1.csn-dot-net (8.6.12/8.6.12) with ESMTP id OAA08374 for <tesla-at-grendel.objinc-dot-com>; Mon, 22 Apr 1996 14:59:31 -0600
In response to Alan Sharp's queries...
> My present circuit is essentially a big flyback transformer. Set up
> a current in the primary, switch it off and in theory the energy has
> no where to go but into the secondary. In practice, very high
> voltages appear in the primary, with unfortunate effects on the
> electronics!!! It runs better clamped but it looses a lot of
In fact, the core energy will go anywhere it is able to. The voltage
that appears across the primary will be also appearing across the
secondary multiplied by the pri-sec turns ratio. In a switch-mode
power supply, this is controlled by use of an energy recovery winding
clamping all outputs to the supply rail x their respective turns
> I may in time try a full bridge but I havent tried driving high
> side FETs.
I would use either a half or full bridge without hesitation in this
app as the primary is always clamped by either the switches or
> Could an electronics wizard tell me if Ive got all this right:
> 1. Abandon the strip board for a double sided PCB. One side devoted
> to ground (ground plane). Power connections at least 1cm wide, hot
> shoe the FETs. Short path to gates, snubbers as close to FETs as
> possible. Ill get some Ferric Chloride and a paint brush and some
> gloss paint. (Ill get into photoresist some day). I remember in
> the physics faculty etching PCBs in seconds using a concoction of
> hydrogen peroxide and conc hydrochloric acid - wicked!
Exactly the etch I use. But a word of warning - using conc. solutions
can lead to copious chlorine production. Better to dilute a bit. It
gives a much more controlled and accurate etch I find.
Won't claim to be a wizard, but : I don't think I'd use a ground
plane. The multiple current paths are going to couple signals all
through the circuitry. I think a star connection for all the high
current paths is a better option. I have seen all sorts of trouble
in MF radio transmitters caused by ground planes. You get a plethora
of parasitic inductances and capacitances. Just my experience.
All connections MUST be as short as possible. You may find you
need resistors in series with the FET gates to damp responses caused
by track inductances and the gate and Cdg capacitances.
> 2. 2 IRF740s on each side, mounted on a 2 c/w heatsink have
> handled 2A -at- 150v, 200kHz continuous.Ill use 4 to allow more power
I'd definitely go for a lower frequency if possible to minimize
the time the FETs are in transition. You should get less heating
and better power handling. Also, get the transition times as swift
as possible. I know that small inductances and capacitances respond
to the high frequency content of fast transition times, but minimizing
these should make for a reasonably small amount of snubbing with
> 3. Transformer design.
> I have the formula, giving the minimum number of primary turns, in
> terms of voltage*period, and B max flux density. B max is 320mT but
> this reduces for push pull and further reduces because of core loss
> to about 32mT. I guess that I want at least a 10:1 step up? Build
> carefully lots of insulation between layers. I was considering
> lacquering each layer as well. Any tips?
You'll find that you get far more out of a given core when using
push-pull because you are swinging the core in both directions.
Single ended stuff only works on one side of the B-H curve and
usually works well up on the curve. In such a situation, controlling
maximum switch current is the only way to keep the core from
saturating. The problem with this in pushpull is different - keeping
the core magnetization equal in both quadrants.
The easiest way to view a transformer is to imagine the primary
inductance being in parallel with whatever load is applied to the
secondary (transformed by the turns ratio). So you can control
core flux by tailoring the primary inductance. Usually, the practical
limitation on primary inductance is the amount of wire you need to
fit on the core (turns vs wire current handling capability for both
primary and secondary). If you do use a transformer in a pushpull
circuit (bridge or 1/2 bridge), I'd suggest applying current limit
to the switches in each half to prevent core saturation. If not,
ensure that the waveforms can't let the core flux get imbalanced.
This can be veryy difficult to achieve without some kind of current
monitoring. An airgap in the centre-leg of the core can help, but
adds enormously to the leakage inductance and won't allow perfect
transformer action. By the same token, an air-cored transformer
suffers this disadvantage.
> 4. New secondary.
No comment, except to say that I think the bottom line of the formula
should be : 9r + 10h
I hope this helps. I really think an air-cored transformer for the
driver is a better option as power handling is only limited by the
wire size used (you can use it as is for operation at any power).
Also, you won't have to worry about core saturation caused by
imbalances in driving.
Finally, the driver step-up ratio is going to be ideal for only
one operating power level with a particular secondary. I am unable to
quantify this at the moment but am working on it.