[sdiy] Switching power supply transformers
Batz Goodfortune
batzman-nr at all-electric.com
Thu Jan 13 05:48:45 CET 2005
Y-ellow Karl Harry et all.
At 10:15 PM 1/12/05 -0800, harrybissell wrote:
>NO NO NO....
>
>switching power supplies run at very high frequencies...
>50/60Hz on one of those transformers is saturation / fire
>for sure
I've seen it. Lots of fun for the whole family. If your going to do that,
don't forget the marshmallows.
OTOH, Lots of synths use switching PSUs. The latest MOOG offering for one
it would seem. There are switchers and there are SWITCHERS however.
Switchers who's only job in life was/is to run a dumb terminal or other
digital-only device are probably going to be less than impressive. A
millivolt of noise or two isn't really going to be a problem. Makers of
general purpose computer supplies had to up their game however because
people wanted to run audio and other high quality mixed signal devices from
them.
You should be alright in terms of the 5 volt supply for the logic, Though
it would be prudent to observe the rules for mixed signal PSU routing
overall. That is. Separate termination for analogue and digital grounds.
Having said that, you might be surprised at what manufacturers get away
with in the real world. The Yamaha TX7 for example, has a small, and very
simple switcher. It produces 5 volts at some current for the mostly digital
circuitry. But it uses a pair of ordinary 78/79 15 volt regulators for the
analog. I'd like to say that this arrangement works well but given that
mine actually did pack up and burst into flames one day seems to suggest
otherwise. I'm sure that's not exactly an issue with you at this point
though. (Yet!)
The basic rule of thumb with transformers is. The higher the frequency, the
smaller the size. (for a given current/voltage output.) But since these
switchers are often running at RF frequencies these days it also introduces
several new advantages and some disadvantages. One of the advantages is
that at frequencies of 1 meg or more, a designer can take advantage of the
skin effect in the winding of the transformers. Smaller wires in the
windings but more windings. I don't know off hand what the calculation is
for that but effectively they double the surface area of the windings and
at higher frequencies, electrons tend to want to move around on the skin of
the wire rather than through it.
The disadvantage of course is that if badly designed, you're going to blow
away radio reception for miles. The people who can deal with these issues
and many others concerning SMPS design are better men than I Gungadin. This
is the real voodoo in electronics. There are lots of people now who can do
this stuff but at one time there were only a handfull of people in the
world who knew how to cast the spells. And they got paid big wads of cash
for using their magic. Apparently you can pretty much trace modern SMPS
design back to just 6 people.
But I digress.
Needless to say, when you enter the world of switchmodes, you're not in
Kansas anymore Toto. And hopefully before any harm is done, and just in
case it would be, I should just point out that these things can kill you.
They are dangerous to poke around inside unless you're absolutely clear
about what's going on in there. Any mains supply is dangerous, Mains
powered switchmodes are lethal.
Out of interest, the start-up loading needs only be enough to stop initial
in-rush overshoot on the output filter caps. But this depends on the
specific SMPS design as well. An ATX supply for example, needs an auxilury
switcher or linear PSU in order to provide a constant 5 volts to operate
the on/off switch. Subsequently the SMPS chip(s) themselves usually have
their own power supply. Otherwise the PSUs are what they call "Self
Starting" What that means is that the first few pulses at turn-on are full
spike. The filter caps charge up pretty quickly. This powers up the SMPS
circuitry but if there's no load, the filter caps keep on charging. The
SMPS circuitry detects that it's overshooting it's regulation target and
immediately pulls back the pulses. The cap remains floating, often
over-voltage with nothing much to pull it back into line and the chips keep
sensing an over voltage condition.
The chips don't require much current and so even though there isn't much
current actually available at the output, there's nothing but the chips to
use it. Thus every now and then the voltage will fall below the threshold
and the SMPS circuitry will fire off a few pulses. Which of course, will
immediately charge the caps up again. If there's something attached to the
supply that also uses bugger-all juice but is particularly sensitive to
over-voltage conditions (Say a 5 volt CMOS chip) then this chip will almost
certainly be toast.
The SMPS chips don't fry because they're designed to handle a wide supply
range and the transformer ratio is designed that even at full clip, it
can't spit out more voltage than the ab-max of the circuitry. In single
supply switchmodes this is not usually so much of a problem because the
feedback to the switch mode chips is monitoring the actual output. However
with computer supplies where there are also +/-12 volt rails, these rely on
the feed back provided by the 5 volt rail. The net effect is that you'll
get spurious output from your 12 volt rails as the 5 volt rail bounces
round. The 12 volt rail will bounce over and under voltage and tends to
destroy things.
Ergo; You must have sufficient load on the output but not necessarily
attaching a bar-radiator to it.
Take the 5 volts out. Filter it a bit more if you have to, but don't muck
round inside the things. And if it's not already in a safety cage of some
kind, put it in one before you use it.
Hope this helps.
Be absolutely Icebox.
_ __ _ ____ International Nihilist ____
| "_ \ | |
| |_)/ __ _| |_ ____ ALL ELECTRIC KITCHEN
| _ \ / _` | __|___ |Your source of Armageddon in a musically crass world
| |_) | (_| | |_ / /
|_,__/ \__,_|\__|/ / Disgusting-> http://all-electric.com
/ ,__ Music -----> all-electric.com-click music/download
Goodfortune |_____| Cult -----> http://www.subgenius.com
More information about the Synth-diy
mailing list