all your lovin' and all your switchmodes too:
I know it's been a while but I've been busy. Well.
That's when I haven't been comatosed.
bought a new oscilloscope with many thanks to Eric
for facilitating that one. Now I can actually see
inside things again. I can't thank Eric enough for
the tireless effort he put in and the donation of
probes and other wonderful bits he's given me to make
the whole thing a going concern.
I've been delving the mysteries of switch mode power
supplies and PWM (Pulse width Modulation) in general.
And before I go any further on that, if anyone would
like to donate their old computer power supplies,
working or not, I'd be most grateful. In the past
few months these things have gone from a mysterious
grey box to a vast wealth of fascination.
that end there are two new projects (quasi projects
for they do not have PCBs with them.) Published in
[Bread and Circuits.] One
is an bench audio test set and the other is a PWM/Switchmode
mini-drill speed controller. Both built almost entirely
from junk. As recycled as possible.
WE DRILL HOLES:
PWM drill speed controller is a dual channel device
based on a linear pre-regulated supply but a switch-mode
speed controller. The reason is that PWM allows for
much higher torque at low speeds. This box also controls
a bunch of bench fans with a view to extracting soldering
fumes out of my face. If you ever wanted to investigate
SMPS technology, this might be a good place to start
and a lot of my research is noted.
it's designed to be built out of what ever you have
laying round (apart from the SMPS chips themselves)
there is no PCB layout. I used some darlingtons with
a pretty strange package arrangement. You might chose
differently. But if there's enough call for it, I
might just consider doing a PCB layout for it. But
I figure it's more of an idea thing that might spawn
the imaginative process. And of course, you don't
have to control just drills with it. You could quite
easily use it as a DC lamp dimmer or other motor controller.
it's current form it can deliver up to about 3 amps
per channel depending on the base PSU. But you can
adjust that to suit and also the max/min voltage range
delivered to the device. The chips used are quite
common SMPS chips and there's nothing particularly
tricky about them. Except that I use them in a somewhat
unconventional way. So this may be of interest in
a lot more research went into these things than one
would normally wish to spend on a hole drilling implement.
It's predecessor wasn't much more than a pot and a
rectifier. But I can say that these little DC drill
now have one hell of a kick in them using this system.
Even at low speeds. And I can even grind and mill
with one of them. It was worth the effort.
Then we take New Noise:
Audio test set I'm not quite as happy with. In that
the oscillator isn't quite as flexible as I would
have liked. But I'd spent so much time (or so it seemd)
on it that I thought it would be better to at least
get something happening that I can use and worry about
the cool bits later. So it has no sweep arrangement.
does have a useful VU meter section using a pair of
mechanical VU meters augmented with a series of 3
peak LEDs. This was an adaption of something Rod Elliott
of ESP fame published on his web site some time back.
But I had to modify it for more gain and to add the
amplifier (monitor) section was ripped out of an old
portable cassette radio. It saved me the bother of
having to actually build an amp, which would be way
too boring, and it came with it's own front panel.
Which presented some interesting mechanical challenges.
way through that part of the project I decided that
it would be far better to mount the whole thing inside
an old cassette deck. One of many I'd collected for
just such a purpose. So with some cutting, grinding,
painting and general remodeling, I managed to marry
the two different devices together as one. In the
pictures it kinda looks like a cassette deck but then
gain it looks nothing like one. How many cassette
decks have BNCs and RJ45s on them?
used an RJ45 to connect the test set to it's remote
box. The idea here was to have a little box with some
banana sockets and terminals etc, which would bring
the action closer to a project under development if
necessary. The BNCs proved access to scope probes
and other arrangements if I want to use it like that.
I can even directly BNC it into the scope if necessary.
also provided a square wave output and a pulse out
for triggering logic circuits if need be. When working
in slow mode, A little LED flashes whenever the square
wave is at the top of it's travel so you can see if
it's actually pulsing. Apart from this pulse output,
all the other I/Os are isolated by little 600 ohm
audio transformers. The impedance of the stereo amp
inputs are raised a little with resistors so that
it doesn't load a circuit under test quite so much.
not perfect but it's convenient. And it might be of
some interest for a number of reasons. None the least
of which is that I have used recycled junk to make
something really useful. As I usually do.
AND FOR MYNEXT TRICK:
current project however, is an investigation of old
computer switchmode Power supplies. I needed a new
bench supply and wondered if I could do it using SMPS
technology instead of linear. Normally one uses linear
for such things but they're big and bulky and lets
face it, they're just no fun. It's a bit like building
mixing desks one channel at a time. You might design
the best mixer in the world but after the second channel
it's all just slave labor.
a warning though. Switch modes are astonishingly dangerous
to mess with. After line filtering, they directly
rectify the main voltage at mains potential. So if
you thought 240 volts AC was nasty, try 340volts odd,
of DC. At least with AC you stand a chance of letting
go at the zero crossing point. DC never gives up.
once it grabs you that's it. And 340 volts WILL GRAB
if you're wondering how you get get 340 volts DC from
240volts AC, it's because the mains voltage is an
RMS value. The simple way to work out the actual rectified
voltage is to multiply the AC voltage by 1.414. This
actually makes it 339.36volts DC. But this is a handy
number to remember for working out what your linear
supply will give you after rectification.
you knew that.
floating round on the primary side of an SMPS is a
really nasty set of conditions you don't wanna mess
with. And unless you're absolutely confident and totally
awake and aware, don't mess with this stuff. And having
an ISOLATED mains supply through an ISOLATION TRANSFORMER
is mandatory at the very least. An RCD (Residual current
Device) is also a really good idea. But it must be
on the isolated side of the isolation transformer.
Just having one in your mains house fuse box isn't
going to be of any help because the isolation transformer
won't trigger it.
to cut a long story short. I pulled apart an old AT
PC power supply and started prodding around It was
pretty fascinating and I've mapped out the relevant
parts of the circuit now. But it only took me about
an hour of scrutiny to figure out how to make the
PSU variable. It was then I realized that if it was
variable, then the capacitors would probably burst
at all the extra voltage. So they had to be changed
immediately. So far I've isolated it down, mapped
it out and finally worked out how it does it's current
sensing. With any luck, tonight I'll have a proof
problem here is that each switch mode is going to
be a little different. These Power designers are sneaky
bastards and know how to cut corners with the magnetics.
Very clever but it also means that each designer has
their favorite way of doing things and the magnetics
are always going to be largely a mystery. So it helps
to do lots of research. And it also means that publishing
any project along these lines is going to be difficult.
But I'll endeavor to see what I can do that might
help as a guide to anyone who wants to risk their
lives playing with this shit.
the mean time, Send me all your junk. Don't send it
to land fill and there will be more of this crap to
Be absolutely Icebox.