Re: Power Supply Rectification and Smoothing

On Sat, 20 Oct 2007 05:21:23 GMT, Ross Herbert
<rherber1@xxxxxxxxxxxxxx> wrote:

First up, Ross, let me thank you for your most valuable assistance in
helping me understand these new (to me) concepts. Most appreciated.

I wrote:

Thanks Ross. Makes sense. Did you see my question I posed about
turning the mains off and then on quickly with the transformer core
still magnetised, and the phase angle at switchon such that the
transformer goes "boing"?

No, I didn't see that question. However, it is well known that rapid
on-off switching on the mains side of large transformer based power
supplies can result in unintended surges and overloads. Mostly, these
simply cause a circuit breaker to pop but there may be other
consequences depending on the psu design. It is generally accepted
that this practice is frowned upon and belongs in the realm of "worker
vandalism".

Understood. I guess some important machines are wired with a time
delay to prevent this intentional or unintentional mishap.

We used to get a lot of <1 second power dropoutss which would turn off
some clocks and computers. Seems to have disappeared since we got the
nice underground power. Now THAT is a nice conductor, I wonder how I
could wind my tranny with that :)

To avoid the need of caps, and keep the voltage stable, what about
another suggestion I've had to build "a simple series pass regulator
using a 36V zener like a 1N5936 and a 60A NPN transistor like a
MJ14002. it would be tons cheaper than the humongous capacitors you
would need. Since these need a couple of volts headroom, wind the
transformer to provide a slightly higher voltage say 32V or so. Also
the transistor must be put on a largish heatsink."

You could use this arrangement with a single large electrolytic across
the zener reference as a means of producing a "capacitor multiplying"
effect but you would still need a fairly hefty bulk storage capacitor
across the output to provide for instantaneous current demand due to
sudden load changes. You wouldn't save a great deal if any and in my
opinion it is not required for your application.

Thanks for that. Understood.

I doubt that a single series pass transistor would be suitable either.

OK

Or even "to parallel the outputs of 5 or 6 LM338K 3 terminal
regulators. Since the max output is 32V you would have to float the
regulators above ground using a 3.9V zener like a 1n4730.
Again you would need a big heatsink."

Solid state linear regulators (well designed) for high power DC
supplies have been used successfully as far back as the 60's and are
still used today when the most rugged operational conditions are to be
endured. $ for $ they will withstand almost anything you can throw at
them and while heavy duty switch mode units are also very capable, the
inclusion of safeguards against sudden blow-ups makes them more costly
and complex to design.

Fair enough. I really want to keep this project fairly inexpensive and
simple. I mean, if all I wanted was a DC variable speed drive for my
lathe, I could just buy a lathe with it on and pay the $2000 extra.
a). I can't afford it,
b). I want the learning and experience gained from homebrewing it and
c). It's not all that important for a LOT of unwanted effort and
expense.

This one weighs 11 lb, and some others weigh up to 15. Have you a
ball park figure for a reasonably stable load regulation?

It depends to a large extent on how you are going to drive your motor.
If it is going to be running at near full load for several hours at a
time then an 11lb transformer is totally inadequate. For such a
requirement (36V @18A) I would specify a transformer of 1KVA to be on
the safe side. As an example, I have a 1KVA isolation transformer and
this weighs 16Kg (35lb), so your 11lb unit is a pip-squeak by
comparison.

Yes, I understand.

It will be driving a hobby lathe, and thus will be used for probably
only a matter of seconds at full load to perhaps a minute or two at
half to three-quarters full load. Otherwise, idling.

This microwave transformer runs at around 1kVA (2000V at about half an
amp) in the microwave, but it is only intermittent use, and is fan
cooled.

I imagine your isolation transformer is rated at 1kVA continuously,
without fan cooling and in possibly hot ambient conditions?

Apparently 1000W output microwave ovens only have a 15lb transformer
and that would have nearly 1500W through it.

I want to run a mere 650W intermittently through the thing, while
blowing air over it and monitoring its temperature.

I first saw these transformers being used in homebrew welding
machines. I've already got two stick welders, so I got to thinking
that it might be nice to put a few more turns around the secondary
core to get 24 or 36 V for a DC motor of which I have a number. The
welding fraternity only wrap a few turns around the trannies to get
their few volts of heavy current. Admittedly, they gang several up
together, depending on how big a bridge they want to weld. They say
they wind them with flex and seem to get around 1 or two turns per
volt.

The primary turns on this one that I happened to have lying about are
either 14 or 15 turns each side of a square window, so around 200 to
220 turns in all. It may be better to use a 15 pounder that I hope is
in the other old microwave I have yet to plunder. I'm saving the
magnets in them to eventually make a wind turbine, multipole
generator.

I'm interested in these projects, partly because it is my nature -
I've been that way for most of my 60 odd years,
(emphasis on the odd :).

This lathe motor is not essential. It already has a perfectly good AC
induction motor, but I figured a variable speed drive would be a nice
feature and teach me something in the meantime. You are helping with
that teaching. Thanks again.

I mean, I can get 2 dozen of these for around $US18 . Cheapest
capacitance I can find. Even three 12V leadacids would cost much more
and last for much less a time.

It seems to me that you have yet to grasp that your project is not a
simple "use what you can find at the lowest cost" type of project.

Understood, but if it is expensive, I won't do it, and will understand
why and thus have gained some knowledge and insight

I looked for the most suitable caps at *** Smith, Altronics and
Jaycar. This seemed to be the 50V 4700uF electrolytics.
As luck would have it, I found at Oatley Electronics, simiar spec
electrolytics at around one sixth the price. Pure coincidence.
I think they only offer 4 different caps.

As apparently little smoothing is required, I was thinking of just
adding $18 worth of these caps wired up as 100V and giving ~28,000uF.
I'm not quite clear whether smoothing caps are of any advantage at all
if in fact no smoothing is required.

Is the upside, a slightly higher effective voltage, and less
likelihood of 100Hz hum, and the downside, perhaps expense?
Are there any other downsides?
Am I wrong in assuming that any smoothing is worthwhile?

I've just had a thought occur to me, and it's probably ridiculous, but
could anything possibly undue occur if the frequency of the PWM of the
controller is a harmonic of 100Hz of any ripple left. It just keeps
nagging me that chopping a variable voltage might have unpredicted
consequences. Seems that no-one I've heard of has ever tried what I'm
doing, so I've had no input from actual experience, just educated
guesses.

What you are trying to do is build a fairly heavy duty transformer
based dc power supply and to do this you have to start out with an
appropriately rated transformer in the first instance. There is no
substitute for an adequate laminated core mass when good load
regulation is required. Grabbing an old microwave transformer and
slapping on a few turns of pvc flex as a secondary is totally
inadequate in my opinion. If you had used something like 10AWG Formvar
insulated magnet wire then you would achieve far greater magnetic
coupling efficiency and get a result closer to what you were looking
for.

So the proximity of the windings to each other is quite important
vis-a-vis output voltage stability? I'm totally clueless about
transformer design, but I'd assumed that if a secondary turn went
through the secondary core windows, it was induced with a voltage
dependent on the primary windings. I'm sure I've seen some
applications where they actually space their windings apart with bits
of cardbord.

See below for a description of this flex..

BTW, according to the current carrying tables, 10AWG can only handle
14.8A for power transmission.

You can't use a transformer with poor load regulation AND get by with
the lower voltage rated capacitors - it's that simple. You either use
a good transformer which produces little voltage sag between low load
and full load where lower voltage caps can be used or you cover the
available output voltage range using higher rated caps. It's one or
the other and you have a cost penalty either way.

I understand. Thanks. That's a clear explanation.

I can of course now see ways of perhaps avoiding or minimising these
trade-off problems you point out. I'm not trying to be argumentative,
but if I am aware of the likely voltage rise on light loads, I can
look out for it, by measurement, and wire my caps in series to make
them safe to 100V.

I will of course be monitoring temp rises when testing.

I'm not trying to manufacture an idiot-proof item for public
consumption, but a Heath-Robinson contraption that I understand fully
and know what to watch out for.

Like some of the old banger cars I used to nurse about the place. If
anyone else tried to drive them, they would fall apart at the first
intersection. They all did me sterling service!
I can remember an RAC breakdown mechanic refusing to help with my
LandRover because it was so non-standard. I got it home without his
help.

This stuff is automotive, I think. It has quite a thin PVC sheath.
Its overall diameter is a whisker over 3mm. Drawn to scale, I can fit
at least 54 turns in the transformers secondary window.
I intend to have a fan blowing continuously over the whole shebang,
caps, transformer, bridge rectifier and motor, and I will monitor the
temps of everything to begin with to make sure that nothing is
getting too hot.

It is a long time since I wound a secondary to get a desired voltage
but from memory I had to wind something between 3 and 5 turns per volt
(240Vac primary) to get what I wanted. Depending upon the number of
turns on your primary will determine your required turns per volt at
the secondary and you should experiment to determine this. 54T sounds
a bit on the low side for a 36V,18A load to me.

Of course I will test it, but as I said above, I believe there are
around 220 turns of the primary windings, so theoretically, 33 turns
should give me 36V across the secondary windings. Gives me some room
for adjustment.

From my experience, depending upon the supplier, automotive shops
describe their flex with rather dubious current ratings. If the
overall diameter of your flex is 3mm the wire diameter will be about
half that at best and I would question its ability to conduct 30A for
any appreciable time without drastic losses and severe temperature
increase. See the chart here http://www.powerstream.com/Wire_Size.htm
to get an idea of wire gauge for a nominal current capacity. For
example, 30A power transmission conductor would be in the region of
3.7mm diameter without any insulation.

It is labelled 41 0.32
That is a 3.3mm^2 conductor.
Nominally rated at 30A

I've measured it at 41 0.30.
That is a 2.9mm^2 conductor.
Nominally rated at 25A.

I will want it to carry 18A
intermittently, and with fan cooling.

BTW, that 3.7mm diameter wire mentioned above is 10.75 mm ^2 or 7AWG.
It is rated at 89A for chassis wiring.

I sometimes run 18A (240V) through a 2.5mm^2 conductor and it gets
perceptibly warm after an hour or so.

Surely that power transmission rating is downrated for extra long
distances where a low maximum voltage drop is specified?

Fig 8 lamp flex of 0.75mm^2 cross section is rated at 7.5A which
multiplied up would mean a 2.9mm^2 conductor would be rated at 29A.
*** Smith calls it 30A - Jaycar calls it 25A, and charges twice as
much. Altronics don't appear to have anything similar

I think *** is getting out of the wire business and hasn't bothered
to raise his prices along with the price of copper. Everyone else
seems to have doubled the price of their cables, even though the
doubled price of copper is only relevant to a quarter of the cable
price.

Thanks for that, Ross. The speed controller is described by my
friendly ee who has experience with them on his bike, as:
"they have big input caps on the supply voltage in the controller,
and the circuit after that can be thought of as a buck switching
supply. All the logic and control runs from a low voltage (14.4v in
Crystalyte) that is internally regulated, so as long as your ripple
does not drop below that or go above the voltage ratings of the fets
or caps, you should be fine."

Since it's a "chopper" type speed controller then a bit of ripple on
the dc output will not be a problem. Your main problem is your
transformer load regulation as I read it.

That seems to be the general consensus, now (the ripple) but as I'm
only running an intermittent 18A max through the secondary windings,
perhaps if I doubled up the caps in series for a 100V rating, and
adjusted the secondary windings such that at low load, the voltage did
not creep too high, it is not worth a try?

Of course, it is up to me, along with any responsibility for any
injury, but I would appreciate any learned comment.

I am, BTW, so cautious, checking everything repeatedly, that some say
I'm neurotic. Friends would buy equipment and be so anxious to
assemble it and switch it on, that they would be tripping all over the
packaging, whereas I would take a day or two savouring the assembly
and thinking about every step. So there! ZAP!!! Where did that
bugger come from? :)

I've also got a couple of good quality brushless 24V 400W motors
with built in brushless controllers. He rebuilds these to reverse
the rotation. I will ask him about these too.

I've posed the question and await the reply.
These are beautiful machines - black, finned, machined aluminium with
just two heavy power leads emerging and a small control loom for a
pot or a Hall effect twist grip "throttle".
These are too nice for a lathe, best for my trike :)

I haven't any experience with this type of motor so I can't add
anything.

You've helped me enough already, Ross, thanks again, jack
.