Thanks, for your excellent review of transformer design. Coming down to
a more practical problem that has bugged me in the past when I've tried
to rewind the secondary of a transformer is how to deal with the last
two laminations. Taking it apart is usually possible provided there
isn't too much varnish. Rewinding the secondary, to give a different
voltage, is a bit painful ensuring that the wire goes onto the bobbin
evenly but usually isn't too much of a problem with low voltages, but
then there's getting the core back together. It doesn't seem to matter
how much you clean up the laminations the penultimate one usually goes
in with a bit of a struggle with the aid of a blunt instrument but then
the last one ends up bent and useless.
So far the transformer I've done this with seem to work OK despite
missing one lamination but I'm sure there must be better ways of winding
and assembly we can use on the bench.
On 14/09/2011 19:23, Manfred Mornhinweg wrote:
> Hi Gene,
> I would like to take the opportunity to clear up some misconceptions.
>> Core selection appears to be fixed mainly by the voltage across the
> Not so. There are several factors to play with. Firstly, core materials
> are not all alike, and some can work at a much higher flux density than
> others, at a given level of loss. How much core loss is acceptable will
> depend on the application, kind of service, environment (temperature,
> forced air cooling, oil cooling). Then, of course, the designer of a
> transformer can trade core cross sectional area for turns per volt,
> maintaining the flux density constant. And once the cross sectional area
> has been decided on, a taller stack of smaller laminations, or vice
> versa, also leads to different iron/copper ratios.
> For these reasons two transformers of the same nominal power and
> voltages could have very different core sizes.
>> (so that there is sufficient inductance in the primary)
> In most power transformer designs one doesn't even calculate the
> inductance! It's usually high enough to be of no concern. Instead the
> flux density in the core is calculated, and this defines the amount of
> turns that has to be used on a given core, for a given frequency and
> voltage. Only in special applications would the inductance become important.
>> and the peak current that flows (so that the core doesn't saturate).
> Here you express another, very common misconception! Transformer cores
> DO NOT saturate from excessive current (at least if you mean load
> current). Instead they saturate from excess voltage per number of turns!
> Interestingly, a transformer that is idling (no load) will be closer to
> saturation than when it is delivering full output current. This is
> because the current causes a (usually small) voltage drop, which reduces
> the effective voltage setting up a magnetic field.
> Related to this is the misconception that transformers that saturate
> will cause flat-topping of the output waveform. That won't happen! If a
> transformer saturates, it will do so when magnetic flux is highest, and
> that doesn't happen at the waveform voltage peaks, but at the zero
>> So designers don't have that much flexibility in iron core selection,
>> implying that they can't vary it as much as they can wire diameter.
> Not correct. Actually there is MORE flexibility in core selection than
> in wire diameter! The wire diameter depends on the maximum current that
> will flow over a time long enough to exceed the thermal time constant of
> the wire, it will also depend on the allowable heat rise, which in turn
> depends on the kind of insulating material used, and some other factors.
> Also in many cases it's necessary to consider the total allowable
> voltage drop due to wire resistance. At the end, it boils down to a
> rather small range, from about 2A per square mm of coppper for a large
> transformer made with class A insulation and working in air, to maybe 4A
> per square mm for a small transformer using class F insulation.
> Core selection instead is far wider ranged, since the designer can
> choose between several different qualities of material, and then he can
> choose different shape factors, at the same time as the size.
> Transformer design is not like first selecting the core, then
> calculating the windings. Instead it's an interrelated procedure, in
> which both the core and the windings are altered and the results
> calculated, until the best compromise is reached for the application on
>> Then, I tried to do an "on paper" conservative design, and the thing
>> would have weighed over a hundred pounds if I had made it.
> That has happened to me too! That's why I learned to do optimized
> designs, rather than conservative ones! ;-)
>> It was also unclear how much money I would have saved after I bought
>> the iron cores at a single-unit price.
> I don't know how well your costs (in the US) would relate to mine (in
> Chile), but down here I can save about half the money by buying the
> core, wire, insulation material, and doing the work myself, over
> ordering a custom-made transformer. Instead if a ready-made Chinese
> transformer is available in the rating needed, that's far cheaper than
> winding my own!
> One can often find suitable cores in a well stocked junk box. But that's
> not too much of an advantage, because new silicon steel laminations are
> inexpensive, compared to the wire. And the wire has to be bought new
> anyway. Trying to recycle magnet wire from old transformers is a recipe
> for disaster.
>> I did not have the facilities to bake it or do a decent job of
>> laquering it
> My trick is to pour the varnish into the windings, let them soak, and
> warm up the whole winding assembly by passing DC through the wire. That
> way it dries reasonably fast. Without such heating, it takes forever and
> a day to dry.
>> or making sure that there was sufficient insulation between the
>> primary and secondary.
> A simple voltage multiplier, built from a stack of diodes and
> capacitors, powered from the line through a lightbulb as current
> limiting device, easily generates enough voltage.
>> It also probably would have looked like junk if I had made it.
> Maybe the first time. Mine did too. But after making a few, you will get
> the hang, and make good looking ones!
> By the way, the very first transformer I wound, at age 13, is still
> serving well! It does look like crap, with unprotected magnet wire
> pigtails coming out from odd places of the assembly, but it works!
>> The plate transformer he made for me-- 3KVA output, 99 degree F
>> temperature rise for CCS, and capacitive input -- ended up weighing
>> 47 pounds, and I estimate that about 2/3 of that is iron.
> Clearly he used high quality steel laminations. By using standard
> quality material, it would have ended up a bit heavier, with at least
> 3/4 iron, to reach that same temperature rise.
>> I like home-brewing, but I'm glad I did not try it myself.
> Hey, I challenge you to change that attitude! Homebrewing transformers
> is fun!
>> There is a LOT that goes into making one of them, make sure that you
>> know what you are getting into first.
> That's right. For a novice transformer homebrewer, I would advice
> starting with medium size, low voltage ones, such as used in 13.8V power
> supplies. Those are the easiest to make. Very small ones use large
> numbers of turns, and very thin wire, and high voltage transformers also
> have lots of turns.
> Manfred, XQ6FOD
> Visit my hobby homepage!
> Amps mailing list
Amps mailing list