[Amps] HV transformers I want to series primary andparallelsecondary

[Tom W8JI]

> Now you've got me worried.  I am building an amp using a 
> 4CX1500A and didn't
> have a filament transformer that would supply 5V at 42A 
> but I do have 2
> identical 25A transformers.  Am I going to run into 
> trouble here?  I plan to
> use a step start circuit to control start up inrush.

I think the problem of directly paralleling identical 
transformers is highly exaggerated. My guess is the paranoia 
comes from people who assume few percent turns difference 
will somehow behave like a shorted turn. That isn't the case 
at all.

 The only potential for problems comes from a sloppy winding 
method that has a quite large percentage of turns error 
between two transformers so the voltage difference is large 
compared to the combined ESR of the transformers. I suppose 
some transformers might have such grave errors, but they 
would be easy to test for. This also would ONLY apply if the 
transformers had parallel connections on both the primary 
and secondary sides. Place either or both sides in series 
and the perceived problem vanishes, although a secondary 
insulation breakdown problem might exist with series 
connected HV secondarys.

Let's look at an example. Say we had one half volt per turn 
transformer with a 10 volt secondary that had 10.5 volts 
open circuit and 10 volts at a full load current of 10 amps. 
Let's assume they don't monitor the 240 turn 120 volt 
primary so well in winding and it is 250 turns on one 
transformer and 240 turns on the other. Now we have a 
secondary NL voltage 96 percent of the ideal voltage on one 
compared to the other, or 10.08 volts compared to 10.5 volts 
on the other no load. That's very poor transformer QC.

Since we know the full load voltage of a single transformer 
drops 1/2 volt with a 10 amp load, we know the ESR is .5/10 
= .05 ohms in each transformer. We have .1 ohms total ESR 
with the secondary difference driving the largely 
dissipative resistance (some may be due to flux leakage and 
not cause heat).  1/2 volt across the .1 ohm ESR is .5 
squared over .1 ohms ESR or about 2.5 watts of extra heat 
split between two transformers. That's due to the .5 over .1 
= 5 amps of no-load circulating currents flowing through the 
.1 ohm ESR.

The operating dissipation of a perfect transformer would be 
somewhat more than secondary I squared times secondary ESR, 
or 10^2 * .05 = 5 watts. The higher voltage transformer 
would get a few percent more heat than the other since it 
has a little more voltage at the same current, but not much. 
We've added about 2.5 watts maximum  heat to each 
transformer that would normally be dissipating about 5 watts 
at full load.

To simplify this explanation I've left out a couple things 
that would have a small effect (like the small heatless flux 
leakage for one), but you get the general picture.

You can see in this case the error between transformers, 
even in the case of very stiff transformers, has to be 
pretty large to add significant heat. In this case a ~5% 
open circuit voltage error produced an additional 50% heat. 
I can tell you something from my experience at a company 
where we wound transformers and coils used in the 
aftermarket automotive industry. Where no one cared about 
precise turns our oldest poorest coil winding machines would 
be within one turn plus 3% of total turns. In more critical 
devices we would use machines that were always exact. We 
could wind a 1000 turn coil and have 1000 turns every time. 
The only error would be if the operator unwound a turn when 
making a connection.

I suppose we could have two identical transformers off a few 
percent, but it would be easy to check for. Just measure the 
primary no-load magnetizing current of the two transformers 
with and without the secondary paralleled. You can calculate 
the maximum dissipation increase of the pair by knowing that 
current increase and the primary voltage.

My bet is this is one of those things that got started 
because someone somewhere thought a two turn difference 
would act like a shorted turn in one transformer, which it 
would not ever do. So now we are all paranoid about nothing. 
When I want to read much ado about nothing, I read the amps 
reflector. :-)

73 Tom