Iron core toroid test at 1 kHz square wave

[P E Schoen]

As "promised" in my previous thread, I performed some tests on a specially
wound iron core toroid. The primary consists of two coils of 8 turns each,
about #10 AWG, and the secondary is 100 turns of about #18 AWG. The core is
rated 80 VA at 60 Hz.

I made a push-pull driver consisting of a PIC16F684 driving a pair of
IRL2203 MOSFETs rated 30V, 115A, 7milliohms. I have 0.1 ohm sense resistors
from source to ground, and I'm driving the gates from the PIC through 100
ohm resistors and 1k to ground.

I used the PWM module to generate a 1 kHz square wave with 50% duty cycle
and deadband of 7 cycles with an 8 MHz clock or 3.5 uSec. I used an
adjustable lab power supply for the voltage to the center tap of the
transformer.

Under no load conditions, I got:

4V 0.58A 2.32W 96V P-P
8V 1.01A 8.08W 192V P-P
12V 1.42A 17.0W 293V P-P

With a 1k 10W resistor load:

4V 1.10A 4.4W 89.6V P-P 2.01W
8V 2.00A 16W 180V P-P 8.1W
12V 2.89A 34.7W 266V P-P 17.7W

This is intended for about 1 kW output, but that will require a more
powerful voltage source and conductors that can handle 50 amps. But it
appears that the transformer has no more than 17 watts of core loss with the
1 kHz square wave, which will be less than 2% at the design power rating.

I found that the rise time of the output was about 1 uSec from zero to a
200V peak, then ringing for about 8 uSec to settle at 100V. There was almost
no ringing with the 1k load. From these results I conclude that the toroid
transformer has good performance at 1 kHz and is probably usable up to 10
kHz. If the 2% core loss is the major source of inefficiency, a DC-DC
converter should be able to get close to 98% efficiency, although output
rectifiers and filters may lower that somewhat.

Now I can use the same setup to test a ferrite core transformer at 50 to 200
kHz. I think the best way to approach this design is with multiple switching
units in parallel. It is rather impractical to run PCB traces for 50 amps,
so for this I would probably use five circuits to keep currents in the 10
amp ballpark, and use like #16 AWG for each, and a bus bar to parallel the
primaries and connect to the battery. In that case, I could make the outputs
30 or 35 VDC, and wire them in series to get the 150-180 VDC for the motor
controller DC bus. And 200 watt switching supplies are much easier to handle
than 1 kW.

But this was still of value in determining the usefulness of iron core
toroids for high frequency.

Paul

 

[Phil Allison]

"P E Schoen"

This is intended for about 1 kW output, but that will require a more
powerful voltage source and conductors that can handle 50 amps. But it
appears that the transformer has no more than 17 watts of core loss with the
1 kHz square wave, which will be less than 2% at the design power rating.

** Are you on the same planet as the rest of us ??

That puny core is suitable for an 80 VA tranny for a simple reason - temp
rise.

An 80 VA toroidal tranny has about 10 % losses - mostly copper loss.

So, the running temp reaches a safe max at 10 watts loss in free air.

What power loss are you looking at ?

30 or 40 watts right ?

The thing will be smoking hot unless you blow a ton of cool air over and
through it.



.... Phil

 

[Jamie]

I have wound more transformers than i can count, using a guide printed
in Popular Mechanics, Little Library of Useful Information #41,
"Home-Built Transformers" second printing 1944 and for all practical
purposes there was NO copper loss at full power and all of the
transformers ran cool.
Furthermore, there was always a little extra room in the window(s).
And for toroids (which was not covered eXplicitly) there is always
plenty of room by default if one follows the guides.

There were only two exceptions: i wound them specifically for high
current load and short term use; in fact i still have one of them; i
believe it is a 50W core, output at 1A test load of 1.45V with 20A max
design semi-continuous load (wire gets Baer-ly warm) and capable of 80A
for short term use (wire gets hot after about one minute).
This is a loose-goosey wiring where one large-sized conductor (or
better) flat strip would give much lower copper loss.
I used 5 turns each of four #18 twisted wire, knowing that was not
adequate for lower loss (#18 was largest wire on hand).
I should have used at least six #18 wires in a flat lay arrangement;
the twisting ate up too much room in the windows (no...NOT Micro$uck)
but there is still room to spade nonetheless.

Talking about mags.

I still have every issue, - the lost one's of Popular Electronics
since I was thirteen up until they stopped publishing on the rack. And a
few here and there when they attempted to restart it with some variant
authors that didn't seem to touch in areas of my interest.

Lots of interesting reading there.


Jamie

 

[John S]

Talking about mags.

I still have every issue, - the lost one's of Popular Electronics since
I was thirteen up until they stopped publishing on the rack. And a few
here and there when they attempted to restart it with some variant
authors that didn't seem to touch in areas of my interest.

Lots of interesting reading there.


Jamie

How egotistical to think anyone cares about your pathetic personal life,
Maynard. Shut up!

 

[Jamie]

How egotistical to think anyone cares about your pathetic personal life,
Maynard. Shut up!

Well, you took the time to read it and respond, so I guess I have some
interest out there.

Jamie

 

[John S]

Well, you took the time to read it and respond, so I guess I have some
interest out there.

Jamie

And you're presumptuous, too.

 

[Jamie]

And you're presumptuous, too.

I see

Jamie

 

[Martin Riddle]

As "promised" in my previous thread, I performed some tests on a
specially wound iron core toroid. The primary consists of two coils of
8 turns each, about #10 AWG, and the secondary is 100 turns of about
#18 AWG. The core is rated 80 VA at 60 Hz.

I made a push-pull driver consisting of a PIC16F684 driving a pair of
IRL2203 MOSFETs rated 30V, 115A, 7milliohms. I have 0.1 ohm sense
resistors from source to ground, and I'm driving the gates from the
PIC through 100 ohm resistors and 1k to ground.

I used the PWM module to generate a 1 kHz square wave with 50% duty
cycle and deadband of 7 cycles with an 8 MHz clock or 3.5 uSec. I used
an adjustable lab power supply for the voltage to the center tap of
the transformer.

Under no load conditions, I got:

4V 0.58A 2.32W 96V P-P
8V 1.01A 8.08W 192V P-P
12V 1.42A 17.0W 293V P-P

With a 1k 10W resistor load:

4V 1.10A 4.4W 89.6V P-P 2.01W
8V 2.00A 16W 180V P-P 8.1W
12V 2.89A 34.7W 266V P-P 17.7W

This is intended for about 1 kW output, but that will require a more
powerful voltage source and conductors that can handle 50 amps. But it
appears that the transformer has no more than 17 watts of core loss
with the 1 kHz square wave, which will be less than 2% at the design
power rating.

I found that the rise time of the output was about 1 uSec from zero to
a 200V peak, then ringing for about 8 uSec to settle at 100V. There
was almost no ringing with the 1k load. From these results I conclude
that the toroid transformer has good performance at 1 kHz and is
probably usable up to 10 kHz. If the 2% core loss is the major source
of inefficiency, a DC-DC converter should be able to get close to 98%
efficiency, although output rectifiers and filters may lower that
somewhat.

Now I can use the same setup to test a ferrite core transformer at 50
to 200 kHz. I think the best way to approach this design is with
multiple switching units in parallel. It is rather impractical to run
PCB traces for 50 amps, so for this I would probably use five circuits
to keep currents in the 10 amp ballpark, and use like #16 AWG for
each, and a bus bar to parallel the primaries and connect to the
battery. In that case, I could make the outputs 30 or 35 VDC, and wire
them in series to get the 150-180 VDC for the motor controller DC bus.
And 200 watt switching supplies are much easier to handle than 1 kW.

But this was still of value in determining the usefulness of iron core
toroids for high frequency.

Paul

Core loss is usually specified in Watt/LB. ~2ish watts/lb is generally
acceptable.
But it depends upon heating and temp rise and duty cycle and cooling.

Cheers

 

[P E Schoen]

"Robert Baer" wrote in message

Paul:

Excellent!
The data given seems to imply that there is less percentage loss (lower
relative drive current WRT drive voltage) as the drive voltage
is increased; making the overall efficiency higher as drive voltage
is increased.

I would suggest you keep those wires separate all the way to the battery
connector - that way the wire resistance of each leg would
end to balance or equalize the currents.

This test was done under far from ideal conditions, so the figures represent
a "worst case" scenario. The power supply leads were #22 test clip leads
about 3ft long, and the wires going to the transformer were #18AWG about 18"
long. So the supply wires at about 0.2 ohms drops about 1.7 watts, and the
transformer leads drop perhaps 0.5 watts. The source resistors contribute
about 0.25 watts. So about 2.5 watts out of 35 at 12V.

The transformer windings are significant with about 4 ft of #10 primary at
about 0.004 ohms. At the design target of 50 amps, it's 10 watts each, but
they are mostly exposed to free convection so they might not get extremely
hot, and I have room for heavier wire or multiple strands. The secondary is
about 70 ft of #18 or about 0.5 ohms. At 1kVA design target of 160 volts at
6.25A that would be 19 watts. So, yes, there are copper losses of about 40
watts, plus the core loss of 15 watts, which is 55 watts or 5.5% of kVA, for
94.5% efficiency. That might present a problem without forced air or other
cooling.

So Phil is right about his 30-40 watts estimate. But my goal was mostly to
determine the core loss of an iron core toroid at 1 kHz, and then proceed
with more detailed design analysis to determine the limits of what is
practical. Perhaps just 8x (480 Hz) and 640 VA continuous is more
reasonable. I suspect the core losses would drop to less than 10 watts, and
copper losses maybe 6+8 = 14, or 24 watts total.

My tests also indicate that a toroid such as this might be good for audio
applications, especially when used at the design rating of 80 VA. That would
be at a primary voltage of 4 volts 20 amps, where core loss at 1 kHz is 2
watts. If proportional to frequency, it would by 40 watts at 20 kHz, but
unless a system required most of its power at that frequency, a "normal"
audio signal of about 5 kHz average would have only 10 watts loss or 12%.

Paul

 

[Phil Allison]

"P E Schoen"

So Phil is right about his 30-40 watts estimate.

** Of course he is.


But my goal was mostly to
determine the core loss of an iron core toroid at 1 kHz, and then proceed
with more detailed design analysis to determine the limits of what is
practical. Perhaps just 8x (480 Hz) and 640 VA continuous is more
reasonable. I suspect the core losses would drop to less than 10 watts, and
copper losses maybe 6+8 = 14, or 24 watts total.

** Still need a lot of fan cooling.


My tests also indicate that a toroid such as this might be good for audio
applications, especially when used at the design rating of 80 VA.

** Mains rated toroidal transformers make nice audio transformers too - if
the winding voltages are suitable.


That would
be at a primary voltage of 4 volts 20 amps, where core loss at 1 kHz is 2
watts. If proportional to frequency, it would by 40 watts at 20 kHz,


** Oh dear - you really are living on another planet !!

For a given transformer, core losses go DOWN as input frequency rises
because magnetisation of the core is inversely proportional to frequency.

For example, a mains rated toroidal with 120V, 60Hz primary can be used for
audio from 60Hz to 100kHz at rated voltage and power or from 30Hz up if the
input voltage is kept below 60 volts rms.

Core saturation at low frequencies is your only concern.



.... Phil

 

[Phil Allison]

"Robert Baer"

I have wound more transformers than i can count, using a guide printed in
Popular Mechanics, Little Library of Useful Information #41, "Home-Built
Transformers" second printing 1944 and for all practical purposes there
was NO copper loss at full power and all of the transformers ran cool.

** No copper loss at full load ?

The busses don't stop where this dude lives ....


..... Phil

 

[Phil Allison]

"Robert Bare Faced Liar "

Well, obviously there were core losses at full load, but total losses
were low enough that the transformer(s) did not seem to get warm.

** Err - so it was NOT at full load.

Fuckwurt.



.... Phil

 

[P E Schoen]

"Robert Baer" wrote in message

Well, obviously there were core losses at full load, but total losses were
low enough that the transformer(s) did not seem to get warm.

There will always be significant copper and core losses which result in
heating and reduced efficiency as the watts (or VA) per cubic inch increase.
There are always tradeoffs between efficiency and size and cost of materials
and other factors. But any transformer that does not exhibit any noticeable
temperature rise must be of a rather trivial design with very low power for
its size.

I could not find anything about the Popular Mechanics article. I have also
wound numerous transformers, and some of them were not really very good
designs, because I had little concept of winding resistance and heating when
wound in layers. I typically tried to get more voltage out by increasing
secondary turns, and "wound" up getting less than expected.

When I "discovered" toroid transformers, I thought I could make high current
transformers for circuit breaker testing much smaller and lighter than their
E-I cousins, and I thought I could save even more weight (and cost) by using
aluminum bus bars. But when I built the first prototype using three 1.4 kVA
cores, I was disappointed in the output I was able to obtain. It improved
much when I used copper bus, and I also discovered the importance of clean,
tight connections. But I also discovered the limitations caused by internal
impedance and inductance effects of even apparently short sections of bus.

Eventually I designed and built a successful toroid transformer with 5.6 kVA
continuous rating and capable of over 10,000 amps into a short. Even with
this experience, I ran into difficulties with a larger design, but that was
because the primary cores were not designed to provide the maximum flux at
rated line voltage. The design would have worked if I had removed about 20%
of the primary windings, but my employer/customer decided to abandon the
project and instead go with a C-core design. We finally got that right and
now it is capable of producing about 10 VAC at peak currents close to
100,000 amps. The toroid design probably would have been at least as good,
but there were considerable mechanical challenges.

Paul

 

[Jasen Betts]

"P E Schoen"

So Phil is right about his 30-40 watts estimate.

** Of course he is.

_I_ never doubted it.

For a given transformer, core losses go DOWN as input frequency rises
because magnetisation of the core is inversely proportional to frequency.

That's magnetic hysteresis losses right?
And eddy current losses are proportional to what? V^2 like a resistor?
so at, or below rated voltage they won't be a problem.

 

[P E Schoen]

"Jasen Betts" wrote in message

_I_ never doubted it.

That's magnetic hysteresis losses right?
And eddy current losses are proportional to what? V^2 like a resistor?
so at, or below rated voltage they won't be a problem.

In the interests of completeness and curiosity, I repeated the tests at 500
Hz and up to 16 kHz.

Under no load conditions 500 Hz, I got:

4V 0.71A 2.84W 96V P-P
8V 1.24A 9.92W 194V P-P
12V 1.70A 20.4W 286V P-P

With a 1k 10W resistor load:

4V 1.22A 4.9W 93.6V P-P 2.2W
8V 2.25A 18.0W 190V P-P 9.0W
12V 3.15A 37.8W 277V P-P 19.2W 51%

Under no load conditions 1000Hz, I got:

4V 0.58A 2.32W 96V P-P
8V 1.01A 8.08W 192V P-P
12V 1.42A 17.0W 293V P-P

With a 1k 10W resistor load:

4V 1.10A 4.4W 89.6V P-P 2.0W
8V 2.00A 16.0W 180V P-P 8.1W
12V 2.89A 34.7W 266V P-P 17.7W 51%

Under no load conditions, 2000 Hz, I got:

4V 0.48A 1.92W 96V P-P
8V 0.87A 6.96W 192V P-P
12V 1.20A 14.4W 279V P-P

With a 1k 10W resistor load:

4V 1.00A 4.0W 93.6V P-P 2.2W
8V 1.86A 14.9W 186V P-P 8.7W
12V 2.70A 32.4W 274V P-P 18.8W 58%

At 4000 Hz with a 1k 10W resistor load:

4V 0.91A 3.6W 91.0V P-P 2.1W
8V 1.71A 13.7W 182V P-P 8.3W
12V 2.49A 29.9W 270V P-P 18.2W 61%

At 8000 Hz with a 1k 10W resistor load:

4V 0.81A 3.2W 94.0V P-P 2.2W
8V 1.54A 12.3W 179V P-P 8.0W
12V 2.25A 27.0W 267V P-P 17.8W 66%

At 16000 Hz with a 1k 10W resistor load:

4V 0.73A 2.9W 89.6V P-P 2.0W
8V 1.46A 11.7W 181V P-P 8.2W
12V 2.10A 25.2W 266V P-P 17.7W 70%

16000 Hz, No Load

4V 0.26A 1.0W 94.5V P-P
8V 0.50A 4.0W 183V P-P
12V 0.72A 8.6W 272V P-P

I interpret this to mean that magnetic hysteresis losses drop as a function
of frequency. And of course this is at very low current. Will there be other
core losses that increase with current?

Thanks,

Paul

 

[Phil Allison]

"Robert Bare Faced Liar"

ALL of the transformers (except the two mentioned) did not seem to get
warm at full load;

** Then they are NOT operating at full load.

You fucking tenth wit !!!!!!!

Cos that implies there is a free air temp rise in the windings of 65C -
at least.