Thor Meets His Maker

[CJ]

Once upon a time there was a transformer:

(cue jaws theme!)

This is a Thordarson line to line I believe. The voltage ratio is 1:1.
Inductance is about 18.5 henries.
DC resistance is about 65 ohms for each of the two windings.

Were gonna take a closer look.
This varnish was the toughest I have seen yet. Almost had to break out the blow torch. Pulling the first lam after removing the frame:

http://vacuumbrain.com/The_Lab/Thordarson/thordarson_1.jpg

Hmmm, very interesting! Looks like somebody put a cherry on top of this hot fudge sunday. Notice the shiny lams?

Finally down to the last lams.
Hmmmm, do I detect a bi-filar winding structure perhaps? (look close):

http://vacuumbrain.com/The_Lab/Thordarson/thordarson_4.jpg

Lets pull the wrapper and see what the heck is going on:

http://vacuumbrain.com/The_Lab/Thordarson/thordarson_6.jpg

Ahah! Looks like we got some API tricks going on here:

http://vacuumbrain.com/The_Lab/Thordarson/thordarson_8.jpg

Almost done unwinding:

So it is a bi-filar winding structure. Primary is red, secondary green. Or visa aversa. Don't really matter. Two wires wound at the same time. Not litz braided, but just two different spools used. Hand wound obviously.
Here is the start. The pri and sec starts are brought out on opposite sides of the bobbin:

Parts is parts:

OK, summary of construction:

Lams: 625 EI http://vacuumbrain.com/The_Lab/Thordarson/thordarson_3.jpg

0.014 thick. 38 M6 lams--5 50/50 lams. They look like old Mag Metals glazed lams.

Wire: about 0.007-0.008 , probably #33 ga. Length of each winding: 293 feet.

One winding was 62.8 ohms, the other was 67.4 ohms. Probably due to slight differences in wire and hand winding.

222 ohms/1000 ft. 8 layers--120 turns per layer--960 turns total.

No E-shields or interlayer insulation. 18.3 mH air core inductance.

Weight: 8.8 oz.

Leakage: 145 uH each winding.

Supposedly, this is a killer line to line x-former.

I will sweep the other one I have and see how it looks.

 

[adrianh]

CJ
We all need to ask did you write broken on the transformer
to make you feel better or was it really broken?

 

[CJ]

There was a break where the heavy buss wire meets the smaller green wire.
I recieved two. One good, one broke. So it wasn't rally aggravated assault, more like putting someone out of their misery!:razz:
I can easily rewind the broken one.
It uses the same lams as the API output, and I have some in 49 and 80 Ni, so maybe I will soup it up a bit.

This would be another one that would be an easy DIY.

I found a bobbin:

Cosmo Part Number: 7036-0
Part Description: Transformer bobbin without terminals
Quick-Ship Part?: Yes - Next Day delivery available direct from stock.
Part Dimensions:
Lamination EI-625
Core Dimensions 640 X .640
Flange Dimensions 1.218 X 1.218

Length Wall .921
.
Wall thickness: .030


No Middle Flange

 

[electronaut]

By far the best camping spot I've found in the midwest is a beautiful island on Lake Michigan that used to be owned by Thordarson. Apparently he meant to build a house to retire in, but all he ever built was the boathouse and a few other small buildings. The boathouse is a beautifully built stone building that is now a museum, housing one of his early prototypes of the first MILLION VOLT transformer, which he designed and won an award for.

The story of Hjortur "Chester" Thordarson is pretty amazing, and certainly inspiring.

(Sort of on topic, right?)

 

[CJ]

Interesting! I will have to do a search on him.
Don't worry, anything goes on my threads.

You can talk about consh shells from the bahammas for all I care.
:green:

 

[CJ]

Well, dang, this x-former makes my life easy. I don't have to post a frequency response curve. It's just a straight line from 1 hz to 1 Mhz.
I don't have to post a phase shift chart because there isn't any.
Much like the API output.

It has a nice smooth B-H also. Here are a couple of pics of a square wave and sine wave B-H.

Thats a wrap.

:guinness:

 

[PRR]

> It's just a straight line from 1 hz to 1 Mhz.

Ignoring capacitance:

With zero-Z source and 600Ω load, it should be down at 600KHz. Your MHz rating is probably infinite (over 2K) load.

With 600Ω source and no load, the bass should droop near 5Hz.

Let's see: 600Ω in and out, bass goes to 2.5Hz, top goes to 1.2MHz.

Zero-ohm input, bass is flat to 0.5Hz.

Adding in the capacitance, say 300pFd, with high-Z source and no-load there is a ring at 760KHz, and it could be as high as 20dB. With matched 600Ω impedances the Q is around 1 so no ring but a drop around 950KHz. Capacitance may be higher with bi-filar, so performance may not be this good.

But it does show the advantage of bi-filar: leakage inductance very-very low so very wide bandwidth. Advantage is lost at high impedances due to extra capacitance, but at 600Ω you can do very very good.

Dunno why so many classics used sectioned winding. Maybe for voltage reasons. A broadcast output can end up hooked across power line voltages (when two buildings are on very different grounds) so you want more insulation than classic shellac alone can reliably give; you need fishpaper. This is clearly one of the more modern poly insulations (all shellac is brown) and can stand hundreds of volts between strands.

When was this baby born?

 

[Gus]

Thanks Chris

 

[Consul]

Is there a limit to how far you could take bifilar winding? I'm just wondering if it would be possible to twist together 11 wires to make a 1:10...

 

[PRR]

> Is there a limit to how far you could take bifilar winding?

Not a limit, but benefits reduce and winding labor increases with number of strands. It is hard enough to lay one strand nice and neat, harder with two (you see the problem in CJ's photos), and 10 is probably insane.

Also: for optimum design, the total pounds of copper in each winding (primary or secondary) should be about the same. For 1:10, that would mean one fat strand and 10 super-small strands like #40. Such small strands are hard enough to wind on solo, you don't want to try to manage 10 of them all at once.

It does seem like rope-making or textile machinery could pre-wind one #24 and ten #40 into "one wire" that could be wound on the bobbin. But I've never seen anything like that, perhaps because it would not lay neatly and efficiently.

The classic Macintosh tube power amp output iron was quadrafillar-wound on the tube side. Split-load push-pull and close to Class B, so he needed GOOD coupling between windings to avoid spikes at crossover from push to pull. There is also a 30:1 ratio from primary to speaker-output, but I don't think the speaker winding is that tightly coupled to the primary, that would be too insane to wind.

 

[Consul]

First of all, thanks a lot for the answer, PRR. As always, it's a lot to think about.

[quote author="PRR"]It does seem like rope-making or textile machinery could pre-wind one #24 and ten #40 into "one wire" that could be wound on the bobbin.[/quote]

This is actually exactly what I was thinking. Take 11 small-gauge wires and wind them together, then wind them around a bobbin. It was a thought, anyway.

But I've never seen anything like that, perhaps because it would not lay neatly and efficiently.

I'm sure there is a reason somewhere. Still, might it be worth trying, to see if any problems could be solved?

I don't know why I'm so fascinated with transformers all of a sudden...

 

[aurt]

PRR wrote:

The classic Macintosh tube power amp output iron was quadrafillar-wound on the tube side.

Didn't they also have some sort of cathode coupling winding on the primary as well? I seem to remember anyone thinking of doing a DIY Mac completely losing their minds after looking at the trafos.

 

[CJ]

Some used SE iron, others were push-pull with those 70.7 screen taps.

This Stancor I took apart would make a great API output x-former by the way.