turns ratio

This is correct.

tor

ble

.001,

ou

the HP LCR Meter used a single turn [if it could be added] to determine the turns, again sounds like ratio.

"Chieftain of the Carpet Crawlers" "Phil Allison"

** Just your opinion - or is there an actual instrument that works this way ?

Googling finds any number of turns ratio meters that use an international standard method of testing that seems employ low level voltage ratios. Accuracy is claimed to be in the order of 0.1%.

However, the intended use is mostly for 3 phase trannys in the 50kVA range or more.

... Phil

Whenever we were characterizing a new core material candidate, we used ten turns, which is the standard, IIRC.

I think that for miniatures (what we use on circuit boards), Current ratios would be pretty accurate.

It would be good to chart it through different loadings and excitation levels as well.

You can get resolution of 0.01ppm in a "ratio transformer" but I think they use one or more tranformers with tap switches and then a resistive divider to get the resolution.

I have a couple of these guys:

(paid a LOT less than that, though).

They have one toroidal-core transformer inside.

Good for calibrating resistive dividers and such like. It's almost(?) impossible to match the stability, even with very expensive resistors.

Jerk.

OK, That would be an answer to the question, "How do I make two transformers who's ratios differ by about a PPM?"

The original question read more like converting a measured approximate floating point number to ratio of integers for one transformer and have PPM accuracy in the result.

I'm still not convinced you can have PPM turns ratio resolution for less than a million turns on a repeatable, stable, cored transformer. I admit that I have no idea what error terms exist when you try to wind half turn on an E core. I can't imagine putting one-thousandth of a turn on a 1000 turn solenoid.

Sounds to me like a GIGO situation. Trivial in theory, but a bitch to do on real hardware.

If I really, really, really needed the answer, the solution is clear... unwind one. With my attention, span, I'd probably have to keep unwinding 'em until I got two counts that agreed.

I tried that, using prototype transformers from two vendors. The outputs cancel to a few millivolts out of 3 volts.

Both were designed to a spec of turns ratio "1:1.1 +-1%" but appear to both have turns ratios very close to 1.0943. Maybe they have the exact same numbers of turns. Maybe all the transformer companies use the same design software, or use some handbook full of tables. The benefit to me is that the transformers are interchangable.

The transformer companies have the math problem I was interested in: how do you come up with integers that ratio to some real number? Transformer design tangles this with issues like available wire sizes, window area, DCR, insulation, SRF, all sorts of complications.

That must have been a real pain in the days before computers. Google ... there's a lot out there.

--

John Larkin                  Highland Technology Inc
www.highlandtechnology.com   jlarkin at highlandtechnology dot com   

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME  analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

At best, use only 3 digits with "fuzz" on least "significant" digit. Better, call it 1.1:1 or with integers, 11:10. Even tho it is small, that type of transformer uses very fine wire, so expect large turn count..maybe 11,000 to 10,000 or in that crude region.

Here you are talking about PRECISION equipment, _not_ high volume consumer junk (err..stuff) like a simple miniature audio transformer.

how

Yeah, we had two Wurth reps call on us last week. They do transformers, inductors, connectors, emc/esd things. They are a family-owned German company that seems to do a lot of stuff all over the world. Sort of like Bruker. Funny how US companies have to spin off divisions because management can't concentrate their tiny brains on more than one thing at a time.

--

John Larkin                  Highland Technology Inc
www.highlandtechnology.com   jlarkin at highlandtechnology dot com   

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME  analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

The more you disclose, the more confused I get. Just to use round numbers, 3mV out of 3V is a part per thousand. Am I confusing the TLA? PPM == "parts per million"???? I can't comprehend worrying about PPM accuracy for a transformer with a specification of +-1%.

I don't have anything more to add, so I'll just (try to) STFU. ;-)

PPM stability is usually more important, but the two are of course closely related. The main application that I know of where that's a requirement is capacitive position sensors.

Everybody who thinks about capacitive sensing first tries to use resonance, but it turns out that a bridge made with a differential capacitor running against a transformer with a centre-tapped secondary is orders of magnitude better.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058

hobbs at electrooptical dot net
http://electrooptical.net

Lots of nomographs, tables and approximations- probably faster than it is now with $#$#$# computers. It took almost no time in the old days for the pros (Hammond and ex-Hammond) to design a power transformer. The core sizes and steels are pretty much standard.

Or, you could do like my mentor from the 60's who used to own a transformer company: Measure what you make and keep a huge table of recorded data. Interpolate/extrapolate. He said it saved all kinds of time in calculations and got pretty close each time.

turns

I'm not currently designing to part-per-million accuracy, although the product in question has 16 bit ADCs and DACS, and a lot of averaging is going on, so I'm argubly working in the single digits of PPMs. But I want the transformers to be as identical as possible. I don't really have a problem; I was just interested in some of the math that transformers suggest.

In a synchro/resolver system, it's voltage ratios that represent angles, not absolute voltages.

I probably have enough magnetic coupling between transformers to introduce errors in the rough ballpark of 0.02%. But I'm not shooting for arc-second accuracy.

--

John Larkin         Highland Technology, Inc

jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro   acquisition and simulation

For a power transformer of given frame and stack size and material, all the primaries will be the same, so you'd just need a modest library of them. Then there are window area and volts/turn constraints that should make the secondary fairly easy to calculate. Signal transformers are a little trickier, since the primaries aren't standardized.

--

John Larkin         Highland Technology, Inc

jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro   acquisition and simulation

Same way you made odd gear ratios back in the day, or extremely fine pitch threads, by difference. The difference between adjecent ratios can be made very fine indeed. Suppose you thread a coupler with (say) 1/4-28 outside, and #8-32 inside: after 28 turns, the outer thread rose 1 inch, but the inner thread rose 28/32nds of an inch, for an overall (1 - 28/32) / 28 = (32 - 28) / (32 * 28) = 224 TPI.

FYI, the way you implement fractional turns in multilegged cores is by weaving a flux equilization winding between the legs. For instance, on an E core, you wouldn't want to just end a winding halfway, because you get (say)

100 turns within one leg and 101 on the other, for an average 100.5 turns. The extra turn acts independently of the other hundred, so it is all leakage flux. That's not good. By linking the legs with a winding, that problem goes away and balance is enforced.

If you don't mind that your Bmax drops proportionally (in a symmetrical core), you can even make the balancing winding lopsided. Suppose the balancing winding is 1:2 turns: now you can make thirds of a turn by looping around the 1/3 branch once or twice as needed.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms