Audio Transformer Inductance

[ruffrecords]

It has long been clear to me that most audio transformer manufacturers are rather lax in publishing full specifications for their products. This makes it hard for designers to select a transformer, especially if there are two apparently identical transformers available from different manufacturers at very different prices. You expect to get something extra if you pay extra but it is not clear just what that is - is it some nebulous improved quality or just better shielding or lower distortion or what?

As a designer I know that one of the key transformer parameters that manufacturers rarely quote is the primary inductance, which has a profound effect on low frequency response and also a significant effect on the transformer's manufacturing cost. So I have shelled out some cash and purchased a decent LCR meter and tested as many audio transformers as I can get hold of. The results of theses tests are in the attached pdf file. I have a couple more transformers coming my way soon so I'll update the document when these have been tested and if anyone wants to loan me one of their favourite or not so favourite transformers to measure I would be more than happy to oblige.

Cheers

Ian


Edit: 5th Feb, updated document to include groupdiy members results.
Edit. 11th June 2012 corrected OEP information by Mathis D.
Edit:17th June 2012, added mouser transformer info provided by NY Dave

[mitsos]

Thank you Ian.  I haven't seen such a concise explanation anywhere else.  I guess I haven't read everything out there, but I've tried.

Maybe I missed it, but what frequency are you measuring L at?

[PRR]

Test conditions not specified except the meter brand/model.

Iron-core inductors' inductance varies a LOT with signal level, frequency, and DC.

May triple from "zero" signal to half of saturation. (See permeability curves.) (This is much less in gapped cores.)

May triple as you go down from 1KHz to 50Hz. ("Skin effects" throw the flux out of the center of each lamination.)

It might be more useful to set aside the inductance meter, drive transformer from sig-gen through "matching" resistance, at levels 1% 10% and 50% of rated, and plot frequency response.

[PRR]

"Agilent U1371C LCR meter" is perhaps the U1731A ?

http://cp.literature.agilent.com/litweb/pdf/U1731-90035.pdf

[ruffrecords]

Test conditions not specified except the meter brand/model.

Iron-core inductors' inductance varies a LOT with signal level, frequency, and DC.

I actually made measurements at the three frequencies the meter has available viz. 100Hz, 120Hz and 1KHz but only quoted the 100Hz values because they are the nearest to 20Hz.

May triple from "zero" signal to half of saturation. (See permeability curves.) (This is much less in gapped cores.)

Indeed. Measuring primary inductance is but one parameter.

May triple as you go down from 1KHz to 50Hz. ("Skin effects" throw the flux out of the center of each lamination.)

It does vary significantly from the 1KHz to 100Hz  that I can measure but the low frequency response is what I was most interested in.

It might be more useful to set aside the inductance meter, drive transformer from sig-gen through "matching" resistance, at levels 1% 10% and 50% of rated, and plot frequency response.

Quite right, and if I had the time I would do that too but I don't plan to do a Ph.D. in transformer characterisation just yet.. The point is that transformer manufacturers rarely if ever provide anything like that sort of information but I would very surprised if they did not have it. They could however quote primary inductance which would reveal a lot about how much they were prepared to spend on material costs and also a lot about what you would need to do to in order to design using it. Transformers are hellishly complex and I was just trying to demystify it a little and put some simple numbers to a key parameter.

Cheers

Ian

[bockaudio]

I feel you on this issue, especially @LF.
You really gotta use a bridge to do this. Modern LCR's, even my good one w/variable frequency/Q/DCR etc doesn't tell the whole story.

[ruffrecords]

"Agilent U1371C LCR meter" is perhaps the U1731A ?

http://cp.literature.agilent.com/litweb/pdf/U1731-90035.pdf

1731C. Sorry about the typo!

Cheers

Ian

[alexc]

Thanks Ian and contributors

All very helpful info and serves to bring the picture into a little more focus

Best regards

[Mzaar]

more turns of wire results in a larger inductance

Yep....but don't forget by doing so, you'd need a thinner wire for a same core size -> DC resistor goes up and more turns means more winding capacitance if not properly wrapped.

Yet for an ideal Tx in theory, DCR and winding capacitance =0 , winding reactance= infinity.

So it's all about compromise. 

[AZ999]

Hi Ian,

a question about primary inductance and some other ussually unknown parameters for mic inputs came up a few years ago. IIRC only Jensen and Cinemag gave max input level at 20Hz and described other parameters in their datasheets. Sowter only showed max level at 40 or 50Hz, i don't remember how it was with Lundahl.
I asked David at Cinemag to meassure primary inductance for their CMMI-7 mic input transformer. He kindly offered to "bridge" it for me and i got result a few days later. If you want i can try to find that email and let you know what he wrote.

[ruffrecords]

more turns of wire results in a larger inductance

Yep....but don't forget by doing so, you'd need a thinner wire for a same core size -> DC resistor goes up and more turns means more winding capacitance if not properly wrapped.

Or you could use the same gauge wire on a larger core but that would make it more expensive - which is just what I am trying to get at. I have a funny feeling that the cheaper manufacturers achieve their cost savings by cutting some of these corners. DCR is another good indicator of 'quality' so I think I will add that to my measurements.

Yet for an ideal Tx in theory, DCR and winding capacitance =0 , winding reactance= infinity.

So it's all about compromise. 

No argument about that, in fact I think you hit the nail on the head - the question is which manufacturers make which compromises and how do these relate to cost and performance?

Cheers

Ian

[ruffrecords]

Hi Ian,

a question about primary inductance and some other ussually unknown parameters for mic inputs came up a few years ago. IIRC only Jensen and Cinemag gave max input level at 20Hz and described other parameters in their datasheets. Sowter only showed max level at 40 or 50Hz, i don't remember how it was with Lundahl.
I asked David at Cinemag to meassure primary inductance for their CMMI-7 mic input transformer. He kindly offered to "bridge" it for me and i got result a few days later. If you want i can try to find that email and let you know what he wrote.

Yes, that would be very useful if you could find it.

Cheers

Ian

[ej_whyte]

Very interesting, thanks.

[zebra50]

Hi Ian,

This is good stuff! I have been chewing the other end of this lollipop recently, by rolling my own inductors and transformers.

I'm only doing a few, but it is apparent that for big runs the economics of the inductance are really important.
I found that 80% Nickel lams are around 2.5 times the price of 50%Ni/Fe, for the same quantity of lams (buying by the kilogram). And they give you around 2.5 times the inductance for the same size and number of turns.

So switching to the higher cost lams can either save space, or let you use fewer turns and bring the DCR down, which can be desirable in some applications.

The measured inductance does change remarkably with frequency, which does screw with the theory. Eventually of course you have to measure how it performs in the circuit.The document that Marik posted here is really helpful and worth linking again...

http://www.groupdiy.com/index.php?topic=47497.msg597825#msg597825

Of course the more subtle effects of saturation, hysteresis and the elusive 'tone' are harder to measure and understand.

Anyway, thanks again for sharing these results.

Stewart

[AZ999]

Yes, that would be very useful if you could find it.

Cheers

Ian

Ian;

i just found this email from 2007, primary inductance of Cinemag CMMI-7(C) is ~ 30H, meassured (bridged) by David from Cinemag.
Btw, original Redd47's input transformer had 15H of primary inductance.
I'm sure David from Cinemag will help you get more information about their transfomers if you tell him what you are up to. He kindly replied to many of my noob questions back then, analysed circuits (and described them!) to find correct transformers, etc. And i wasn't buying a lot!

[emrr]

Even a crude measurement will weed out many options quickly, and is useful.  Seems many older designs quoted performance with a matching source Z, while many new ones quote it with a bridging source Z, or a zero R source Z.  And they don't tell you about the source spec, which you learn the hard way the first time you try an old school matching interface. 

[gemini86]

Ian, if you were to get your hands on an OEP tx, I would love to see some data there... They seem to be in the same proce range as edcor, but slightly higher quality (and a bit cheaper). I do love that edcor is so small, their facebook page has some great pics of the facility and the employees seem to take pride in what they do.

[abbey road d enfer]

I found that 80% Nickel lams are around 2.5 times the price of 50%Ni/Fe, for the same quantity of lams (buying by the kilogram). And they give you around 2.5 times the inductance for the same size and number of turns.

True but it's only one side of the story. Ni saturates much more quickly than Fe, so if a transformer is designed for high level, such as an OT, for a same level, the Ni core may be larger than the Fe core. Fe is ran at typically 1.5T, Ni at 0.5. Ni has much lower intrinsic THD, but Fe operating with a very low source impedance (particularly using negative-impedance drive) can yield very good results though.
For mic input xfmr's, there is no dicussion, Ni is best.

[ruffrecords]

Ian, if you were to get your hands on an OEP tx, I would love to see some data there... They seem to be in the same proce range as edcor, but slightly higher quality (and a bit cheaper). I do love that edcor is so small, their facebook page has some great pics of the facility and the employees seem to take pride in what they do.

OEP are made here in the UK so perhaps one of my fellow Brits will loan me one to measure?

Cheers

Ian

[lassoharp]

i just found this email from 2007, primary inductance of Cinemag CMMI-7(C) is ~ 30H, meassured (bridged) by David from Cinemag.
Btw, original Redd47's input transformer had 15H of primary inductance.

That sounds huge for mic source impedances.

Did he happen to give you sec L as well?