Building a high quality gainclone toroid coupling transformer with "tubeness" winding

Building a high quality gainclone toroid coupling transformer with "tubeness" winding

Since many of us have already walked the long hard road to the perfect coupling cap, it's a pleasant idea to start trying out with transformer coupled gainclones, as these, in opposite to coupling caps, can be diyed!

I'm planning to build a toroidal input coupling transformer (1:0.7 to 1:3, prefering 1:1) and face many choices:

· Which core material, target inductance, number of windings and size would be ok for a 1 Vrms output signal amplitude?

· I have many computer smps toroid cores, would these give good audio quality?

· I plan to add a tubeness winding to create a constant field flowing trough the core and add 2nd and 3rd harmonic (Yes, true tubeness as real tube amplifiers have transformers with constant dc bias and this does contribute to it's sound signature). This winding would be loaded with a ccs (Niceness knob), any suggestions regarding number of turns and current? Near-saturation behaviour is very material-dependant, any advise?

· Would a faraday shield (putting the toroid inside a metal candy box) and additional mu-metal shielding arround be enough to keep it dead quiet? Should the faraday shield be connected to power ground?

Thanks. I will post sonic feedback when completed.
 

BWRX

Ex-Moderator
2005-01-17 5:29 am
Pennsylvania
This is definitely an interesting topic for discussion.

You will probably need small diameter wire and many turns to make an input transformer. Small diameter wire, many turns, and toroid cores are not the most DIY friendly combination but it could be done if you have a lot of patience! I'm not sure what kind of core material you'd want to use but it should be suitable for use in the 20Hz-20kHz range (I'm sure you already knew that!). Different core material will give different saturation characteristics as you said. Iron cores will provide a nice rounded characteristic as saturation is approached.

You could add foil shields between windings and external mu metal shielding but why bother if you're just experimenting. It would be beneficial to use them in a "final" version if/when you get to that stage ;)
 
Thanks for the reply, getting down to 20 Hz means winding a lot, and i've noticed the problem with small cores :(. Thanks for your interest, it will help me in keeping windinding.

My first calculations give 870 turns (aprox) for a -3dB point of 20 Hz on a 47K load (this seems a sensible value for the input impedance in a power amplifier) using with a radius of 1.5 cm and 0.5 cm^2 section. Sounds reasonable?

Brian, can I PM you for a suggestion?
 
Thanks, i will try to get a better core before doing the 870 x2 turns. The computer power supplies i've seen come with two kinds of cores: ferrites and dull, green-painted ones. May these be powdered iron cores? Not all of them where used as transformers, some were chokes.

The idea is not to buy it done, but to try different arrangements and load impedances plus the third winding to slightly saturate the core.
 
The entry barriers here are pretty high. Having hand wound more than my share of toroids, I wouldn't want to do 800+ turns without a winding machine. And toroid winding machines are hard to come by outside commercial transformer companies. IMO, toroids are no magic answer to transformer design, and you'll have more properties to play with if you go with a conventional laminated design. Another big problem is finding anyone who will sell small quantities of the very high grade steel laminations you'll want. It might be worth looking at commercial line and MC transformers, and seeing if there's enough room to add a winding. There is nothing SMPS related that's useful for wide band designs. One interesting idea is the use of stacked toroids of different types to extend the range, but I don't know if it applies to audio frequencies.
 
I know that it's a hard job, and that maybe isn't worth the effort, but i'm interested in trying it.

Any more advise in core materials? Could a larger toroid be used here?

I know that the devotion that toroids have in PSU transformers is quite exagerate, but since it seems that they do give a wider bandwith (A reason not to use them in power supply units) maybe it's easier to get good results using them in audio.

I belive that the core materials in the SMPS psu could be anything from good to horribe, because they are very different from one model to another.

Is there any way to check the frequency response of a core other than winding some turns and checking HF performance? Will this give a good idea about the quality of the core? Do you know anywhere to buy good ones?

I feel the toroid approach to be more straightworward as it does not need to be glued.
 
You might be able to get a small power transformer, like PC-mount mini-60Hz ones (with E-I laminations), and rewind it for audio applications. Give it a few turns and test inductance to find AL, then you'll have an idea of core properties, or just go googling for typical silicon-steel properties. The mini pc-mount ones will run you a few bucks and then you don't have to go hunting for laminations. I have never tried to make a signal transformer like this, so I couldn't really tell you how well it will work, but it might be something fun to try.

One this is for sure, you'll want to keep your core as small as possible for your power rating (real small), or your leakage inductances and parasitic capacitances might become significant.

Toroids are nice in that they don't have to be glued, but they do usually have highish interwinding capacitance, which may affect response. Not a big issue for smps, but could be important here. I haven't seen too many available toroids made of tapewound steel, which would probably be needed with audio frequencies. You could try with a mini torioid power transformer core similar to the method above.

good luck
 
Re: Building a high quality gainclone toroid coupling transformer with "tubeness" win

ionomolo said:
Since many of us have already walked the long hard road to the perfect coupling cap, it's a pleasant idea to start trying out with transformer coupled gainclones, as these, in opposite to coupling caps, can be diyed!

Well, yeah, one could get a core, a bobbin, and wind some wire around it. And technically it would "work." But unless you're just wanting to do something yourself just for the sake of it and don't really care much about the quality of the result, there's a bit more to winding a good quality transformer than simply wrapping wire around a bobbin.

If money's an issue, you can just get a cheap Edcor WSM10K/10K for $10 and end up with something better than you'll get just wrapping wire around a bobbin.

· I plan to add a tubeness winding to create a constant field flowing trough the core and add 2nd and 3rd harmonic (Yes, true tubeness as real tube amplifiers have transformers with constant dc bias and this does contribute to it's sound signature).

Well, sort of.

In a typical push-pull tube amp, while you have a constant DC current flowing through the primary, the currents flow in opposite directions and the fields cancel, effectively leaving you with no net flux in the core.

A typical single-ended tube amp also has a constant DC current flowing through the primary, but it's unidirectional and there's no field cancellation as there is in a push-pull amp. For single-ended tube amps, the transformer's core is made with an intentional air gap to keep it from saturating. So if you're wanting to emulate something like this, you'll need to add an air gap in the core.

se
 
Originally posted by Steve Eddy
Well, yeah, one could get a core, a bobbin, and wind some wire around it. And technically it would "work." But unless you're just wanting to do something yourself just for the sake of it and don't really care much about the quality of the result, there's a bit more to winding a good quality transformer than simply wrapping wire around a bobbin.

I do for both, dying is fun but it's also compatible with trying to get good results. Money is not an issue since i could recycle the transformer for other designs. I do know how a homebrew electromagnet looks and i plan to wire it better even if it takes weeks.

You are right about fields cancelling in push-pull amplifiers, but the sound all of we love is SET, wrong? The tube comparison was almost a joke but it does not mean it wouldn't sound tubey and the idea of a controlable amount of coloration is really appealing.

Good shaped, well-paralleled turns is enough or it is important to control other factors?

In my understanding the transformer acts like a waveguide and keeps (and distorts) the field inside it. If wires don't cross themselves the dominant factor regarding distortion will be the material response. Inter-wire capacitances may also become very ugly at HF, but i expect this not to be a problem if there aren't many layers involved (the voltage difference between turns should be low.

I would like to do my own transformers as small "masterpieces" to add something of my own to the standard gainclone/LM4702/LME49810/LME49811/LME49830 amplifiers i build.
 
ionomolo said:
You are right about fields cancelling in push-pull amplifiers, but the sound all of we love is SET, wrong?

Well, many do love the sound of SETs, but many also love the sound of push-pull.

The tube comparison was almost a joke but it does not mean it wouldn't sound tubey and the idea of a controlable amount of coloration is really appealing.

Yes. I'd given some thought in the past to using that technique for an electric guitar distortion box. Never got 'round to experimenting with it though.

Good shaped, well-paralleled turns is enough or it is important to control other factors?

You also want to control for parasitics, mainly winding capacitance. That's where layering techniques comes in. You can read more about this in the Radiotron Designer's Handbook (before you click, it's rather big, 25MB).

The easiest to build would be a bifilar, but that precludes a Faraday shield which is commonly used with input transformers to improve common-mode rejection performance.

In my understanding the transformer acts like a waveguide and keeps (and distorts) the field inside it.

I wouldn't call it a "waveguide," even in loose terms.

But yeah, ideally all of the fields are contained within it. Of course realworld transformers aren't ideal so you won't get prefect coupling and what doesn't get coupled manifests itself as leakage inductance, which is one of the parasitics of transformers and contributes to limiting high frequency performance along with winding capacitance.

I would like to do my own transformers as small "masterpieces" to add something of my own to the standard gainclone/LM4702/LME49810/LME49811/LME49830 amplifiers i build.

Nothing wrong with that. In fact I find it admirable. Just letting you know what you're up against if you're looking to create a good quality result at the end of it.

In addition to winding techniques mentioned above, a good input transformer typically uses wire on the order of 42 gauge in order to achieve sufficient turns to get enough primary inductance suitable for a high impedance input transformer.

Not only is the stuff difficult just to see (it's only slightly larger in diameter than a human hair), it also likes to break if you breathe on it too hard.

But if you're willing to give it a go, more power to you!

Good luck!

se
 
Since my bedroom amplifier is a small gaincone i'm thinking in doing something easier to start with transformers. The idea is to do a coupling transformer to use with my ipod. This is intended to drive 32 Ohms impedances and this makes winding really easier, and i hope it will reduce the deep hum that the amplifier produces when the cable is floating and the slight buzz that it shows when the ipod is turned off (of course wiring is OK since it produces absolutely no hum with the ipod turned on and paused).

The problem i've found is that ipod earbuds show 32 Ohms resistance, so they aren't anything near a pure inductive load. Since there aren't any cone resonances involved in DC operation, does this mean that the wire used really shows that resistance? Is there any way to mimic this behaviour in a transformer? The only way i see to archieve this is to put a 32 Ohms resistor bypassed with a huge cap in series with the transformer, but this looks like a really bad idea.

I thought this was a great idea because i could start trying with only 45 turs (theoretically) and check how much headroom should be used to get an undistorted bass response and compare different core materials.

Thanks in advance.