DIY Powdered Iron Tranny Core

[valveitude]

I was up late last night (early morning?) working on my ball mill to pulvrize my iron, so when I sat down to check this thread before bed I was both sore and tired. I was hoping for a little inspiration,so when I got more of the "why bother it won't work" meme I sorta popped.
I didn't join this forum to engage in squibbling, I joined to share ideas. Re-reading this thread I can see we got off on the wrong foot...let’s try again

One thing I noticed, it seems some folks are posting without out actually reading my earlier posts. I say that because I seem to be getting some posts with points that had already been covered/answered.

I know what I am trying to do is far enough out of the diy norm that some skepticism is a healthy thing. What I hope to do in this post, is to lay out a little more clearly what it is I’m trying to accomplish , and show that I have the tools I need to do it.

One of the biggest frustrations I’ve had in the past is needing to do/make something only to find out midway I didn’t have access to a tool I needed. About a decade ago I set out to change that. To that end I have assembled a small hobby level machine shop, and aluminum sand casting foundry. I will never have every tool I want, but I can now make just about every tool I need (the ball mill I’m building now, for example) This has increased my fabricating capability’s several orders of magnitude . Now it’s a question of economy of time as to whether or not a project is worth it or not. I guess you can say I’m a uber-diyer.

Some folks seem to think I want to make trannies out of iron powder as a means to itself..it’s not.
First, IMHO silicon iron (steel) is still the best all-around material for OPT’s , medium permeability, and high saturation point (a little Nickel aint bad either). I settled on brake turnings because of its availability. Commercially available iron (not ferrite) powder is VERY expensive. Now if my tranny core idea(s) work out, I might make the plunge for the “good stuff” .

Why powder instead of laminations ? A couple of reasons. First, as everybody here probably knows, lots of thin laminations has better performance at the upper end of the spectrum, over fewer, thicker laminations. There are two reasons for this, 1) the reduction of losses to eddy currents, and 2) a smaller mass takes less time to align its molecules magnetically. Carried to its logical conclusion, the smallest possible mass/size, will have the fastest/loss-less magnetic path…hence powder. Secondly, by being able to mould the core, I can make any shape I want. The possible downside? Air-gap losses. No matter how tightly packed, or how fine the powder, there will still be a space occupied by the binder (epoxy). This isn’t a deal breaker for me. First this air-gap is distributed throughout the core, minimizing its effect, secondly, since I am building a SE amp with parallel 813’s, I’m going to have about 200ma of dc flowing through the primary. I NEED an air gap, a fairly large one at that. I’ll know it’s as good as it can be if I have to add an additional air gap. If not ,I’ll be forever tormented as to how much better it could be ( since its never good enough anyway, I’ll live).

To recap my methodology (again, I didn’t come up with this industry did, I’m just adapting here) I am rendering my materiel (brake turnings) into as fine a powder as I can. I am then oxidizing the surface of the powder to form an insulating layer. Next, the powder will be mulled with an epoxy binder, and rammed into a mould under about 1000-2000 psi with the aid of a hydraulic press…garnish and serve.

I hope this clears up a few things. I look forward to our future dialog.

[Jaka Racman]

Hi,

Magnetics makes E powdered cores. Here is the link.

I think you will find that use of powdered cores is not suitable for high linearity signal transformers. All comercially available powdered cores (and I think that also yours will be the same) have a highly nonlinear B-H curve, contrary to the discrete airgap cores, where B-H curve is higly linear. But you are free to try.

Best regards,

Jaka Racman

[pinkmouse]

I have no idea if this will work or not. But, I applaud you in pushing the boundaries of diy, and I can't wait to see your results. Positive or negative, ( though I hope the former, of course), you have the spirit of a true experimentalist. Good luck!

Though I would take careful note of some of the safety issues that have been raised.

[Sheldon]

Originally posted by smoking-amp
A source of HV power supplies and components (including xfmrs and capacitors) that is largely unknown in the tube audio realm are electrophoresis power supplies.

Electrophoresis. Now that's a word I didn't expect to see here. I spent 15 years of my professional career involved in PAGE, though on the wet side. Actually worked at Bio Rad (source of the supply in the e-bay ad) at one time in chromatography. Like the PS's in consumer stuff, most of the new gear is SMPs. Lighter more flexible.

Actually, a working electrophoresis ps might be a good bench supply for experimenting. All of them have a wide range of adjustable output and, like the one shown, many can be selected for constant current, constant voltage, or sometimes constant watts.

I'll have to check out e-bay more often.

Sheldon

[mwmkravchenko]

I can't figure out why you are making a ball mill when as an amateur foundryman you could use a rolling mill more for your sand mix and your crushing operation.

And I'm guessing that you have a hydraulic press or will rig up a jack to do the same. Good idea but you still may end up with air bubbles. Mix just enough to get the dough wet. No more than needed and you will entrain the least amount of air.

MArk

[valveitude]

Magnetics makes E powdered cores

This source wins the prize for the closest "off the shelf" solution so far..closest not exactly close. First off..it is a "high permeability alloy not optimized for OPT, second its an E- core..I don't want to be confined to one physical configuration. I want to experiment with a cup/c-core hybrid , think c-core that that wraps 360 deg. providing a magnetic shield.

I think you will find that use of powdered cores is not suitable for high linearity signal transformers. All comercially available powdered cores (and I think that also yours will be the same) have a highly nonlinear B-H curve

Correct.. ALL commercially available cores I've seen have this problem. This is the function of the core material not being simple Fe-Si. These guys are in the biz of making money (no problem there). When the market for tube OPT's becomes a multi-million a year industry (read never) they will pour their R&D into an alloy optimized for this purpose. Until then, they will continue to develop better and better cores for switchers, IF, and RF,and the like. That leaves wingnuts like me to try and fill the need..my need that is. To my knowledge (limited as it is ) there are NO commercially available powder cores suitable for OPT's.

I do appreciate the effort for the link though.

I can't figure out why you are making a ball mill when as an amateur foundryman you could use a rolling mill more for your sand mix and your crushing operation.

Simplicity..I considered roll mills, hammer mills et al. I decided on a ball mill, as it is something I can "scab" together quickly. In addition I think it has the advantage of working by concussion, (the "balls" flinging around smacking into the wall of the drum) as opposed to pressure that a roll mill uses. The limit of concusion rendering is limited only by the porosity of the material.

I have no idea if this will work or not. But, I applaud you in pushing the boundaries of diy, and I can't wait to see your results. Positive or negative, ( though I hope the former, of course), you have the spirit of a true experimentalist. Good luck!

Thank you.

[mwmkravchenko]

http://www.angelfire.com/tx5/hite/muller/muller.htm

Ideas at any rate

Mark

[Dave Cigna]

First, I want to applaud your diy spirit; let the others rehash what's already been done!

I've thought about making powder cores, though it never occurred to me to make my own powder. Maybe one sunny Sunday afternoon this summer I'll take a grinder to an old PS transformer.... One thought I've had is to cast the windings into the core. I imagine suspending the wound coils in an empty container and then adding the slurry. Obviously, that's a one-shot deal, and neither high pressures nor high curing temps could be tolerated.

Anyway, I just want to warn you that you might be underestimating the effect of the distributed gap. One way to think about it is in terms of reluctance. The net reluctance is the sum of the reluctances of the iron path and the gap. (Like adding resistors in series.) The reluctance is proportional to l/u where l is the path length and u is the perm. We're talking relative perm, so for air u=1.

So, if the perm of the iron is something like 1000, then adding a gap that is 5% as long as the iron path will increase the reluctance by roughly .05*1000= 50 times that of the iron alone. Restating that in terms of permeability means that you'll end up with an effective perm of something like 1000/50 = 20. Make the gap 10% of the total path and you'll end up with an effective perm of 10. That's still 10 times better than air but it might not be what you were expecting.

My hunch is that if you start with a fine enough powder then eddy currents will not be too big an issue even if you use a very thin binder and maybe even skip the oxidation step. It will take only a tiny fraction of an ohm electrical resistance between grains to reduce eddy currents. Obviously, I haven't done any real analysis, just offering my intuition.

[smoking-amp]

Here's a technique that a home Diy-er could use to make an unconventional output xfmr with possibly very good results. Start with a spool of Litz wire with maybe 20 strands of wire in it. (I see Litz wire on Ebay regularly) Unserved Litz is fine. (the serving on Litz is the fiber wrapping around the wire strands to hold them together and provide extra insulation)

Now you need a spool of approx. 1/2 inch wide amorphous magnetic alloy tape. Most likely have to buy this new from the manufacturer. This stuff is very thin. Be careful when handling this stuff, its got razor sharp edges on it. Now you need a rig that can spin the tape spool around the wire as the wire gets pulled thru. Essentially, one would be putting an amorphous film serve on the Litz wire. Can experiment with how dense a spiral layer gets put on by altering wire pull rate versus tape spool spin rate.

The final output xfmr. would be configured by connecting half of the Litz wire strands (the little wires in the Litz) in series for the primary and half the strands in parallel for the secondary. Should probably use triple enameled wire strands to reduce capacitance between "turns", but at least leakage inductance will be near zilch. (its the product of leakage inductance and distributed capacitance that determins xfmr bandwidth) Another approach to lower distributed capacitance further would be to use two (or more) Litz wires with serving on at least one or more of them. Wrap the amorphous tape around both (or more) Litz wires. One Litz wire gets used for primary turns and the other for the secondary.

The whole final length of Litz could be coiled up for compactness, but this won't have any effect on magnetic coupling. The equivalent of this technique is often used to make broadband RF transformers using ferrite beads strung on multiple wires.

You have to get enough inductance in the primary to handle the lowest frequency. While this technique does not get to take full advantage of a large number of turns to get inductance, it does take advantage of minimal magnetic path length around the wire circumference, which will provide a large boost to inductance.
Inductance = u*N*N*Area/(magn. path length)

Don

[PRR]

Ignore my discouraging remarks. If you are right, all the nay-sayers in the world can't make you wrong.

The lack of positive ideas is, of course, because you are working SO far outside the norm that none of us have caught up with you. Most of these thoughts must have crossed your mind too. Give us time to argue out loud. A new idea-seed needs time to grow.

> as everybody here probably knows, lots of thin laminations has better performance at the upper end of the spectrum, over fewer, thicker laminations. There are two reasons for this, 1) the reduction of losses to eddy currents, and.... Carried to its logical conclusion, the smallest possible mass/size, will have the fastest/loss-less magnetic path…hence powder.

To nay-say some more (to make you sure of your analysis): there is a diminishing-returns kink in the optimization. We use one lam-thickness for 50-400Hz power, somewhat thinner lams for HiFi. Thinner is better, yes. But at any specified frequency (and material resistivity) there is a size that gives "small" losses, and a thinner size that is "insignificantly" less loss, but significantly higher price to produce.

Roll or grind small, yes; but rolling/grinding "too small" is not better and consumes money/time that could be better spent on other imperfections. (Yes, in uber-DIY, money/time takes a different perspective; us hasty-hackers don't often commit enough hours or dollars to our dreams.)

Note that eddy-current is a 2-dimensional problem. It is a function of the smallest circle that can be drawn perpendicular to the flux lines. Indeed, long broad thin sheets have the smallest eddy circles per fabrication dollar. Drawing to wire or grinding to powder is much more work. On the eddy-losses arguement, you should be looking for thin foil, not powder. Of course, at some point the many layers of insulation reduce your core effective area.

Have you considered using the brake-shavings directly? Their thickness is somewhat thinner than "good audio" transformer lamination. Your cost of fabricating to that thickness is, as you say, zero (prepaid by brake-shop victims). Stacking-factor could be horrible without much shaking and pounding.

Oxidation is now the standard insulation for power transformers. It can be much thinner than varnish, won't burn. They even shave that by oxidizing just-enough to meet a loss spec, not to reduce loss to "zero".

Is oxidation needed when you have a non-conductive binder? A simple resistivity test on a scrap blob will quickly tell you if it is low-R, high-R, or too in-between to judge without further test.

Sintering sounds wrong (and darn hard to DIY sintered iron). It clearly increases the eddy-circle size bigger than the particle size. But if the sintering is light, the contacts are microscopic. Current-crowding at the contacts might give sufficiently high resistivity to make eddy loss small.

Hammer-out a closed core. It does not have to be a final product. Wrap a few turns, take the measurements, extrapolate the winding resistance for a full winding. My best-guess is that inductance and saturation will be lower than iron, copper loss higher. The other issue would be "fine sound": different irons (and lamination thickness and insulation) have different sound. The ear is a very fine instrument. If the powder material has a "better sound", the size and cost are secondary issues (or even features) for many audiophiles.

I say "copper loss" from habit. But since this is clearly a very specialized process for a well-heeled market (even a market of one), I'd be impressed even if you have to use Silver winding to get performance comparable to iron/copper materials. At this level of investment, Silver wire is an obvious detail anyway.

> 2) a smaller mass takes less time to align its molecules magnetically.

Will the entire particle flop as a unit? I believe magnetic domains are smaller than grains of powder, though I certainly am over my head here. Maybe there is coupling between domains, less so across particle boundaries?

IS there a time element? Being a mere audio-hack, I believe domain-flop is only about flux, not time. That's clearly wrong: nothing happens in zero time. But is the time anywhere near audio rate?

> by being able to mould the core, I can make any shape I want

What is the ideal shape? It is a many-splendored problem. But for the big power transformer business, optimization is profit, and quantity is high enough to forge novel processes.

The square-stack E-I core we usually use makes the winding wire take a long path. Graduated widths can approximate a round core, round windings, least winding length (resistance) for a given core area. GE used to strip-wind iron sheet cut on a taper to get this core-shape.

But big-power is different from audio. In POWER, reducing loss from 10% to 5% means half the cooling surface needed. MVA transformers are a lot about cooling, so half the cooling is a very significant advantage. But in audio, reducing loss from 10% to 5% means 0.5dB more output, inaudible.

Topologically, all the closed-core forms reduce to two conjoined toroids. In a common cheap tranny, the iron is a square toroid, the coil is another toroid. The squareness is not electrically optimum, but convenient for production. In a "toroid tranny" the core is round doughnut shape, the coil is a toroid stretched-out around the circumference. Good, though the winding can get a little tight inside the doughnut. This could be improved with vari-diameter wire: thin where it passes through the hole, fat around the other 90% of its length; this would be tough to produce.

We come down to two key dimensions. The mean length of turn is constrained by the core area needed. And the mean length of flux-path is constrained by winding window area needed. Both interact with each other, and with cost of copper versus iron, and design frequency and losses. The general optimization (ignoring cost) seems to be the common toroid, or the inverse (copper doughnut wrapped with iron; pot-core).

Offhand, I assert that either shape can be approximated to 60% perfection with plain square stamped iron, or to 80%(+) perfection with tape-wound(tapered) toroid.

What shape are you going for? Maybe there is an outside-the-box shape I can't imagine.

Silicon: as I said, in foundry work, silicon makes iron more fluid, easier casting of complex shapes. This is of course irrelevant to transformer steel which is brutally rolled, not cast. In transformer steel, increasing silicon from 1% to 4% increases resistivity, cost, and brittleness. 4% Si iron has about half the (power frequency) loss of 1% steel. This has to be weighed against cost of steel, cost of punching, and the possibility of using a thinner lam of cheaper steel. AFAICT, the resistivity is the main point of Silicon in transformer iron; if you drastically reduce the eddy-circles with a powder technique, is this still an important factor?