Amorphous alloy tape wound toroid for ribbon mic transformer ...

[pyjaman]

Hello, everybody.
I have this ribbon mics project on the bench for months now. basically, the motor is made like this:
gap/ribbon lenght: 50mm
gap width: 5mm
gap depth: 5mm
no pole pieces, two "N50" grade 5x10x50mm neo magnets sit directly on each side of the gap.
ribbon material is 2,5 micrometer Al.

So, in order to keep this project as DIY as possible and to save money (!) I'm looking for a way to wind my own transformers. In my quest for a suitable core, I found this:
http://www.spulen.com/shop/product_info.php?products_id=140
datasheet here :
http://www.toshiba-tmat.co.jp/eng/product/pdf/e_amo006.pdf

Basically, my idea was to experiment with two 9 or 10 turns of 0,9mm dia primaries paralleled in a humbucking arrangement and two 130-150 turns of 0,1 to 0,3mm dia secondaries in serie.

Unfortunatelly, My maths skills, english and background theory are too weak to allow me to fully understand everything about it.

I may be totally on the wrong way, but I know that some people here are very good at ribbon transformer theory and experimentation (Marik?) and Magnetics (CJ?).

So any inputs about the suitability of these cores for my project, any suggestion about the winding arrangement, or any other suggestion for other core sourcing WILL BE HIGHLY WELCOME.

Thanks in advance for your help!!!

Laurent.

 

[PRR]

You want your secondary to be "150 ohms". Transformer is inductor, impedance falls at low frequency. Select the lowest frequency you really "need", find enough inductance to exceed 150 ohms at that frequency.

Or design like a mike input: over 2K impedance at 50Hz. This is "no"-load on 150 ohm mikes over the useful audio band. However that impedance may be hard to reach. And too-large impedance probably means too-small wire and excess copper loss.

You don't need a "perfect" answer. The cost of core and wire is so small, that you can wind one and try it. If it could be better, you can cut-off the wire and wind again; if you can't cut it off, a new core costs less than a lunch.

I *think* e_amo006.pdf shows the actual impedance in the graph ohms per n squared (bottom left). They plot for 1KHz on up. I have assumed (see below image) that impedance falls linearly with lower frequency. This may be wrong: they show a change of slope for >20KHz. But plain iron core does the same thing, only at 400Hz; we may generally estimate the bass response with a linear approximation.

At 50Hz one turn gives 0.1 ohm impedance. You want 200 to 2,000 ohms at the secondary. You want 2,000X to 20,000X more inductance. Take the square root: 45 to 141 turns.

Marik may have more to say about ribbon impedance. Using general electrical principles, you want a winding ratio to lever your ribbon (plus leads and primary) resistance to be similar to a dynamic mike, 100 to 300 ohms. If you have 1 ohms in the ribbon system and aim for 150 ohms you want 1:12 ratio; if 0.2 ohms to 600 ohms then 1:55 ratio. This is a tradeoff between output voltage and loading effects.

For experimentation, 5+5 turns primary and 100+100 turns secondary seems to give usable inductance and several useful ratios.

Pick wire size to almost fill the core, giving equal space for each winding. Bigger wire gives lower copper loss, but it will be hard to bend 1mm wire around a 7mm core. Aim to have secondary copper resistance less than 150 ohms, preferably near 50 ohms.

No fancy winding technique is needed for audio frequency response. Low impedance means capacitance does not matter, and the high permeability and light preamp loading means leakage inductance is not a problem. Just get the wire through the hole. You may however want to consider two strands of 0.63mm wire instead of one strand of 0.9mm wire; easier to bend.

 

[PRR]

Argh! I screwed-up that chart. One turn impedance at 50Hz is only 0.01 ohms. You need 3 times as many turns as I computed, or accept lower impedance. You do the math; I'm obviously too tired to do it right.

 

[pyjaman]

Never mind ! Thanks a lot !

Your help is still precious and enlighting. Now I think I get it. Based on that curve (and your estimation of its behaviour down to 50hz) I could reach around 1 ohm imp. at 50hz with ten turns pri. (might be enough regarding the thickness and width of my Ribbon, if good connections can be achieved). Or even down to about 30hz with 14 turns... but now, If I want around 30db gain, I get around 900 ohm secondary impédance (assuming this 1 Ohm motor impédance wich certainly can be overestimated or/and lowered).

So this is a start. I ordered the cores today and see what I can do. Starting from a tenth of turns and see if I can lower it to lessen copper resistance (enough to make 4 primaries paralleled fit could be cool).

Now what I'm missing is an estimation of the motor (as described in the first post) impedance, witch could help a lot to make better calculations.

 

[CJ]

put the ribbon motor on a scope and see what voltage when screaming.

i do not know how you would load a ribbon motor for this test, it might be a matter of measuring current, not voltage,

hook a raw vu meter up to the motor, if you can peg it, you have 200 uA, etc.

then, once you get your current and voltage levels, you can pick the right core.

you want to balance the lowest pri dcr vs turns vs core size

so if you get the core size, the turns and dcr fall right into place.

it would probably be hard to saturate even the smallest core with a ribbon,

i bet that 23 mm OD is a bit large and all your vocal details will get sucked up by the core, but it is amorphous, sp you need a bigger core per it's alloy partner, 80 Ni, in order to get the same pwr handling.

the amorphous has more of a natural airgap, it is more delicate than 80ni, so you can not tape wind the core as  tight.

i would start with a 80 Ni F lam, (self shielding)  until you get your gauss and turns all figured out, them maybe transfer the data onto a toroid, if your heart is stuck on the noise rejection.

it is a lot easier to bobbin wind a quck layer, rather than hand wind a bunch of toroids

plus, with a bobbin, you can wind the copper foil pri, which is ultra low dcr and which also lays really flat, so you can get more copper than with wire.

it is hard to use coper tape on a toroid, you have to insulate every turn, so a bobbin is better.

you want the window to be as much copper as possible vs air.

start with a MC type transformer, look up a Peerless K-241-D, it is a good MC xfmr,

then scale the core back a  notch, copy the turns, i bet you would have a dang nice starting point,

do you need hum bucking these days? just shield the heck out of it,

but not too close, the mu metal will steal some flux if you get too close.

here is the 241 D info, you want the 11 ohm tap,

some company wanted my domain name so i got leveraged out of my website.  ???

OK, structure:

1/2 pri--sec--1/2 pri

Single coil, bi-filar wound primary, , pri is 42 ga wire, sec is 48 gauge wire.

Pri turns: 492
Sec turns: 6022

Two primaries, two secondaries.

DCR:
each pri- 17,8 ohms, tap at 11 ohms.
each sec-1.465 k ohms-no taps

Inductance:

each pri-2680 milliHenries-tap at 481 mH
each sec-340 henries

Leakage:

Capacitance-115 pf from any pri lead to any sec lead.

Inductance: 1.51 mH with sec shorted

Total weight: 13.42 oz.  Coil/core: 5.37 oz.

Three mu can construction.

1:12 turns ratio

 

[PRR]

> each sec-340 henries

That sounds like a 47K phono pickup interface.

We want a 150 ohm dynamic mike interface.

> 48 gauge wire

!!!

 

[Marik]

Hello, everybody.
I have this ribbon mics project on the bench for months now. basically, the motor is made like this:
gap/ribbon lenght: 50mm
gap width: 5mm
gap depth: 5mm
no pole pieces, two "N50" grade 5x10x50mm neo magnets sit directly on each side of the gap.
ribbon material is 2,5 micrometer Al.

With that ribbon size and thickness you will probably get about 0.1-0.15 Ohm.
The magnets are of a good size, and depending on the frame you make you will get some 0.4T-0.5T in the gap. Your sensitivity will depend not only on that value, but also on the front-to-back signal path. Doubling it will double the sensitivity, but will also halve the top corner frequency, so pick your compromise.

So, in order to keep this project as DIY as possible and to save money (!) I'm looking for a way to wind my own transformers. In my quest for a suitable core, I found this:
http://www.spulen.com/shop/product_info.php?products_id=140
datasheet here :
http://www.toshiba-tmat.co.jp/eng/product/pdf/e_amo006.pdf

Basically, my idea was to experiment with two 9 or 10 turns of 0,9mm dia primaries paralleled in a humbucking arrangement and two 130-150 turns of 0,1 to 0,3mm dia secondaries in serie.

Hard to tell. It will depend on how you make things, what is the tolerance of the ribbon installation, tuning, etc.
Get a Radiotron 4th edition (I think PRR once posted a link to the online copy), open transformer section, and you will get all fomulae you need to calculate the turns.

Most likely, you won't need humbucking, as the toroids have very good rejection.
If you are doing the torroid by hand expect at least an hour or two to get your secondary done. I promise, you will curse everything during that time   8)
The EI, or UI are much easier to experiment if you don't have toroidal winder.

Best, M

 

[PRR]

> Radiotron ... all fomulae you need to calculate

Yeah, but for bulk iron stamped and stacked to your liking.

These pre-sized cores seem to have a short-cut to find the turns. "Impedance VS. Frequency" shows an inductive impedance (up to a point past 10KHz) in units of Ohms over square of turns. If we assume one turn, it shows impedance directly.

So (using the blue/black curves, small core), one turn gives 0.2 ohms impedance at 1KHz. This would be a suitable value for a 0.1-0.15 ohm ribbon, but we want to go deeper. 40Hz is 25 times deeper. Square-of-turns must be 25 times higher. So we need 5 turns primary. That would seem to give 0.2 ohms at 40Hz. Then to get 200 ohms secondary at 40Hz we need square-of-turns 1000 times higher, or 31 times as many turns, or 155 turns.

155 turns around that core is about 9 feet of wire. Without a toroid threader, I'd be inclined to try three 3-foot strands, so I could reach the far end while holding the core. Like hand sewing a button. Then solder in series.

 

[Marik]

155 turns around that core is about 9 feet of wire. Without a toroid threader, I'd be inclined to try three 3-foot strands, so I could reach the far end while holding the core. Like hand sewing a button. Then solder in series.

Can be messy, painfully slow, and most likely end up with nuts. With hand-wound I used a shuttle--long and flat bar with V grooves on the ends. Aluminum works better than plastic, as it doesn't bend when you put the wire on. Bifilar will speed up the process. That way I made about 5 and then bought a small toroidal winder. Helpful, but still, inaccurate and... messy. 
I ended up getting Gorman winder with 4" belt head, similar to this:

http://www.gormanmachine.com/videos/proii/proiivid.htm

Pain to set, but once done works beautifully. Really, those small cores are meant to be wound with an automatic, or CNC winder.

Best, M 

 

[pyjaman]

Hi, and thanks a lot to join this topic, gentlemen. I'm really grateful to have your interest on the subject (especially you 3, who are of my major "kind-of-forum-heroes"  ).
I've been REALLY busy for work these days, so it has been very frustrating not to participate more to the discussion. Today has been a loooooong day too and I'm almost sleeping, but who cares...
Anyway:

RDH4... : Maths and foreign language... Sometimes I wish my brain were younger again... But I definitely have to go back to it...

UI : Yes, much simpler to wind for experimentation, but I can't get my hand on any small UI NI core...
Anyway, those toroids are ordered already, so let's go ! I ever wound one (power trans secondary : 160v 100mA) so I know it's a PITA. I  have a kind of technique that "works" for that, copying the way automatic winders work (arranging the wire as a large "ring" that goes through the hole of the toroid).
Since it's for DIY and a 2 pieces job, I'll take the time...

Any considerations for pri/sec arrangement (interleaving,  P-S, S-P, P-S-P, S-P-S...) ??? I've read many different things about that...

Still many things to think about, learn and experiment...
I'm dead-tired now...
almost can't read myself.
I'll be back.

Laurent.

 

[PRR]

> Any considerations for pri/sec arrangement

No. Just get the wire on there.

Whatever you do (without waste space), the capacitance will be a few hundred pFd. Say 300pFd against a 0.1 ohm ribbon: it looks like 5 GHz on my calculator. Check my math, we already know I spill decimal-points, but I am sure that even on the secondary side the C will not give trouble until MegaHertz.

The other problem is leakage inductance, the amount of primary inductance which does not couple the secondary. For an air-core winding this may be large, similar to total inductance. On iron-core, as long as both windings are on the same leg of the iron, the mutual inductance rises as the permeability while the leakage inductance stays the same. So with permeability near 100,000 we expect leakage inductance to be 1/100,000 of mutual inductance, and loaded bandwidth to be like 100,000. Audio band is only 1,000:1, so any reasonable choice of bass limit will imply leakage inductance which will not give trouble until far above the audio band.

When you get into 5K and 50K impedance windings on lower permeability iron, the leakage inductance may give some trouble but the real problem is that the self-capacitance against the leakage inductance gives a resonance which is lower than you expect and very hard to control. A 4K winding on plain silicon iron (perm ~~ 1,000) will give 40Hz-15KHz or more even with non-interleaved winding; an 8K winding for anything better than AM radio will want a little interleaving, and to get a 10K winding on large iron smooth and flat to well beyond 50KHz (for heavy NFB) Williamson used a dozen interleaves (which turned out to be overkill).

But at <=200 ohms and no supersonic NFB, just put the wires on the core. If you can put one winding all the way around, then put the other on top all the way around, that might be best (more supersonic response). But if the wires won't fit that way, just get them all through the hole.

 

[CJ]

wind it

spectacles > wallet  > watch  > testicles 

or, translated, use four pies for the sec,

1 at 12 oclock, 1 at 3 oclock, 1 at 9 oclock, and the last one opposite the first, 6 oclock.

coupling with a toroid is great, so the leakage L is no biggy.  it's always the leakage C with the toroids, but you do not have that many turns.

when you get up past 2,000 turns, then you really have a tough time with a toroid..

 

[Marik]

wind it

spectacles > wallet  > watch  > testicles  

or, translated, use four pies for the sec,

1 at 12 oclock, 1 at 3 oclock, 1 at 9 oclock, and the last one opposite the first, 6 oclock.

Depending on the wire gage I'd calculate so that it runs whether one direction, or goes back-forth, again evenly covering the core.

coupling with a toroid is great, so the leakage L is no biggy.  it's always the leakage C with the toroids, but you do not have that many turns.

when you get up past 2,000 turns, then you really have a tough time with a toroid..

With that kind of permeability you'd rarely want to go past 2,000 turns... unless you want to get well past 400H   With tape wound core the toroids essentially don't have an air gap, so the losses are pretty small.
Anyway, the sectionizing, or even progressive sectionizing is not a problem.

Best, M

 

[CJ]

has anybody measured the ergs coming out of your best ribbon motor?

what kind of volts and amps are we talkin here?

if you decide just to put on the sec. in an even fashion, you can play with the laps,

try winding first with three laps around the ring to make your turns count, then six.

there should be a roll off difference, find the best way for HF response.

if you only have a few turns on your pri, you could try adhesive copper foil tape,

just make sure there is space between each turn.

this will give you real good tight coupling of the small mic signal to the core.

then what happens after that does not matter ,

 

[Marik]

has anybody measured the ergs coming out of your best ribbon motor?

what kind of volts and amps are we talkin here?

Look at the AEA sensitivity, keeping in mind a 1:28 transformer ratio.

Best, M