• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Transformers for tweeter amps on ferrite donuts

Hi!

I have some common mode input chokes from power supplies. They are wound on donuts with 20 mm internal diameter, 40 mm external diameter, 12 mm thick.
Each winding has 41 turn, 18.2 milli Henry.

The question is, what material can it be? What is saturation induction, and how much power I can pass through them for a power amp from 3 kHz frequency?
 
Well, common mode chokes are often wound on higher permeability ferrites (mu ~5000 or more) to allow a large amount of common mode inductance with relatively few turns. For the higher perm ferrites, saturation flux is usually less than that for power materials. Losses are higher, too, but that will hardly be a concern for an audio transformer. I back calculated and got an inductance coefficient of 10,000 for the core, meaning that it probably is made with high permeability ferrite. To get a ball-park idea of what you're dealing with, you can take the dimensions of the core and compare with the specs of other cores with similar dimensions. The Ferroxcube and Magnetics, Inc. web pages are probably a good place to start. A picture of the core here would help, as I might have a EWAG as the the maker, judging from the color and overall shape. They may even be marked.... Also, toroid core shapes/dimensions (at least for ferrites) are less standardized among manufacturers than other shapes like E cores, so often the exact dimensions of a core can offer some clue as to who made it.

These could possibly work to make a high frequency push-pull transformer to drive tweeters. You basically have two concerns, the first that you have enough turns to support the required volt-seconds, in order to avoid saturation. The second is that you have enough primary inductance so that you are not stealing a lot of power that should be going to the load in order to drive the magnetizing inductance.

The power you can pass through the cores will depend on the volt-seconds and the amount of copper you can load on. I don't think the core losses will be of any great consequence at audio frequencies. If you wind for a max peak flux density of 2500 gauss or so each way at max drive level, minimum operating frequency, you're likely going to be safe, as that's a very conservative level even for wimpy high-perm ferrite materials. Winding for this flux density also will most likely ensure enough primary inductance to get the job done.
 
Last edited:
L = Mu (4 pi N^2 A 10^-9)/le

L in Henries
A core area in cm^2
N turns
le magnetic path length in cm
pi = 3.14159
Can figure the Mu from this and your data

Usual power ferrite material will be from 2000 to 5000 Mu
Bsat around .35 Tesla (thats useable Bmax, the max spec maybe more like .45 to .50)

10^4 Gauss = 1 Tesla

for design:
B = (V 10^4 )/(4.44 f N A)

B in Tesla
V in Volts
f in Hertz
N turns
A core area in cm^2

You can figure on 4 to 5 times larger area core needed in ferrite than in steel (toroid) to do the same job.

edit: at 10,000 Mu it would be about twice the losses of the usual power ferrites, maybe .4 Bmax Tesla (spec) and .3 Tesla Bmax useable. For comparison, Bmax spec for Ni permalloy 80 is around .7T and steel is around 1.8T I think. 1.5T Bmax useable is what I see often mentioned for grain oriented toroids for audio. For EI core its more like .7T for real HiFi OT and 1.8T for a guitar OT. Virually all of the low cost OTs around are rated as guitar OTs (Wattage rating).
 
Last edited:
Thank you gentlemen!

It has FDK stamped on it, the core itself looks shiny, almost like a hard ferrite, so 10,000 may be a good guess.

Looks like probably I can pass 100W through it easily.
 

Attachments

  • fdk-core.jpg
    fdk-core.jpg
    72.6 KB · Views: 238
I am going to torture one using generator, resistor and oscilloscope...

100 Watt! Maybe at 70 KHz.

You need 4 or 5 of these (attached) 6 inch ferrite toroids stacked up to do a 100 Watt standard audio OT. Like $50 each unless you get samples.

I will check... The idea is, to use it's primaries in series, secondaries in parallel with 8 Ohm outputs, with Edcor's 10K P-P 100W transformer to equalize it's response on high end.
 
Last edited:
Smoking amp - you really need to take the Bsat data for ferrites with a grain of salt. Most of the Bsat numbers given are at 10 Oe or so MMF where the material is really in deep saturation. The real numbers are something like about 3500-4000 gauss at room temperature, about 1000 G less at 100C. That's for power material. The high perm materials have a lower saturation flux density, so actually, the 2500G number I gave may be a little bit on the wild side, with 2000G being a more sensible number. You also want to accomodate an asymmetric waveform without going into saturation.

The FDK designation comes from Fuji, so the cores are on the old side. It doesn't appear that Fuji is in the ferrite business any more, but I may have some old data squirreled away (no promises, though). Trying to do a 20-20k transformer with these cores would be insane, but a transformer working from 5k-up would be doable. The equations given by Smoking-amp are a good guide, especially the second, which is the classic "transformer" equation. Rearrange it so you can figure out the turns needed for a given voltage and flux swing. Once you get the volt-seconds issue dealt with, the throughput power will be determined by the copper losses, keeping in mind that the saturation flux density of the ferrite goes down with increasing temperature, so it is inadvisable to let the core get too hot (it will get heated up by the windings).

I've been thinking for quite some time of doing a dedicated tweeter amp with a ferrite output transformer, or augmenting a transformer with limited bandwidth with a small, HF transformer.
 
"The real numbers are something like about 3500-4000 gauss at room temperature, about 1000 G less at 100C. That's for power material. "

That's what I meant earlier by "useable" versus "spec" Bmax. In the final analysis, one has to work out the magnetizing current one can live with. Hopefully a tweeter OT is not going to get hot. We are in basic agreement here. In any case, on Wavebourn's 10K Mu material, he should find the Fuji datasheet to check the relevent curves for certain. I just took a quick look at Fair-Rite's #76 material, which is a similar 10K Mu matl.

"I've been thinking for quite some time of doing a dedicated tweeter amp with a ferrite output transformer, or augmenting a transformer with limited bandwidth with a small, HF transformer. "

I thought about this once also, RDH4 has a frequency splitter/combiner circuit for a dual freq. band OT. (page 888) Maybe for use with a dedicated speaker setup.

Trying to do the LF or the whole band (ie 20 Hz) looks impractical to me with ferrite. (5 six inch cores!) Huge cores. Winding capacitance issues at the top end. I've heard some negative comments on line before about some attempt at a ferrite OT, ie sounded bad, but I doubt if they really used sufficient ferrite.

No reason to expect much difference really from a steel toroid except maybe bandwidth and initial permeability. Might be interesting to do a hybrid version with a steel toroid and a ferrite toroid in the same windings. The ferrite should give better initial Mu at low power and the steel should handle the flux for high power. Sort of like pin-striping with permalloy.

I'm still turned off by the idea of 1000s of turns thru a toroid though (at least without a toroid winder machine). If I use ferrite it will be for a Berning type OT. Or a switched capacitor OT. There was a thread here a few years back where these got worked out conceptually. Once one gets the PC board for the SS stuff worked out, it should be easy to do any OT that way. Even does the HV B+ for free.
 
Last edited:
I'm still turned off by the idea of 1000s of turns thru a toroid though (at least without a toroid winder machine). If I use ferrite it will be for a Berning type OT. Or a switched capacitor OT. There was a thread here a few years back where these got worked out conceptually. Once one gets the PC board for the SS stuff worked out, it should be easy to do any OT that way. Even does the HV B+ for free.

I don't think it is practical: I am going to use toroids for HF only, in addition to EL-core OPTs. Can you imagine a 100WPC tube amp with 20 Hz - 200 KHz bandwidth? ;)
 
I tried a little experiment on a cut core OT here. I measured the leakage inductance with and without the core in place. With no core, it only increased about 20%. Then I tested it for frequency response without the core and it worked for 3 KHz and up fine.

So no point in bothering with hard to wind ferrite toroids. Just wind it on a cardboard tube.

A couple n*100 turns (single layer) on a long tube should work for the primary.
A long thin winding is optimal for low leakage L. If you want to get the low frequency down further (maybe 500 or 1000 Hz), try winding it on a large ferrite rod. I've seen 3/4 inch diameter by 8 inch long rods. You will only get around 60 Mu effective that way due to the air flux return path (see the attached diagram from Fair-Rite products catalog. These curves assume there is some 1 or 2 inches of bare rod at each end like a typical rod antenna. Winding out to the ends will lower the effective Mu since there is less rod to radiate flux out on the end). But you can improve that considerably by stacking a bunch of toroids over the whole rod & winding assembly to provide the return path. This still won't get near the intrinsic Mu of the material due to the residual gapping, but maybe a few times better than the plain air return effective Mu.

Note: the typical ferrite rods around are low Mu (20 to 300) and low Max Flux (1/2 of power ferrite), since they are intended usually for HF radio antennas.
Although some called "Impeder Rods" for welding (arc noise suppression) may be higher Max Flux. So if the scheme works out, and a bunch more are planned, I would get some custom ferrite rods made using power ferrite material. (Fair-Rite Products obviously does such stuff). Or you could try stacking up a bunch of small toroids (ebay) to make the "rod" core, with bigger toroids stacked outside to form the outer return path.
 

Attachments

  • Rod_Mu.gif
    Rod_Mu.gif
    57 KB · Views: 100
Last edited:

bigwill

Member
2004-12-25 8:36 pm
UK
I don't think it is practical: I am going to use toroids for HF only, in addition to EL-core OPTs. Can you imagine a 100WPC tube amp with 20 Hz - 200 KHz bandwidth? ;)

I've also had the idea of combining transformers of different types to extend bandwidth, kinda like woofers and tweeters, but with transformers!

I can't say I know how I'd connect them together without "fighting" though (ie LF from the big transformer saturating the HF one)
 
Last edited:
"I can't say I know how I'd connect them together without "fighting" though (ie LF from the big transformer saturating the HF one) "

Well, there is this setup from page 888 of the RDH4 (attached). Also see page 185. But that takes DC inductors. A more practical setup would be to just do two amplifiers (a small one for the HF) on the same chassis using the same B+ power supply with a freq. splitter (RC) network at their inputs.
 

Attachments

  • pg888.gif
    pg888.gif
    7.8 KB · Views: 86

bigwill

Member
2004-12-25 8:36 pm
UK
Well, there is this setup from page 888 of the RDH4 (attached). Also see page 185. But that takes DC inductors. A more practical setup would be to just do two amplifiers (a small one for the HF) on the same chassis using the same B+ power supply with a freq. splitter (RC) network at their inputs.

In the context of achieving a single wide bandwidth output, would this work? I'd imagine you'd have to cap couple the HF secondary to the LF secondary, and then you'd be shunting it with the LF sec which probably has too low an impedance at HF, although the in the case of the RDH4 diagram that would be mitigated by the inductors in the primary of the LF transformer presenting less of a load

There must be some way to do this neatly...
 

bigwill

Member
2004-12-25 8:36 pm
UK
OR, how about having two PENTODE output stages, each one driven by its own phase splitter. A 1st order "crossover" driving each splitter at say, 5khz or so. Both OPTs DC connected to anodes, but the HF secondary is cap coupled to the LF secondary. The HF amp has to drive the LF secondary, but thanks to the pentode connection the impedance wouldn't be too low to shunt things much, and the stray capacitances would be pretty easy to drive I'd assume.

Then you should end up with a pretty flat gain / phase response up to quite high, wrap it in tons of NFB and end up with an SS amp :D
 
Umm, seems one either has to reverse the LC networks again at the secondary side to join them again or try what you suggested. The pentode side would lose the ability to do any local feedback (to keep high Z out), and plenty of global feedback would be needed to lower the final Zout.

Better to just leave them separate and consider it as the crossover network for a multi driver speaker.
 
If you have a cut core OT, just pull the core out to test. But make sure no low frequencies can get into the amp that way, or something is going to get HOT. I don't see any issue with a 3 KHz and up air core OT except minimizing leakage L by design (long thin windings). But for a mid range, gonna take some ferrite, or just use a small steel OT, since it can be grossly uprated in power at higher frequencies. (until the copper gets hot) Would be nice to find some other way around the LF OT though.
 
Last edited: