Ferrite cores for audio transformers?

[CJ]

The Al number will depend on the metarial as well as the physical dimensions of the core, ie; cross sectional area etc.

The lower the Al, the higher the freq response, but more turns needed so self canceling benifits.

The lower the Al, the higher the loss factor.

The higher the Al, the higher the initial permeability.

The higher the Al, the worse the temp coef gets.

The higher the Al, the lower the coercive force becomes. (Oersteds)

 

[Guest]

[quote author="CJ"]The Al number will depend on the metarial as well as the physical dimensions of the core, ie; cross sectional area etc.

The lower the Al, the higher the freq response, but more turns needed so self canceling benifits.

The lower the Al, the higher the loss factor.

The higher the Al, the higher the initial permeability.

The higher the Al, the worse the temp coef gets.

The higher the Al, the lower the coercive force becomes. (Oersteds)[/quote]
... Sometimes Al depends on the air-gap, which is milled onto core.
There are cores with the some material with different Al inside two orders.
And some cores are milled on the one-side only and
you can complete from two low-Al cores one with big Al without gap.

xvlk

 

[CJ]

Yes above info was for torroids.

 

[johnbessa]

From my reading, ferrite cores have less permeability, but are better in every other way.  Si Fe is only good at audio frequencies (whatever that means).  If permeability is as good as Si Fe in new ferrites then it should fly.

 

[gemini86]

Thread necromancy?

 

[gyraf]

..But still very interesting, as no real conclusion was drawn back then.

I'll see if I can find and re-upload the mentioned B/H curve I made of the T38 core material.

Jakob E.

 

[bezen4uk]

Hi Jacob,
Did you tried nanocrystalline Fe-based cores?
That is the killer material for trafos: Al up to 60uH, Bsat=1.2T.
Example:
http://www.magnetec.us/en/nanopermr-products/technical-data-nanopermr/

 

[CJ]

fun to experiment with but for production you need a toroid winder,

unless you are locked up and have a lot of free time to hand wind those things,

ever been to a prison factoery outlet?

many cool inmate items for cheap, rattlesnake belt buckle (actually killed by the inmate during weeding a field) for 12 dollars, real leather brading and solid steal backing need i say more?

they get money for smokes, you get cool stuff,

i have a san quentin mic capsule that rocks,

 

[PRR]

> solid steal backing

In other stores that would be solid steel. But I guess it is appropriate.

We have a huge prison store down coast, also smaller prisons with very elaborate work-shops. Underfunded but overstaffed. One does woodbending on contract to a snowshoe maker.

 

[bruce0]

Thread from the dead - but still interesting to me.

I have been experimenting with Ferrite cores for audio, for inductors in EQ circuits.  I thought I would do a bit of a "Memory Dump" of what I have learned.

I find the RM7 and RM8 cores the easiest to work with ( They both handle enough turns to get inductance values for audio using 32, 34 or 36 AWG magnet wire) and  get parts for (ChrioN uses RM8, Carnhill use RM7, and you can get some parts at Mouser). 

Oddly, while there is lots of ferrite around to play with, the bobbins seem to be the hardest to buy in small volume.  I can't find pot core bobbins for love or money, and they cost a fortune for RM7/RM8 at mouser, but you can get them.  (You don't need he round insulating washer for our purposes, the square one is only needed if you want to keep the ferrite from touching the PCB.

I have also been playing with and ordering some Pot Core's to mess around with.  These are named by size not by the more standards based RM7 and RM8 approach, and they seem to be a little be different from each other (number of openings etc) but in general they use names like 26/16 and 22/13 which are named for diameter/height of a stack of two half cores.  Carnill inductors also come in 26/16 and 22/13 as well as RM8 which is newer for them.

Here is what I have learned so far.

If you can count turns, it is very easy to precisely get a multi-tap inductor that measures out to within 2% or so (as long as you are willing to make a bobbin with 100 turns on it, and assemble it to each new core to measure a precise "AL" value for each core).  The cores drift with tempurature and how hard they are clamped together by another 5% pretty easily.

While the LCR meters may not be valid measurements, they are repeatable, and my cheap LCR meter (Victor) has proven pretty accurate to the frequency curves I later measure on the EQ in RightMark.

Some of the cores have a threaded insert glued into a center hole.  These are NOT intended to attach the cores together (there are clips for that) The threaded holes take little ferrite slugs for RM7 and RM8 mouser sells these slugs (which are tiny complex little beasts) for about $2 apiece, but using one of these you can trim the AL value after you wind and really "zero in" the inductance, if you care.  You can easily tune them to within 1%, again temperature drift aside.

Each ferrite is made of a "Material" which has an AL value.  Each manufacturer has their own "Materials" which they name or number or code.  They all seem to be proprietary but well spec'd. But the ferrite cores are "ground" with or without gaps (grind shortens the center post creating an air gap ), the air gap lowers the AL value, bigger gap, lower value.  And I assume the lower AL means harder to saturate?  But of course you need more turns to get the same inductance with a lower AL? 

Anyway for each core pair you will get an AL value and a material name (like P48/400)

The pot cores come in "pairs" of halves, sometimes gapped equally, sometimes the gap is only on one half for "mix and match" (that way you could get a big gap using 2 ground halves, a small gap with one ground half, and no gap with two unground halves).

Some of the cores have no hole in the middle, these are said to be for transformers, and I assume it is because they are harder to saturate.

None of the manufacturers ever mentions a frequency below 100K Hz.  (So they don't see analog audio geeks as a big market).  But they do have lots of specs and descriptions of each material.

I have played with cores with AL values from 8000 (ungapped) to 250 (gapped), and I can precisely measure the AL value but I don't know how to measure when the inductor is saturating (CJ, can you shed any light on this?).

I don't know what material is best for audio inductors and transformers but I have been trying to suss that out...If you pick the one that the Manufacturers suggest for resonant circuits below 100K then I have been making a list of those, as I go along, but there are lots of manufacturers here are a couple.  Promising materials seem to be:

Fair-Rite - Material 77 (Also possibly 75 or 78)
EPCOS - Material P48 (Possibly N26, T38 and N30)


I think the key is to choose materials with Low Hysteresis to reduce distortion... (CJ or anyone want to chime in?).

The materials are often stamped on the core on at least one half, along with grind or AL information coded differently for each manufacturer.

Carnhill has pictures of two of thier inductors (the VT9046, and VT9047) which they "Advertise" to use P48/400 which is EPCOS material P48 which has an AL of 2300 (Permeability) without a grind, but which they grind to an AL of 400.  (The way they advertise that is that the picture has P48 400 stamped all over it).  I think that is very nice of them, and I wish they would show pictures of the other core stamps.  Anyone have a close up look at the other Carnhill cores?  Can you read the material stamp on it?

The VT9046/7 are RM7 based inductors, but it seems hard to get that material/grind in an RM7 core which they use, without special ordering it.  But AL of 250 and i think 650 are readily available in RM7, and you could put the adjuster screw in the AL250 stuff and bump it up somewhat.


It has a hole, I don't know if it is threaded or if they use the adjustor screw, I am just going by the pictures they publish.


So my question for the smart magnetic's guys and Mr Hacksaw...are:

What is a good material for Audio?  Do we want the lowest AL value that will let us get the Henries and DCR ( Q  ) we need so we saturate less (does it even work that way)?  Do we want lowest hysteresis?  How does coercive force recon into our target material?


I would love to know how to measure when an inductor begins to saturate... can anyone help there.  I didn't understand the description you gave above (CJ) maybe because the picture is now missing... (it is 8 years old or something now!).


For the time being, I have been trying out low frequency materials shooting for AL of 400, and P48 if I can find it for the inductors I am using.  I am also trying out MPP torroids because apparently that was used in Pultec EQ's but it is an archaic material.


I hope some of this helps folks out, it may all be obvious but it took a long time for me to figure it all out, and now with one point to point G-Pultec passive section under my belt, I thought I would leave some bread-crumbs.. and ask for a little help.

The inductor winding thing is kind of fun once you get the hang of it.  It is a pretty simple spreadsheet to figure out how many turns fit on the bobbins... and how many turns you need for a given AL value, and what DCR that will give you etc.  Then you just wind them, although I would be interested in some discussion of winding patterns, reducing capacitance, etc.

There is a picture here of 4 I have wound.  All are for the g-pultec with 22/69/169/269mH.  They have AL values from 7982 to 254, and they all work.  The two RM based ones are RM7 and RM8 (RM8 is bigger), and have a copper shield.  They are all on little cards with 1mm pins so I can pop them in and out of the passive section to try them out, hear them and measure them.

 

[CJ]

i like the 78 material from Fair Rite also, seems to have the best perm, torroids have no gap so they tend to be more stable,

you can use super glue to keep those gaps in lockdown mode,

saturation testing is easy if you have enough output from your generator, just measure the voltage across the inductor as you raise generator voltage, when you keep turning the voltage knob and nothing is happening across the coil in the way of voltage rise, you have reached saturation,  you can also plot input current vs voltage across the inductor, you will see a flat line at saturation,

they make those ferrite cores in huge lots, and every lot tends to be different, so you might see variance between cores that have the same part number,  ???

siemens made some good stuff and their catalog was huge, but they sold the farm a long time ago, however, ferrite cores get better and better each year, audio is not a huge market for the ferrite stuff so we have to try and use all of that switch mode stuff and make it work for audio,

good work!  how do they sound?

 

[bruce0]

My generator is wimpy... But I have a couple of those NV73 Neve 1073 clones, and a couple of tube preamps that can drive serious signal and I could drive the torroids with that.  I assume all this has to be done with AC, I can't just do it with DC.

Don't think the gaps make the inductance shift, it is the shifting of the alignment two components.  On one pot core I got less inductance as I tightened a screw holding them together, I think I was flexing the mating surfaces out of alignment (flexing ferrite). 

I find the units vary from core to core in the same lot on the same reel of components. ( It is hard to get a homogeneous mix of a slurry, I have some experience with that.) That's why I test each core with a test wound 100 turn bobbin.  That works very well. 

The different core configurations and materials do sound a little different, most are similar.  The RM7's P48/250 and the RM8 N87 ungapped and the little hand wound torroid all sound pretty good, the big pot core with the AL of 8000 sounds a little wierd (not bad really) I think I may be saturating it.

I only built the passive filter section, so I have to use another preamp as Make Up Gain, and it is very sensitive to the impedance of the preamp that does the makeup gain (you will say of "course? but this is new to me) and I am still working that out.

The bass really sounds nice, but what surprised me was how much I like the 6k - 8K ish presence boost and the air above that. It made me wonder if I was getting saturation on the low end ...

I have some RM8-P48/650 cores and I will wind a couple of those up to see how they are. It has a solid (no hole) core and should be harder to saturate. ( I like the inductors mounted on pins like op-amps, it makes it easier to compare).  Today I was lashing up the second passive EQ so I can deal with the two channel signal. 

I will try the saturation testing you suggest, thanks CJ.

b

 

[CJ]

if you are careful you can use 60 hz line pwr, that is what i had to do to get the Neve output to saturate, it is a big M6 core, a variac works nice here, just tape down your cables so you do not have the Hot floating around the bench,
if no variac you can make a voltage divider with some big resistors,

isolation transformers are helpful when dealing with line voltage,

or you could hook up a step down transformer to the 120 AC, or use a step down xfmr backwards and run it off the generator if it has enuff power to drive it,

 

[12afael]

thanks for the post Bruce a lot of good information.

a friend of mine found a board with a lot of nice size inductors, we remove them with a heatgun, the two parts of the ferrite core where glued together and I found that the best way to open it is the heat so super glue is not a problem .

the problem with these cores is that the AL calculated is very low. 0.37 and 0.5, I will take a look if the core is gaped, I was suspecting something like that.
have sense an AL so small??? I have been trying to find an error on my calculations but they seem to be ok... :-\

 

[bruce0]

That AL Seems too low how are you calculating?  The AL value of air is 1.

Put 100 turns on the bobbin and measure the inductance when the core is assembled make sure you clean off the glue from the core faces using a razor blade ( warming the core first helps with that) And meet the cores tightly, clamp them with something but ferrite is very brittle so not too hard.

The wire needs to be magnet wire of course. Or insulated wire otherwise you'll get shorted turns and that would explain that Low AL

With 100 turns if you get an AL value of .5 mH, Then the core AL is 50

The calculation I use is:

Turns=SQRT((( mH /1000)/( AL /1000000000)))

 

[bruce0]

if you are careful you can use 60 hz line pwr, that is what i had to do to get the Neve output to saturate, it is a big M6 core, a variac works nice here, just tape down your cables so you do not have the Hot floating around the bench,
if no variac you can make a voltage divider with some big resistors,

isolation transformers are helpful when dealing with line voltage,

or you could hook up a step down transformer to the 120 AC, or use a step down xfmr backwards and run it off the generator if it has enuff power to drive it,

Using +4dBu output from my test setup and running it through RightMark frequency response, all of the inductors except one traced the same curve.  The big 30mm x 19mm inductor with the AL of 8000 was wrapped with #30 wire (I was trying to push the Q,) and that one had a curve that rolled off much lower and wider.

I think I will use a preamp, rather than mess around with Mains.  I just need to pick a preamp that get push 18dbu or so... so the Neve or my Hamptone LCMP can both push a lot, even the 9K preamps can output quite a lot.  I figure I will put the signal generator (a little BK 9v job) into the DI input on one of those preamps, and crank till I see the saturation.

What are your thoughts on choosing inductor materials for low hysteresis?  Any hysteresis creates distortion I think, so as long as i can prevent saturation, the cleanest sound would be from ones with low hysteresis (a closed bh loop).  Is my thinking correct here?

 

[12afael]

I was using this formula

but I already found my error... L was already in mH...  :

here is some good info about design.
http://www.mag-inc.com/design/design-guides/Inductor-Design-with-Magnetics-Ferrite-Cores

correct me if I`m wrong but this kind of graphs should say something about saturation isn`t? if not I don`t get how the AL change with the current...

 

[bruce0]

I found several cross references to Core materials (mostly Ferrite Materials).  Here is a link to a shared folder of some cross references I have collected from around, some are older and may not contain the many new materials.

One file in particular in this folder named "GROUPDIY EDIT cross reference.gsheet" is editable by anyone.

Please feel free to add any cross referenced materials that you want (it is a google sheet, editable with their free spreadsheet app). It is backed up every day if it gets screwed up I can fix it, so don't worry, there are notes inside the spreadsheet about how to mark your changes

https://drive.google.com/folderview?id=0B7zKjFMCjXR8RlVqbmxyTjEtX00&usp=sharing


On other matters I have built a line driver for my signal generator, and so I think I will be able to test some saturation characteristics.  More on that later.

Two other things that have become apparent.

If we want inductors with low THD we want to use materials with low hysteresis (referred to sometimes as "hysteresis material constant").

Companies seem to use "broadband" materials for Audio uses (as opposed to Power or Hi-Q materials) based upon the small number of sample cases I have seen.

 

[bruce0]

i like the 78 material from Fair Rite also, seems to have the best perm, torroids have no gap so they tend to be more stable,

you can use super glue to keep those gaps in lockdown mode,

saturation testing is easy if you have enough output from your generator, just measure the voltage across the inductor as you raise generator voltage, when you keep turning the voltage knob and nothing is happening across the coil in the way of voltage rise, you have reached saturation,  you can also plot input current vs voltage across the inductor, you will see a flat line at saturation,

...

good work!  how do they sound?

I built a line driver for my Signal Generator (a single TL074 doing balanced receiver, gain and DC servo work, and a push pull stage with BD139/140 biased with an LED... it will drive pretty much anything the main problem is that the TL074 limits me to 18V rails, so maybe i will update it with a high voltage opamp.  It has a little LED voltmeter that will let me quickly put it at 0, +4, +16, +18, +20, +22, +24dBu, and the same positions padded down -40dB.  I have to run the TL074 beyond it's supply voltage rating to get to +24dBu)


I did some tests of various inductors I made.  I think I need to figure out how to calculate when saturation will occur because these are dramatically different in terms of when they saturation.  In particular the low frequency signals saturate the low value inductor taps very easily.

I wound several possible inductors for a point to point Pultec I built (they are on plug in modules so I can swap them out).    I wound some on Toroids, and in particular some of them are on Magnetics C055206L6 MPP cores, which are very similar to the measurements and permeability of the Pultec inductor someone on this booard measured.

I am trying to find out what core materials mimic the saturation characteristics of that core with something that can be wound on a bobbin.  Here are the test results so far. 

https://docs.google.com/spreadsheet/ccc?key=0ArzKjFMCjXR8dE9CMzVHZGpEVFE1eGpheUZONlBIaEE&usp=sharing

From that spreadsheet it is beginning to look to me like the core needs to be gapped to lower the permeability so it saturates later.  And the core shape and size effect things too, so for instance and RM8 will need to be in the AL300 range where the RM7 might be AL400 but I don't know what a 26/16 pot core might need... They have a much larger cross section for magnetic path... So I need to test it.

(It is worth noting that these inductors in the pultec are only used for high frequency and the signal is so low inside the pultec that any "near appropriate" inductor is NEVER going to saturate or get anywhere near saturation.  At least that's what it seems to me.)

I will update that link as I do more tests.  I have some 26 x 16 pot cores to try.

As far as how they sound.  I can hear the difference, so far I like the toroids better, but now that I know what is saturating, I will have to see if I can hear the difference between the ones that saturate around the same levels.

 

[CJ]

holy cow, that is a lot of work on those 2 spreadsheets!

you might want to add a Q category, which is simply Reactance /DCR, (2pifL/R)

then maybe generate some curves with a simulator which would give us a graphical picture of the various cores, then we can compare the orig Pultec graphs with the DIY stuff,