Ferrite cores for audio transformers?

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
CJ said:
...lot of work on those 2 spreadsheets...

No credit for cross reference... I only did 1 sheet, the cross reference is something that came from a manufacturer I modified to keep track, there are several available, this is the best I found.

CJ said:
...might want to add a Q category...compare the orig Pultec graphs...

Really?  How to do that? The Q of the various Pultec sections that use inductors is influenced by the surrounding components and how each inductor was wound.  For instance the high boost section (which is the only part of the EQP1 that uses an inductor) is influenced by a couple of pots and a cap and possibly the interstage transformer/inductance on the way out to the makeup gain. So really the useful comparison needs to be made "in situ" with that componentry replicated.

I can wind with any wire that will fit (as you know) and so I can get any DCR I want (subject to size constraints on low end).

At this point I am comparing core materials to look at saturation and BH loop to see what material AND core shape will replicate the distortion characteristics ( or lack of distortion characteristics ) of the Pultec inductor materials.  I am looking for a NON toroid solution here, something I can wind on bobbins for as many channels as I want (It takes a LONG time time wind these toroids, and the taps break off because of my thumbs...  It is not really DIY friendly and the auto winding machines for toroid use.. well I just can justify this expense to satisfy my curiosity about an analog process outmoded years ago)

From the "Pultec Inductors Again" thread http://www.groupdiy.com/index.php?topic=25482.msg516296#msg516296 we have a pretty good description of the core used to make an EQP1...  You yourself posted dimensions
20 mm OD  14 mm ID  6 mm Height.
A L = 87

The cores I found on Nebraska Surplus are MPP cores from Magnetics (for less than a buck apiece!) and are (MPP which someone said is what Pultec used ... see attached datasheet, I think they might originally have been Arnold).  They are 21 x 12 x 7 coated, and my measured AL varies from ... You later posted:

Now we can get turns:

30 mH = 585 turns 
50 mH = 756
80 mH = 956
100 mH = 1069
150 mH = 1308

So thats the way it was back then, there best audio band powder core could do 87.6
"

The Pultec inductor clones I wound  come very close to these numbers... I will post a picture of them  (I added some more taps because my clones included a Gyraf at first, before I got all historical <grin> ).  The AL I measured ended up between 68 and 76 depending upon the actual core, but I have to say it is hard to lock in the actual turn counts when you are winding by hand (ADD folks... happy to wind forever, just don't ask us to count turns). But by my counts, adjusting AL as I wound I have:
34.9 mH = 716 turns 
50 mH = 857
90 mH = 1050
100 mH = 1212
150 mH = 1485


(Re: your simulated Q suggestion I saw that you posted optimal Q frequencies for the various core permeabilities, but I don't really know how to measure or simulate the optimums.)

Here is the process I was thinking of following, let me know if you think there is a better faster (less winding) sort of way:

1) Build a (stereo) Pultec clone with modules so I can swap in and out various components.  I did this in a Vector rack (I will post pictures).  Solid state in and out boards of my own design to allow me to put whatever in and out trafo I want and not have the external equipment effect measurements as I change trafo's etc.  Stereo lets me AB comparisons, and makes listening better.

2) Wound some inductors and tried them.

3) Using guidance from the above referenced thread, wind some close clone of the Pultec inductor.  My best attept at this is using the above referenced core ( I can post measurements ) hand wound on a AL 70 ish MPP core using 34 gauge wire.

4) Measure that inductor for saturation characteristics ( would love to measure the BH loop but don't know how, only thing I can I think it is measured indirectly with THD ) which I did using the setup you suggested (I think I did that right).  My setup is:  Big driver amp - outputs to 51 ohm resistor - outputs to inductor tap - with the starting tap grounded.  I put an oscilloscope across the inductor, and ran sine waves into it looking for the curve to deform (visually on the XT and also with XY display  (BTW is that XY showing me BH loop?)).  Is that test setup correct?

5) Make a spread sheet and hope CJ or some other brain surgeon comments (which he did,  thanks very much!) to help me along in comparing.

6) Find core shapes and materials that COPY the characteristics of the Pultec toroid clone, but can be wound on Bobbins.  I want to copy the saturation levels, and the BH loop, because both of these seem like the can effect the sound beyond the Q shape (which can be manipulated by winding differently).  This is where I am now... still winding and testing....

BTW: I find I can not deform the curve above 1k Hz on pretty much any of these inductors across the 150uH tap below 11 volts RMS, levels I am pretty sure are never seen in the padded down world inside a Pultec.  So I guess I am going to have to compare the inductors based upon the smaller taps, and below 1K, even though they are used above 1K (this makes me uncomfortable because the inductors are used at higher frequencies (but while I see how to do it, I cannot see the "Audio" relevance of the test method you described using a Variac, the levels are just too high and the frequency too low to be relevant inside a Pultec, I am already testing at 10X the max RMS levels seen in there, 100X seems overkill)


Early indications are that the practice over at Carnhill of using gapped cores (in the case of their RM7 cores... AL 400 ) is not far from what I come to using this test method.  My test cores in the spreadsheet using N48 RM7's gapped to AL250 can take 2 volts plus at 20Hz well beyond what the Pultec toroid took.

Also... I get really weird shapes in the wave from  some inductors when the saturate.  Some saturate differently.  They do effect the driving amp (which is why I put the 51 ohm resistor in to allow the core to distort before the amp so I could tell when it was the core and when it was the amp clipping.  So far it is always the core except above 10V rms)




 

Attachments

  • C055206L6.pdf
    86.8 KB · Views: 20

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
Promised Pics

Test Interface Driver Amp - Inputs on the back, Unbalanced out on the front to 24DBu, meter is within 0.2dBu (the yellow, green, yellow light trio lets you zero in on 0dBu, +4dBu, +16dBu... then just green for 18,20,24 ( about 1 db of error progressively from 20-20K, reads high by 1 db at 20Hz)
photo_zpsc95d8ef0.jpg


Stereo Pultec Modular test platform
(the mini bar knobs "feel" like they sound good <grin>)
photo_zps4e030eed.jpg


And the "robot candy" that fits in there
photo_zps1a5ccc88.jpg
 

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
More candy for robots:  Pultec Clone Toroids.  I have built 3, here are the specs and pics of the last 2, using the Magnetics C055206L6 cores.  I can tell from my measurements that my lack of attention to counting and not the core is at fault for my variations.  Which are up to 15% off at a couple of taps.

ID: Black Heat Shrink  Toroid 
Type: 8 Tap Pultec/G-Pultec MPP

Taps
Start Purple/strp
uH      DCR    Color
  21.5  10.8  Pink
  34.6  14  Red 
  49.8  17.1  Orange
  64.1  19.6  White 
  89.1  23.5  Lt Green 
  98.9  24.8  Yellow 
152.2  32.1  Green 
176.8  35.3  Blue       

ID: Blue Heat Shrink  Toroid
Type: 8 Tap Pultec/G-Pultec MPP

Taps:Start Purple 
uH    DCR    Color
22.1  11  Pink 
35.1  13.9  Red 
55.9  17.9  Orange 
74.2  20.8  White 
91.6  23.4  Lt Green 
101.9  25  Yellow 
164.2  33  Green 
191.5  36.1  Blue

Note:  In testing these for saturation levels, I find that the curve that occurs upon saturation is VERY different from that of the curves of certain ferrite inductor cores.  I will show pic's so folks can make a determination as to why that is.  It is as if saturation of the flux capacity of the core collapses faster or reverses when saturated and is not symetrical.  I don't really understand the curve... Pics to follow

But these are pretty little inductors aren't they!
 

Attachments

  • robot candy.JPG
    robot candy.JPG
    714 KB · Views: 42

shabtek

Well-known member
Joined
Jun 8, 2005
Messages
1,973
Location
midwest usa
NICE r&d. what is your trick for winding on the donut?
I am guessing you just will the wire on their.
 

CJ

Well-known member
Joined
Jun 3, 2004
Messages
14,147
Location
California
i bet he uses a small stick, like a popsicle stick with notches on both ens to hold the wire,


there are 2 Q's we can talk about, one is 2 pi f L  / R for the inductor, and  one for the EQ graphs,

"A high Q resonant circuit has a narrow bandwidth as compared to a low Q
Bandwidth, Δf is measured between the 70.7% amplitude points of series resonant circuit.

Bandwidth is measured between the 0.707 current amplitude points. The 0.707 current points correspond to the half power points since P = I^2R, (0.707)^2 = (0.5)."

the formula is  2 * Δf / f-rez = 1 / Q

from  http://www.allaboutcircuits.com/vol_2/chpt_6/6.html

 

Attachments

  • bandwidthQ.png
    bandwidthQ.png
    5.7 KB · Views: 13

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
CJ, I use a big stick..... (Just kidding).

Thanks for the link, I now understand that the inductor has it's own Q as well, I will read up and I will endeavor to measure it.  I thought it was just inductance and resistance that defined the Q but I guess there is more to it.

As far as how I wind, see picture, I cut a slice off the edge of a plastic "take out container", it makes a U shape channel that is strong, flexible, smooth and can be made to hold wire.  Then I estimate the number of turns I will need, and wind that much onto a spool using my turns counter, and then transfer it to the "Popsicle stick" (otherwise it is hard to estimate the length needed, I could just measure length but I lose count).  I tin the last bit and leave it accessible on the inside of the stick channel, so I can check inductance or resistance while winding.  Note... the inductance is tempurature sensitive, and warm hands effect it when winding.

The plastic stick has the advantage that it is smooth and gentle on the windings when it gets tight, it also deforms and gets smaller when the hole gets really small, and you can trim it thinner.

It is time consuming with low permeability cores, and takes forever.  That's why I am looking for the best duplicate using a bobbin wound system.


 

Attachments

  • popcicle junk.jp.jpg
    popcicle junk.jp.jpg
    754 KB · Views: 85

shabtek

Well-known member
Joined
Jun 8, 2005
Messages
1,973
Location
midwest usa
bruce0 said:
It is time consuming with low permeability cores, and takes forever. 
oh i thought you had a way around this.
Yours is a few steps above the popsicle stick, nice innovation.

edit (to avoid clutter):
did the # of views of this thread double in the last 12 hours?
holy moly, great thread
:Dsorry I'm bringing nothing to the table...
 

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
CJ:

I am a bit confused.

I read the allaboutcircuits link I now get the calculations... but won't they just come out the same as the Pultec one that was measured?

Because If I understand correctly the Q of the inductors I made can be calculated from their DCR (which I have measured) and reactance (which I can calculate for any frequency because I have measured Inductance).  So I have R and L and everything else is a constant or in the case of frequency as sweep.

But I built the inductor to mimic a measured pultec inductor, so wouldn't the calculations and curves just pretty much match?

Sorry if I am being thick, I am looking for ways to measure these devices so I can mimic them with a bobbin wound inductor.

So far the things I can think to measure other than DCR and L are:

Saturation point and BH loop
  I am measuring and reporting Saturation (based on your suggested method, which is working)
  I don't yet know how to measure how square the BH loop is.
Also
  While measuring saturation I am finding that when the curves start to distort, they do so very very differently for the different core shapes and materials.  I don't know why this is yet, any ideas?  I guess I should try to find which materials distort like an MPP Toroid.

Thanks for your suggestions, If there is some measurement I can make regarding Q let me know, or if you have any idea how to measure the BH loop because I think that is the most likely to effect the sound (creates THD) when the inductor is not saturated (which is almost always)... I would be happy to measure if I knew what the test circuit is to do it.  I can measure Voltage on an Oscilloscope, I don't have a current probe.

bb


CJ said:
there are 2 Q's we can talk about, one is 2 pi f L  / R for the inductor, and  one for the EQ graphs,

....
the formula is  2 * Δf / f-rez = 1 / Q

from  http://www.allaboutcircuits.com/vol_2/chpt_6/6.html


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

CJ

Well-known member
Joined
Jun 3, 2004
Messages
14,147
Location
California
you can make a BH tester with a 100 k resistor and a .1 cap,

use your scope in X-Y mode,

here is some BH stuff,

http://www.ndt-ed.org/EducationResources/CommunityCollege/MagParticle/Physics/HysteresisLoop.htm

do you have the circuit for the BH loop?

 

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
I don't have a circuit no. 

When I looked online (just now again) all of the circuits require the use of transformers.  I could wind a transformer onto the core, but it doesn't really model the actual situation.  The magnetic flux will be less than I have now ( I can't really add many turns to the existing cores, which as you can see are quite full ).  I could wind an MPP core as a transformer, and then test that, but the result would be pretty generic, I would probably get something like I could get off a material data sheet.  I was wondering how much of the loop is being traversed at normal signal levels in a pultec eq inductor for example.

Do you know of a circuit to test the BH loop of an inductor core without turning it into a transformer?



 

CJ

Well-known member
Joined
Jun 3, 2004
Messages
14,147
Location
California
you can make a prediction of saturation levels by using BH curves from core catalogs.

http://www.fair-rite.com/cgibin/catalog.pgm#select:part5

get the gauss off the B-H curve,

most Ferrite curves level off real nice so the number is easy to read,

then you use the formula to examine levels vs turns and frequency and voltage,

area is fixed so it can divide out which simplifies the equation down to 3 variables:

V,f and N,

pick the freq for the band you are working on, now you simply have 2 variables:

Volts and Turns,

use a spreadsheet and you can build charts and stuff,

B-max = 25,000,000 * Volts-rms / Frequency * Area-(cm^2) * Turns

this test setup is just a theory, forgot how i did it on the bench, see if it works, you might have to play around with the 100 ohm value to get off the noise floor,

use a square wave if you have it,

and resize your pics from 12,000000 x 8,00000000 down to 800 x 600 so we can look at them,  :D
 

Attachments

  • inductorBH.png
    inductorBH.png
    17.8 KB · Views: 44

CJ

Well-known member
Joined
Jun 3, 2004
Messages
14,147
Location
California
here is the easy saturation checker thing, adjust R value depending on input current,
make it low enuff to get a non fuzzy trace, that way you minimize it's effect on the measurement, notice scope is in normal mode, you can also use sine wave for this,

waveform may deviate from sat definition depending on core material, so V-Sat is an inexact science to say the least,

 

Attachments

  • v-sat.jpg
    v-sat.jpg
    85 KB · Views: 48

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
CJ... You are amazing... Wonderfully helpful!  Thanks.  I am going to try these.  Actually the second one I have been doing already to get my results so far... but I will do the first schematic on all the inductors I have tested so far.  i can make square and sine and the scope does xy... so I should have no problem. It will take a couple of days though ... leaving town... will post results.

And I will post them at 800 by 600, sorry about that... 
Is the problem on the single attached pictures or the several in one post that I reference at PhotoBucket?

b

(Apologies... I didn't realize that anyone still used a browser that didn't resize with ease,... I think I have done this on every post I ever did!.... I use Chrome now mostly  )

b
 

CJ

Well-known member
Joined
Jun 3, 2004
Messages
14,147
Location
California
no problem on the pics, everybody does the same thing,

doing some calcs, looks like you have to use a gapped lamination if you use regular lams,

a 094 EI lam with 1300 turns will give you 148 mH if it has a perm of around 250,

normal 29gaM6 has a perm of around 8,000 to 10,000, so you need a gap to reduce it otherwise you get about 5 henries for 1300 T on an ungapped sq stack,

you can stabilize the gap with a good core bracket, and ferrites have a temp coef also,

once you get the EI lam gap solid, the perm becomes very stable,

094 EI is very small, 1/4 inch by 3/8, so you need smaller wire which is going to throw off the Q,

so it looks like using the same dimensions for a toroid core would unfortunately be the best way to duplicate a Pultec inductor, unless you used a pot core like the Trident eq,

so if you want to duplicate the EQP 1a coil you should probably stick to a ferrite material in a pot core config if you want to bypass the winding headache,


FYI: the Lang version of the Pultec uses a Torwico 5663 core, but Torwico has been gone for a while, however you might be able to dig up some info on the old cores,
 

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
Magnetics Inc says that gapped ferrite cores have a cliff like saturation point instead of the MPP soft saturation.  So I figure either I need to stay far far away from that point, or find a lower permeability material (none of the modern ferrites) or laminations of metal as you mention.

I will post some scope traces, but here is a chart magnetics uses to illustrate this soft saturation
 

Attachments

  • soft saturation.jpg
    soft saturation.jpg
    59.6 KB · Views: 26

johnR2

New member
Joined
Dec 20, 2011
Messages
1
bruce0 said:
Because If I understand correctly the Q of the inductors I made can be calculated from their DCR (which I have measured) and reactance (which I can calculate for any frequency because I have measured Inductance).  So I have R and L and everything else is a constant or in the case of frequency as sweep.
You also need the winding capacitance to calculate Q of an inductor. There's no resonance with just L and R. It's probably easier just to measure the impedance vs. frequency and see where the -3dB points are, as winding capacitance is difficult to measure.

Edit: BTW knowing the Q of an inductor coil on its own is mainly only useful for RF circuits where there might not be an external capacitor. In audio you usually just need to know the Q of the complete resonant circuit.
 

PRR

Well-known member
Joined
Jan 30, 2010
Messages
11,144
Location
Maine USA
> You also need the winding capacitance to calculate Q of an inductor

That's at self-resonance frequency.

In audio tone/EQ systems, we "never" work a coil at self-resonance. We always have an R or a C.

The Q that CJ is talking about is "useful Q", a clue to how it could work *in an audio filter*.

Say I have a 1 Henry with 500 ohms of resistance.

Inductive reactance of ideal 1H at various frequencies:

20Hz = 126r
50Hz = 314r
100Hz = 628r
200Hz = 1260r
500Hz = 3140r
1KHz = 6280r

For an EQ we add a cap and resistor.

The silent fact is that most film-caps have Q much-much higher than most coils. So for a first-whack we can simply assume the cap will not degrade the Q.

We can add resistance which will lower the Q. We can't buy NEGative resistors to raise the Q. (We can do that, but it becomes easier to lose the coil and do an R-C+amp filter.)

Maximum possible Q of my 1H 500r coil:

20Hz = 0.25
50Hz = 0.63
100Hz = 1.25
200Hz = 2.5
500Hz = 6
1KHz = 12

What Q do we need? 0.5 is very broadband, and can often be done as well (cheaper) with R-C networks. Q=5 is quite sharp, makes a very minor dent in music. Large Q might be used to dip a single-frequency whine, notably AM adjacent carrier or FM stereo carrier. For most musical sweetening, Q is in the range 0.7 to 2.0

So this 1H 500r coil is useful down to 100 maybe 50Hz.

Note that my example coil is probably gapped (reactance/frequency is flat). If laminated iron plays a large part, the inductance will drop above a few hundred Hz (eddy currents throw the flux out of the lams) and Q will not rise proportional to frequency.
 

CJ

Well-known member
Joined
Jun 3, 2004
Messages
14,147
Location
California
every core has some type of natural gap due to imperfect surfaces and other factors,

in a Pultec, the torrid gap is caused by the iron particles being spaced out in the bonding medium,

<If laminated iron plays a large part, the inductance will drop above a few hundred Hz (eddy currents throw the flux out of the lams) and Q will not rise proportional to frequency...>

so there must be a frequency where maximum Q will occur, unique to every core and every stacking method, and gap,

or another way to look at it, from the formula:

X-L= 2 pi f L

so frequency and inductance will be fighting each other, as one goes up, the other comes down, now if inductance drops off at a low and a high frequency, then Q-max will occur when L is rising faster than f is dropping, until eddy currents start to slow the rate of inductance rise,this happens somewhere in between the low cutoff point and the high cutoff point of the core,

how do we compute this frequency?

f -  optimum = (DCR * R-e) / (2 pi f L),   

R-e is called shunt eddy current resistance, which is in parallel with L, both of which are in series with the DCR. this is called a series-parallel LR network, which is what we use for our model in this situation.

now the maximum Q of the inductor will occur when copper and core loss are equal,

and what will be the value of our optimum Q?

Q-opt = 1/2 * (R-e / DCR) ^ 1/2

now if you hit the inductor too hard, the flux will go up which means core loss will go up, which means that the optimum frequency will shift due to the loss factor changing,
so we try and run the inductor at low flux levels to keep the resonant frequency of the bell curve the same no mater what the signal level, so you really do not need to worry about the rapid saturation core with the flatline B-H if you design the core for low gauss operation, it will never get close to the saturation voltage anyway,

by having an inductor with a gap, you minimize this effect by putting some of this energy into the air gap.

if a high Q filter is desired, you find the frequency where Q is max for your inductor, then you match the capacitor to the impedance of the inductor you have picked at this maximum Q,
 

Attachments

  • Qcurve.jpg
    Qcurve.jpg
    159.4 KB · Views: 12

CJ

Well-known member
Joined
Jun 3, 2004
Messages
14,147
Location
California
since were doin the toroid thing,
need new camera,  but you can read this with some bifocals or maybe a mag glass,

 

Attachments

  • toroid-Grossner.jpg
    toroid-Grossner.jpg
    911 KB · Views: 50

Latest posts

Top