Pultec Inductors again...

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Your calcs were spot on, so no comment was needed.
I was amazed that you got that close, that approach has a lot of variance, as the winding pitch will change the mean length, etc.
Plus the wire DCR moves around a bit also.


Here are the levels at the different freqs:
B max = 10,000,000 Volts(-sine-)/(4.44* Hertz * Turns * 0.64 sq cm )

30 mH = 585 turns 12 kHz B Max = 0.50 V
50 mH = 756t 10 kHz B Max = 0.47 V
80 mH = 956t 8 kHz B Max = 0.46 V
100 mH = 1069t 5 khz B Max = 0.66 V
150 mH = 1308t 3 khz B max = 0.90 V

OK, all that is left is Q:

(lifted from Wikpedia, there are many definitions for Q, this one is handy for plotting Bell curves.

"Q factor

An ideal inductor will be lossless irrespective of the amount of current through the winding. However, typically inductors have winding resistance from the metal wire forming the coils. Since the winding resistance appears as a resistance in series with the inductor, it is often called the series resistance. The inductor's series resistance converts electrical current through the coils into heat, thus causing a loss of inductive quality. The quality factor (or Q) of an inductor is the ratio of its inductive reactance to its resistance at a given frequency, and is a measure of its efficiency. The higher the Q factor of the inductor, the closer it approaches the behavior of an ideal, lossless, inductor.

The Q factor of an inductor can be found through the following formula, where R is its internal electrical resistance and ωL is capacitive or inductive reactance at resonance:

Q = 2 pi f L/R

By using a ferromagnetic core, the inductance is greatly increased for the same amount of copper, multiplying up the Q. Cores however also introduce losses that increase with frequency. A grade of core material is chosen for best results for the frequency band. At VHF or higher frequencies an air core is likely to be used.

Inductors wound around a ferromagnetic core may saturate at high currents, causing a dramatic decrease in inductance (and Q). This phenomenon can be avoided by using a (physically larger) air core inductor. A well designed air core inductor may have a Q of several hundred.

An almost ideal inductor (Q approaching infinity) can be created by immersing a coil made from a superconducting alloy in liquid helium or liquid nitrogen. This supercools the wire, causing its winding resistance to disappear. Because a superconducting inductor is virtually lossless, it can store a large amount of electrical energy within the surrounding magnetic field (see superconducting magnetic energy storage)."

Hey Pat, what are the Freq Bands on the Dial Plate for that Pultec Inductor you measured?
Thanks!

OK, calculate Q at some basic frequencies:

XL = 6.28 f L
Wire ....Inductance... DCR.......f.............XL...............Q..............D (1/Q)

Red: -----32.2mH ---13.0 ---12,000 Hz = 2,427 XL/13 = Q 187 D 0.0053
Orange: --50.9mH ---16.7--- 10,000 Hz = 3,197 XL/16.7 = Q 191 D 0.0062
Yellow: ---84.1mH ---21.9--- 8,000 Hz = 4,225 XL/21.9 = Q 193 D 0.0052
Green: ---97.9mH--- 24.0--- 5,000 Hz = 3,074 XL/24 = Q 128 D 0.0078
Blue: -----147mH--- 30.9--- 3,000 Hz = 2,770 XL/30.9 = Q 90 D 0.011
 
core data and winding tables from magnetics, just realized that also the winding length is mentioned.
http://www.mag-inc.com/powder/catalog.asp

with pats new data:

the closest core is 20,3mm OD (page 16 on the core data pdf)
asuming 40% winding factor
so 20% is the average turn, which is 24,1 mm

30,9ohm resistance #34 wire = 30,9*0,88368 = 27,305712 meter

27,305712meter/0,0241meter = 1133 turns for 147mH
Al = 114.5

getting closer :green:
 
If those Q numbers seem funny, don't panic, here are the numbers you might get on a normal checker that checks L at 1000 Hertz.

If you take the Inverse Tangent of D, (1/Q). you get the phase angle.
The Degrees data below represents the angle at 1000 Hz.
Since the 2 pi f L formula is linear, simply divide the angle by ten for the angle at 10,000 Hertz.
So the 50.9 mH tap would have a angle of 3/10 = 0.3 degrees, not too bad.
A perfect inductor would have zero angle.
The angle is a function of DCR.
Capacitors have a very high Q, so they typically do not enter into angle calculations.

So you might get close to these numbers on a normal hand held meter.
Plus or minus 30 percent because thats the way it is in inductor land.
Some checkers, like the old Gen Rad, have a external input jack for checking Q and L at different frequencies.

5 Band Pultec Inductor @ 1000 Hz Estimated Values
Red 32.2 mH = Q 15.5 D = 0.065 3.75 Degrees
Orange 50.9mH = Q 19 D = 0.053 3 Degrees
Yellow 84.1mH = Q 24 D = 0.040 2.3 Degrees
Green 97.9 mH = Q 26.5 D = 0.038 2 Degrees
Blue 147 mH = Q 30 D = 0.033 1.9 Degrees
 
OK, Pat said the core was about 23 mm OD by 7mm ID and 8 mm high with the wire.

Since the shuttle has to have clearance when winding so it does not nick the insulation of the wire, the typical toroid is wound such that a 1 inch inside diameter core is filled so that it no has a 1/2 inch I.D.

So you use half the inside diameter for wire.
That means if you measure 7 mm as the I.D., then the ID without wire would be close to 14 mm.

Since the wire bunches up more towards the center, the inside build will be greater than the outside build.

To see this, we can look at a new invention, wire that is formed in such a way as to lay down on the toroid in a nice fashion.

Observe that the wire is formed with a square side , then a round side, then a square side, then round, I do not know how they make it, or how they get it to wind right, but it sure is a good idea:

6fm4pl.jpg


Anyway, since we do not have this wire yet, our coils will be bunched up with wire in the middle, increasing the diameter.

So we double the 7 mm ID measured by Pat to 14 mm for the ID.
Since 7 mm is the wire build on the inside, expect half that on the outside, so we subtract 3 mm from 23 and get 20 mm OD.

Subtract 2 mm from 8 and we get about 6 mm for height.

So we are looking for a core with the following properties:

20 mm OD 14 mm ID 6 mm Height.

A L = 87

After looking through some catalogs, we find that indeed, 20 by 12 by 6 or 7 is a standard size.

Standard Notation is A = OD B = ID and C = Height, so I will use that.
Here are some part numbers:

MGG Neosid: p/n 18-2006 A = 20.32 B = 12.7 C = 6.35 A L = 87 u = 160

Magnetics Inc: p/n 55848-A2 A = 21.1 B = 12.1 C = 7.11 A L = 87 u =
160

Arnold Engineering p/n 8-848032-2
(now part of Mag Metals, so the p/n is probably obsolete)

Ferroxcube p/n TX-20-1364 A = 20 B = 13.6 C = 6.4 A L = 87 u = 160

u is permeability.

If you mix a big batch of ferrite, and pour it into shapes, the more mass, the higher the A L, since the perm of the material is held constant.

So if two cores give the same inductance per turn, and they are the same shape, then the perm, and therefore material, must be the similar.

So for our A L value of 87, and our core dimensions, we find a constant perm of 160 from our various vendors that match our size.


OK, a little note on leakage capacitance.
You want to try and wind just one layer with a gap in order to minimize capacitance.
This is becaus the closer two metal plates are to each other, the more their capacitance.
The same goes for transformer wires.
The closer they are, the worse the problem is with parasitic C.
I have seen winders use fishing line in between turns to add a gap, whic is a real pain.

Anyway, here is a pic of what I am talkin about:

2nc2tco.jpg


Thats it for now.

Oh, one more:

j5i6u8.jpg


ok.
 
> wind just one layer with a gap in order to minimize capacitance.

Sure, for "free" (untuned kick-up) inductors and some HF tuned inductors where you need low-low C and high Q.

In this case.... aren't we going to hang a big capacitor across to get the resonance and impedance down into the audio band? Making-up numbers: 147mH at 3KHz needs 0.02uFd or more like 19,166pFd. If our tiny coil overlap adds 100pFd (surely less?), then we actually buy a 19,066pFd cap, which is still "0.02uFd" at Mouser. The error shifts "3KHz" to 2,992Hz, which is 3 cents in musical terms. Unless Q is way-high and the audio hovers right in this zone, this is not noticeable. Actual Q for most audio EQs is, what, 0.7 to 3? maybe 10 for a notch-filter which typically has to be hand-trimmed anyway?

I say wind it fast and easy. Overlap is no problem to real Audio Men.

That's a cute trick cramping the wire in the hole. They can't be selling it on spools that way, it must be a masher in the winder squeezing the wire just before it lays it in the hole.

May be moot to Audio Men since we often have to add resistance to the tuned circuit to bring Q down near 1 or 2 so it covers a good hunk of spectrum. Economy suggests using the thinnest wire so the design resistance is "free", but wire gauges do not work out neat and you usually have to add 20%-200% with an actual resistor.

We won't easily reduce core-size because that is set by Energy: the voltage applied and the frequency/ impedance it works at.
 
There are a family of optimum Q curves for these toroids.
They show the frequency vs the Q for the same core made out of different perm materials.

The peak Q happens at an optimum frequency for the inductor.

This happens when the copper loss equals the core loss.

Here are some plots for an Arnold core of different perms, but whose core size is about the same size as the Pultec core:
You can see that the higher the perm, the lower the optimum Q frequency.

25813cw.jpg


Copper loss is easy, thats just DCR.
No frequency dependency.

What about core loss?
It depends on frequency.

The higher the perm of the material, the lower the cutoff frequency of the core material.
For a 14 mil thick lam:

Supermalloy f-o 200 cycles per second
50-50 Alloy 1300 CPS
M6 Grain - 12,000 CPS.

Powder cores have an even lower perm, so they operate at a much higher frequency than a laminated core.
This is because toroids have about ten times less eddy current losses due to their construction.

In fact, in the Legg equation:

R/(u' f L) = hBmax + ef + c

the coeficients represent loss constants.

hBmax is hysterisis loss, ef is eddy current loss, and c is any other losses not related to h and e.

We can ignore c, and the e coef is so small for powder cores that we will ignore it also.

So we have hBmax as our main loss that shapes the Q curve.

Since Q is related to losses, we can manipulate the above equation to examine the relationship between Q with B max and f.
 
Here is a link for a toroid program that computes turns if you input enuff data.


Download link hidden at bottom of page as usual.
http://www.dl5swb.de/html/mini_ring_core_calculator.htm
 
drpat said:
It takes less than an hour to wind one of these things, and that's on a toroid core. Parts are only ten bucks, including the mag wire. The only thing that's remotely difficult about it is the math, but that's not even that hard. Plus you'll be learning something, and you can get your DCR spot on with the original Pultec inductor. Certainly much closer than what's currently out there on the market. You'll also be able to use the original Pultec inductance values that I listed previously.

Or, like the above poster said, buy a Cinemag for 25 bucks. The big advantage of this route is the ability to use modern cap values instead of trying to find the odd values that Pultec actually used. You'll lose the advantages of the toroid core, and your DCR won't be original Pultec values, but it should still sound decent...

As of 5/25/09  the CML-150 (unshielded, no can) is 44.00.  If 25.00 was accurate for the previous price,  explanations for this nearly doubling in price would be very interesting to hear. . .  . . ;) 
 
As of 5/25/09  the CML-150 (unshielded, no can) is 44.00.  If 25.00 was accurate for the previous price,  explanations for this nearly doubling in price would be very interesting to hear. . .  .

There's the CML-150 and there's the CM-2251-x series. 

Maybe you're confusing the two?

The CML costs 40+

The CM-2251 costs 20+ (because they are not quite as complex, as Cinemag says)
 
owel said:
As of 5/25/09  the CML-150 (unshielded, no can) is 44.00.  If 25.00 was accurate for the previous price,  explanations for this nearly doubling in price would be very interesting to hear. . .  .

There's the CML-150 and there's the CM-2251-x series.   

Maybe you're confusing the two?

The CML costs 40+

The CM-2251 costs 20+ (because they are not quite as complex, as Cinemag says)

Must be.  The quote from the old post was probably referring to the 2251-x which I wasn't aware of.  I was originally looking for the CML-150 with the can which they apparently no longer make.  Thanks for pointing that out.
 
I have 3 real Pultec's(4 if you count the hp/lp filter) and built one clone. I used the Sowter inductors and seriously you can't tell the clone from the real thing.  The difference is no more than the slight difference from unit to unit in the real pultecs.
 
I disagree.

The Sowter & Cinemag (IE CORE) inductors MAY sound similar when it comes to FREQUENCY...
BUT they do not "Saturate" the same as the Originals (Toroidal Core).

IE Cores are much "Cleaner" sounding to me (& many others)

I've given detailed explanations of what I've heard if anyone care to read it... just search.
I think in this thread.

Obviously... everyone has different opinions & EARs... so, believe what you want OR use your own ears. ;)
 
niI have a nice LCR  Meter FROM 100 hz TO 25khz. Cost is 250 USD + Shippng. You may check specs at [edited] ... Bhagwat
 
-Such items need to be listed in the WHITE market, after first meeting the requirements of posting there.

You would of course be most welcome to post there, and your item looks very interesting, but A couple of forum users have indicated that they're uncomfortable with any users getting 'free advertising' here while others do not.

With that in mind, I've edited your post.

Thanks for your understanding,

Keith
 
Has anyone wound the 3 inductors for the MEQ-5?  Was getting ready to wind some of these for the unit I'm building and wondered how well the AL values used for the EQP1-a  would work on the MEQ-5 inductors.

Anybody ever used these?

http://www.bisoncores.com/products_mpp_specpg3.htm
 
Wire Inductance DCR
Blue: 147mH 30.9
Green: 97.9mH 24
Yellow: 84.1mH 21.9
Orange: 50.9mH 16.7
Red: 32.2mH 13

Wound on a toroid core
 
It takes less than an hour to wind one of these things, and that's on a toroid
core. Parts are only ten bucks, including the mag wire. The only thing that's
remotely difficult about it is the math, but that's not even that hard.
You can get your DCR spot on with the original
Pultec inductor. Certainly much closer than what's currently out there on the
market. You'll also be able to use the original Pultec inductance values.
Or, buy a Cinemag for 25 bucks. The big advantage
of this route is the ability to use modern cap values instead of trying to find
the odd values that Pultec actually used. You'll lose the advantages of the
toroid core, and your DCR won't be original Pultec values, but it should still
sound decent
 
The original inductor has an AL value of 875μH, and is about the size of a
quarter, so ferrite is probably not the original core material.
 
They used both 5 tap and 4 tap inductors in the original EQP-1A, depending on
when it was manufactured. The earlier models had a 4 tapper in it.
The inductor is located in the same filter can as the HF filter caps. The can is
filled with actual bees-wax.
The original inductor is about the same dimensions as 5 US quarters stacked
up.The wire does not appear to be wrapped in any sort of orderly fashion. it's easy
to see criss-crosses and loose wraps all over the place.
The original inductor was NOT potted, however it's circumference was wrapped
in a yellow cloth tape. The tape appears to be holding the wire leads into place.
The original inductor is mummified in a scrap of a Pultec price sheet, then
wrapped in a thin sheet of shiny flexible aluminum. The beeswax was outside
of this aluminum shroud, although a small amount did seep in through the
cracks, but ended up mostly on the yellow tape.

The wire is copper

The core was most likely "molly perma powder
core"
 
Back
Top