V72 Humbucker Choke

[CJ]

how do you figure out how much distributive capacitance there is in an inductor coil?

there are a ton of methods that you can research on the internet, most are related to RF coils and done by ham radio operators and PhD's, these are the guys you want to avoid unless you are in prison and have a lot of time on your hands,

so we reach for the bluebook, the bible, RDH4, and turn to page 225.

then we back that up with a resonant freq calculator like>

http://www.deephaven.co.uk/lc.html


so we use the data points off the "inductance" graph for rez peak freq and Henries to get another ballpark figure,

shown below is the RDH4 method, you start with the length of the winding, which in out case is about 7/8" inch for a 625 EI bobbin, then follow the red line to the mean turn length of a single winding, then you traverse horizontally over to the winding thickness, which is the build height of ll the turns (blue line), then you angle up to the right edge of the chart (green line), then you come back horizontally to the number of turns-21,000 (yellow line) then slant up to the right again to red the distributive capacitance in picofarads (black line).

so we read about 60, but remember we divide by 9 since we have aa 3 chamber bobbin, (3^2=9),

this is a crude method, and so is the graphical solution, but they both get you pretty close,

so C-d = about 7 pf.

now if you punch in various graph values into the LC calculator, you cn check yourself to make sure you are not way off the mark for C-d, punch in 7 pf and fiddle with the choke value til you hit 1800 hz,  in this case we get 1100 Henries, which is different from the chart, so maybe the RDH4 value can be tweaked to give a higher Henries reading, like 4.5 pf and 1800 Henries,

why does inductance go up?  core loss matches copper loss and the Q of the core is maxing out for that particular gap,

now we can adjust our graph to give  better approximation of actual inductance, see next post>

 

[CJ]

we have corrected the low end freq plot for DCR,

Z is the impedance of the choke,

this number is simply Volts-ac / Amps-ac through the choke at any given frequency,

Z is a combination of the resistance of the magnet wire in the choke (DCR) with the AC resistance (XL) of the choke

Z will change with frequency as XL (Inductive Reactance) changes with frequency,

however, DCR will not change with frequency (unless we go into RF skin effect which we can ignore at audio freqs)

so as frequency goes up, DCR will stay the same as XL goes up,

we can ignore the 8 pf distributive capacitance in the windings at these low freqs,

(calculate XC for 10>40 Hz and see that it is very large compared to Z and XL)

so DCR is a factor at low freqs where XL is low, but as you increase the frequency through the choke, XL becomes much larger than DCR  (XL=2 pi f L)

this means that Z nd XL will approach the same value, which means that DCR will not be a factor when determining inductance at higher frequencies,

DCR in the V72 choke is a factor up to about 40 Hz.

so we correct the freq graph from 10 to 40 Hz.

we can not simply subtract DCR from Z to get XL

since voltage and current are 90 degrees out of phase in an inductor, we have to use the formula for the sides of a 90 degree triangle to do our math,

you know, A^2 + B^2 = C^2

here is the freq plot of the V72 choke showing this correction,
notice that Z and XL go up but inductance remains mostly the same,
this is because as Z and XL go up, so does frequency, which divides into Z and XL to give us our inductance reading,
Z and XL values were scaled to fit the graph so their values can not be read off the Y axis,
now we have a nice flat inductance curve leading up to the self rez peak,
we will correct the high end curve for distributive capacitance next,

 

[CJ]

more experiments and calculations done on this V72 choke,

DCR from the magnet wire in the choke will shift the peak resonance curve to the left (lower frequency),  so the actual peak due to inductance and distributive capacitance will be further to the right, probably be around 2400 Hertz,

http://www.allaboutcircuits.com/vol_2/chpt_6/5.html

at self resonance, the choke presents a very large impedance to the 20 volts AC applied during testing, this leads to errors in the inductance calculation because the current through the LC circuit is very small,
this leads to a large Z, but not large XL,  thus, we can not use the current at the rez peak for calculating inductance, so the self res peak should be edited out of the graph,

so we have corrected the low end, the peak, now we should look at the high freq section of the graph as far as error caused by distributive capacitance,
in a parallel LC circuit, total current = I-L + I-C,

so if we subtract the current due to the distributive capacitance from the total current, , we will be left with the current through the choke, which can then be used to calculate inductance,

we do this by calculating X-C at say, 10 pf, for the high freqs above the peak,
and subtracting this from total current measured, which yields inductor current,

as it turns out, 10 pf is small compared to the Henries of the choke, even at high freqs where the core loss starts to kill the inductance,

this means that the current from C-d will be small compared to the choke current,
so when we subtract this C-d current from the total current, inductance will actually go up a bit as the lower the choke current, the higher the inductance,

so the curve for inductance above the self rez peak will not be changed much after the correction for C-d,

here is the graph showing the difference>

 

[CJ]

here is the corrected graph for the V72 choke,

bass flattened, peak removed, high freq bumped a tiny bit,

next: a rewound V72 choke with the original ferrite pot core and 21,000 turns 

you guys like the bass/drums on Spiral Staircase-More Today Than Yesterday?

http://www.youtube.com/watch?v=cAcr7i4jA2I&list=PLSXEWU-h31MfoZl7bwkOJQ_r4mjNzGvuG

 

[abbey road d enfer]

this leads to a large Z, but not large XL,  thus, we can not use the current at the rez peak for calculating inductance, so the self res peak should be edited out of the graph,

Absolutely right.

now we should look at the high freq section of the graph as far as error caused by distributive capacitance,
in a parallel LC circuit, total current = I-L + I-C,

so if we subtract the current due to the distributive capacitance from the total current, , we will be left with the current through the choke, which can then be used to calculate inductance,

we do this by calculating X-C at say, 10 pf, for the high freqs above the peak,
and subtracting this from total current measured, which yields inductor current,

That is all correct.

as it turns out, 10 pf is small compared to the Henries of the choke,

You can't compare picofarads and henries; you can compare the reactances.

this means that the current from C-d will be small compared to the choke current,

No. At resonance, the current in the cap is the same as the current in the inductor, whatever the values, whatever the resonant frequency.

so the curve for inductance above the self rez peak will not be changed much after the correction for C-d,

Yes it will. Nature abhorrs discontinuity. The inductance value should be a more or less straight line.
You need to check your calculations.
Using your graph, with L=450H and Fres=1.8k, I get C= 1/L (2.pi.F)²= 17pF
This is a more accurate result than the RDG chart
At 1.8kHz
Z_L= 2.pi.F.L= 5.1Megohm
Z_C= 1/2.pi.F.C= 5.2Megohm
Close enough.
Indeed the current in the inductor is identical to the current in the cap.
Ther are some discrepancies because of the DCR that makes the current in the inductor not strictly in phase opposition with the current in the cap.

EDIT: I see you have updated the results; that is definitely much closer to expected. It shows the influence of iron losses, although I think you have been a tad pessimistic. The sudden drop in apparent inductance seems to be too steep.

 

[CJ]

<using your graph, with L=450H and Fres=1.8k, I get C= 1/L (2.pi.F)²= 17pF..>

anti-resonance effect of the 25 k ohms of DCR will shift the rez peak up higher if DCR is zero, so i used 420 H and 2.4 k Hz to get about 10 pf,

math for calculating the exact shift in frequency is pretty heavy, but i guess i could use a spread sheet,

Inductance>u * N^2, so as u changes, L changes,

we can look at core loss curves for 29 ga M6 vs freq and flux level to get a better idea of the roll off,

air core inductance is pretty high with 21,000 turns so freq response of the V72 amp is pretty flat depending on the OPT response,

 

[CJ]

ok we are going to explore the effects of DCR on the resonant frequency of a

Parallel Resonant Circuit

this is the V72 inductance in parallel with the distributive capacitance, C-d

it turns out that even with 25 K Ohms of DCR, the self resonant frequency does not change that much,

we found some equations that were not too rigorous, (no j operators or vectors)

with a tighter gap we have 480 Henries for the V72 inductance, if the peak of our old graph is correct, this yields a C-d of 16 pf,

Q of the inductor at resonance will be 2 pi f L/DCR,

= 6.28 * 1800 Hz * 480 H / 25,000 = 217

we can even do damping factor for any propeller heads out there,

DF= DCR/2L = 25,000/960 = 26

here is a pic which shows the math for determining the DCR effects on the resonant frequency, as well as a formula to check our results which uses the Q of the choke,
these are "dead guy" formulas so you have to divide by 2 pi to get the freqs,

 

[CJ]

rewound the V72a ferrite choke and did a plot next to the steel core version,

this V72a amp has the choke in the cathode not the output tube instead of the anode,

it only has about 20 Henries, stock choke is more like 250-400 H,

since the Henries re less, the rez pek is out at about 9500 Hz instead of 1400-1800 Hz like the steel core,

this round coil calculates out to about 14 pf of coil capacitance, not too much different than the 17-18 pf of the rectangular bobbin in the V72 chokes,

here is the graph and a winding pic,

 

[CJ]

did an out of circuit test on the ferrite choke,

looks like it does not like DC, as when DC is removed, Henries go way up,

so the core is saturating,

wonder if a paper gap in between core halves will make this thing work better,

E180F  (output for V72a)  draws just as much current as EF804s, so the tube is not an issue,

here is a plot of the ferrite choke, DC vs no DC>

 

[CJ]

just for kicks we removed the cores from both the steel core choke and the ferrite choke to see how the coils alone reacted to frequency,

getting a bass boost even after DCR is factored out, this could be the 21,000 copper windings acting as a core, copper does have a perm, if that is the case, it does not work well beyond about 100 cycles,

both coils had the self rez peaks at the same frequency,

lower inductance due to no core moved the peaks out to about 10 K Hz, C-d stays the same but inductance drops without the cores, so the peaks move to the right,

there is more copper in the square coil and more Henries, leading me to believe that the copper wire is acting as a core,

will look at inductance formulas for air core inductors to see what the Henries should be for these coils,

 

[abbey road d enfer]

could be the 21,000 copper windings acting as a core, copper does have a perm

Most of the times non-ferrous metals are approximated as vacuum or air. Permeability of Cu is 0.999994 so unless your windings are actually Ni, the "self-core" action seems unmeasurable.
Going from 320H with core down to 10H without the core indicates the relative permeability of the core to be about 1000, which is consistent with most ferrite cores.
It's also consistent with the increase in resonant frequency and the decrease in "Q" factor.

 

[CJ]

tried a couple of Wheeler formulas for air core inductance,

they both say about 20 H for the V72 coils, which tested at 5 H (round) and 8 H (square)

so these formulas do not work well with three chamber coils,

here is the V72a ferrite choke with a 20 mil gap,

Henries came up to about 100, big improvement, and the rez peak came back,

so this choke obviously gets a gap in the V72a amp,

there is a 25 uf output cap in the V72a instead of 4 uf like the V72, so this might help the bottom end as the ferrite choke has 100 H instead of the 250 H like the V72,

DC really makes the pot core halves clamp down on each other, MMF in action,

 

[abbey road d enfer]

here is the V72a ferrite choke with a 20 mil gap,

It's surprising you had to put a shim to make the gap. The AL1000 core has a gap ground in the center jamb. In fact the bottom half of the core has the center jamb perfectly flat with the outer ring. The top half is the one that governs the AL. By grinding the center jamb of the top, it adjusts the AL of the whole assembly; that's why the reference and the AL are printed on the top.

so this choke obviously gets a gap in the V72a amp,

No doubt about it, considering the plate DC goes through it.

there is a 25 uf output cap in the V72a instead of 4 uf like the V72, so this might help the bottom end as the ferrite choke has 100 H instead of the 250 H like the V72,

That's really because the primary of the OT has much less inductance in the 72A than in the 72. About 5 times less.

 

[DaveP]

Outstanding work CJ,

Between you and the Mod, you have given us the best this forum can do.

Stick with the New Year Resolution

Respect
DaveP

 

[CJ]

here is a better graph of the V72 choke,

turns out you can not hook any leads to the choke or you get big erros,

so the cathode resistor of the output tube was used as a voltage sensor ,

as well as cathode current being directly measured by breaking the wire and inserting the ammeter,

Ollie was right , their is no resonance in the audio band, all those peaks are caused by stray C in the leads,

so get the the joke of the week award for posting a bunch of garbage, live and learn,  ??? :-\ :'(

however, at least i think we have it right this time,

here is the new graph for an iron core version,

sorry about that, i will get all the other choke versions tested over so we can compare them correctly,

 

[DaveP]

Thomas Alva Edison, on his Light Bulb research:-

"I have not failed, I have just found 10,000 ways that don't work"

Your Electronics Tutor would be very proud of you, we are 8)

best
DaveP

 

[CJ]

ok i made a cluster out of this thread, but finally getting some good numbers,

disconnected the OPT so the AC cathode current came down which means Henries came up,

getting 250 H at 20 Hz which is seems normal,

notice XL stays flat even though Henries are coming down, this means the freq response will be flat as long as the OPT circuit does not mess it up too much,
Henries come down with frequency as the core perm comes down,
Henries is proportional to turns^2 x perm,
core perm comes down because of core loss and a decrease in flux density
flux density relates to frequency as B=V/f, with V is being held at 20 v-ac,


there is very little ac current going through the 30k screen resistor, so the Henries plot is not changed after subtracting this current from the cathode/choke ac current,

so to test these chokes:

1)  use it in the actual circuit for correct application of DC
2) do not attach any test leads to the choke itself while measuring current, use the cathode circuit
3) disconnect the output transformer,

this graph has a log plot for the Y axis to improve upper freq Henries resolution,

 

[CJ]

just a graph from above with plate and screen currents added,

notice as XL goes down, currents go up, well doh on that, 

notice that ac screen current is about 1000 times smaller than plate current,

y-axis scale numbers are  x10 times higher for currents,
no dual scales allowed with this graph program,

 

[CJ]

here are 3 V72 chokes and 1 V72a choke plotted as L vs f

all converge to about 10 to 11 Henries at 1 k hz and ramp down at the same rate from there,

V72a ferrite pot core is shown with a 10 mil gap and with no gap,

 

[CJ]

when you increase the gap in an inductor with no DC, inductance will always go down,

however, if the inductor carries DC, you can actually increase the perm of the core by adding a gap which means the inductance will go up,

there is a point where a bigger gap will start to lower inductance,

so the trick is to find the gap which maximizes inductance,

this is called the optimum air gap

you can see that on this graph which shows Henries vs Frequency,

a 1 mil gap creates more inductance than a zero mil gap,
the no gap line is the black line with X's

DC will saturate a core which lowers perm, take the core out of saturation a bit and the perm will come up,

a big gap will be like removing the I bars from an EI core, so i did the plot for no I bars also.

notice that different gaps change the Henries mostly from 200 Hz and on down,

why would you use a 5 mil gap if it gives the same Henries as no gap?
because it stabilizes the core, if you have no gap and the choke heats up and the gap changes  a little, perm will change a lot, but not so much with the 5 mil gap,

also shown is inductive reactance for different gaps,

notice the larger gap shifts the peak to the right, so the Q of the inductor is moving up in frequency as the gap is increased,