Telefunken V72 Choke Rewind

[CJ]

today we are going to rewind a V72 choke.

these chokes are very unique in that they use a 3 chamber bobbin, tons of turns,

and a lamination that will have you guessing as to how to take the choke apart,

since this choke caries DC, it has to have a gap.

if you move the gap 1 or 2 mils, inductance will change a lot.

temperature or mechanical shock can upset the gap, but there is a clever solution,

you stamp a lam with a built in gap,

but how do you get the coil on? well, you just bend the tongue a bit.

after all the lams are in place, you stick a shim under the coil to lock the choke together,

the only disadvantage is that you throw away some material,

but if you want a fantastic product, you do things like this,

see if you get it by looking at this pic>

 

[CJ]

audio chokes are just like the primary of an audio output transformer primary,

they need to have better frequency response than a pwr supply choke,

so if you wind thousands of turns, you need to do something to lessen leakage C.

by using three sections, you lessen C by a factor of 3^2 = 9 times.

here is the lead breakout, there is a slight groove for the wire to pass between sections,

now each turn puts a nano newton of force against the wire next to it, so if you have 10,000 or 20,000 turns, this force gets multiplied like water behind a dam, so you need to either watch you winding so that you do not over stuff the margins, or block the unfilled sections with something so that the bobbin does not distort, or both.

 

[CJ]

they wind the sections as 1, 2 and 3,

if you wind it 1, 3, 2, the 1 and 2 sections will have a larger voltage gradient between turns, this means more capacitance as C depends on voltage as well as plate size,

V72 chokes should have about 250 Henries, the V76 has 400.

this is because the V72 uses a different OPT than the V76

the OPT has to match the output tube, in this case we want to drive the OPT with an EF804s.

that is a new power board next to the coil, a bit thicker as the originals get baked and are prone to cracking.

next we will rewind the V72 pwr trans with fewer sec turns in order to match the B+ with the new silicon rectifiers, as the originals used a real funky selenium jobby.

then we will wind an input and output and ship some V72 units.   

we have the dimensions for the pwr and main circuit boards, if anyone is interested, we can post pics,

 

[CJ]

here is a worksheet for the V72 plate choke,

that weird lam with the built in gap has no match in the Mag met catalog,

so we just use the tongue width which was 1/2 inch and find a EE lam has a 1/2 inch tongue.

this would be the 50 EI, 50 means 1/2 inch, so we use those specs for L , A and MPL.

since the gap is on the tongue only, we divide it by 2 since the legs are not gapped.

this lam has a built in gap of about 16 mils so we use 8 as our gap.

we use mills since gaps rarely go over 30 thousandths, so 1 thru 30 is a good notation system for gap length.

notice the dramatic drop in perm with the gap, as well as the drop in Henries.

this takes core perm variations out of the equation which is desirable.

also notice that will all those turns, flux is very low, less than 3 kg on a core material that will take 18 kg, so distortion is very low.

tubes like the EF804 want about 400 Henries when operated as a pentode,

this is because the plate resistance is high and we do not want to drag the tube down.

now with a gapped inductor or transformer, you have to take into account the DC flux as well as AC flux, this DC flux is caused by the plate current, so we add the flux caused by the signal voltage and the flux created plate current.

there are also calculations for different wire sizes just in case we can not squeeze on all the turns and have to jump to a thinner gauge wire.

some V72 circuit show 300 H, the V76 uses a 400 H choke, so we go right in between with 350 H which gives us a nice round figure of 18,000 turns which divides nicely by 3 into 6000 turns per bobbin chamber, which should be a piece of cake with #44 wire.

to get MTL, or mean turn length of wire, we just take a piece of stiff wire and bend it around the bobbin about 2/3 rds of the way out , straighten it out and measure it,

then we can get wire length by turns multiplication which can then be extrapolated the DCR from wire tables. there is a wire table link listed at the bottom which is the only one on the net that has #48 thru #60 specs.

wire tables will have turns per sq inch, so multiply the bobbin width by the height, this will give you your winding area in sq inches, then simply multiply the two numbers to get the theoretical amount of turns on that bobbin section.

EF804s only uses 0.6 ma plate current, it has the coldest running cathode of just about any pentode out there at 680 degrees C, that is why you can barely see them glow,

voltage drop across even a choke with 12,000 ohms will only be .6 times 12 = 7 volts.

note that on some Supersonic V76 prints, the plate voltage on V2 is shown at 175, when in fact the B+ is 245. this would mean a DCR of 100,000 ohms on that choke which is obviously in error, so watch out for that, plate voltage will actually be about 245 - 7 = 238 VDC, not 175 VDC.

this means most of the 245 B+ volts can be used for amplification. 

 

[CJ]

we have 3450 turns and tons of room, we might make 7000 per section for 2100 turns = 475 Henries gapped,    13,000 un-gapped,  :  you want to fill the window with as much copper as possible as to make the Q go up, we should have 60,000 ohms pri X-L @ 20 hz,

but what about the extra hi end loss with the extra 3000 turns?

we can cut back to 400 Henries by leaving room for an extra layer of tape at the end, or we could find a way to extend the high end by using tricky winding techniques,,

original V 72 choke has about 30 Turns per Layer, we can use 20 T/L and improve turn to turn C which gets multiplied by the extra turns,

and there is another trick, it is called fold-back winding,
you lay down 20 -25 turns per layer, then traverse back in 5 turns and lay down 20 more, traverse 5,
what this does is it lowers the voltage between turns getting the wire back over to the start quicker, since voltage goes up as you work your way thru a coil, the turns over the start will be closer to the start, which means less voltage between the start and these turns as had you waited to wind a complete layer which means you 1.3 V over the start instead of 2V.

use fold-back and more T/L, you can dramatically lower the leakage C and raise the high end while maintaining killer bass.

attach leads to #44 Before you start winding, otherwise if you get to 21,000 turns and snap the wire while hooking it up, you have to start over.  :-[

if you snap the wire while doing your leading it at the beginning of the wind, it is no big deal, just pull more wire thru and solder that sucker up.

sand the bobbin to prevent snagging the fragile #44, bakelite color revealed, they use the killer insulators in these modules, they hold charge better than nylon,

this is the stuff rupert neve used to shock his customers with a non energized xfmr by storing charge in the coil by applying high voltage D.C.,
just like a capacitor left on the bench by your covert co-workers,  8)


ever listen to GLT blues radio from Chii Town, lets break it down one time only,

http://wglt.org/streams/bluesPlayer.shtml

 

[CJ]

ok we managed 7500 turns per section = 22,5000 turns,

this is the stock turns for the Tele V72 plate choke, bass should be awesome,

DCR came out to 15,460 ohms, so maybe they use #43 instead of #44,

 

[CJ]

lacing it up like a converse all star,

you can grab 3 at a time and bend, then 2 as the bobbin  fills up, then 1 at a time to get the last few in,

 

[CJ]

this thing is a monster!

700 henries at 5 hz, 700 Henries up to 500 hz where X-L peaks at 2.2 meg.

least excitation current at 1,300 hz at 1.6  micro-amps, that is low.

so this thing goes 5 hz to 40 k like a walk in dead man's park,
X-L at  60 Khz.  is about 100,000 ohms, then things drop off fast the HP craps out and all that.

we can hookup a 9 v batt for 0.6 ma DC and see how the core does, should be interesting, we can see what DC does to the rez freq.

if you wanted to improve hum, you could reverse wind the center section and thus have 2 sections hum-bucking which reduce hum by 67 percent on paper at least,

but then you would have to splice that micro wire, plus, this thing is going inside about 3 layers of mu metal if you count the housing,

 

[CJ]

we are going to check the inductance specs with DC with the coil inside the preamp,

why?

weii, if you hook a 9 volt battery across this thing and isolate the generator from the DC with a big cap, all the juice from the sig gen will go into the battery! doh, 
this drags the HP down from 24 volts rms to 2.5 volts rms, no good

darn, because 9 volts across the dcr of this thing produces exact;y 0.6 ma, oh well.
this is how you learn stuff. it can be painful and embareazz at the same time, 

this thing was drawing 35 ma excitation current from 5 to 20 kc,,and i thought i was gonna have to rewind the coil and put a reverse wind on there to cancel DC flux, but since it is the battery, we will not worry and install and check. 

so if you want to apply DC, you have to use a hi voltage supply and then put a big resistor in series with the DC and coil, even then, the generator will still be feeding a bit of ac current to the DC filter caps on the output of the HV supply via the 100 k resistor or whatever you use,

tech heads would say,"you need to use a hi z source, bendejo maricon!"  ;D

EF804s plate R at 238 volts is 400,000 ohms, so you would need a 400,000 ohm resistor in series with the DC supply to simulate the same feel as the actual circuit,

how much voltage do we need for the DC current simulation?

well the coil has 15,460 and we have 400,000 ohms series R, so we have 415460 ohms  @ 0.6 ma = 249 volts,

you could use the B+ supply from any tube project to do the DC measurements, just be careful when running 250 across the room, use some coax or something like that,

we have a Fairchild 670 pwr supply as our tube rack ext supply, it has a adj knob, so we will tap that one time.

 

[CJ]

we built a test jig to apply the DC offset from a tube supply,

ground, +6 vdc (heater voltage) and 250 variable DC come in on that orange extension cord from the rack supply octal plug output jack, the heater was added in case we want to use this jig for bread boarding a tube circuit,

that is a 470 K 3 w resistor on the board, the DC supply goes from 190 to 300 VDC, so we have the 0.0006 ma dc current dialed in, those parallel caps are non polarized 6.8/250 so we can get the bass excitation currents, they keep the dc off the 600 ohm balanced output from the sig gen.

we have a Optoelectronics freq counter that can also be used as a portable freq scanner to keep track of the local PoPo's secret UHF transmission freqs,  ;D

and also a Micronta 22-164 from rat shat that is the only talking voltmeter i have ever seen, too bad, as it is really a clever idea, and a real help when you are under the bench and can not see the meter, they discontinued it after the first run, so not too many around, there is a small speaker built in and a button on the probe to trigger the robotic voice:
one  ---point------eight -------  five-------milli---------volts-------D------C------

so that B+ and resistor  can be thought of the EF804s as a current source for testing the effects of DC.


be back with the inductance measurements with DC,

 

[CJ]

ok the test works great with the tube supply feeding the DC,

here is the breakdown:

5 Hz  794 Henries - 776 Henries with DC. 
10 Hz  768 Henries - 665 Henries with DC
15 Hz  810 Henries  - 659 Henries with DC.
20 Hz  791 Henries  - 612 Henries with DC.
100 Hz  811 Henries  -  465 Henries with DC.
500 Hz  810 Henries  -  139 Henries with DC.
1 KHz  585 Henries  -  68 Henries with DC.
2 KHz  194 Henries  -  31 Henries with DC.
20 KHz  0.843 Henries  - 0.723 Henries with DC.

Test Level:  24 Volts -28 Dbu

numbers in red are sketch as the caps used to bypass the sig gen may have contributed to bogus readings at the low end, seen this with scopes set on AC, ie, an increase in inductance when you expect a decrease,

interesting that the DC  has the most effect at 1 KHz and no so much the very bottom end, maybe 5 hz is almost DC anyway, so a little DC current does not get noticed down there,

with no DC, the inductance of the non polarized inductor gets extended frequency wise by a factor of about 10, but both start to lose Henries quick, as do most transformers, that is the upper end of core steel frequency response, you do not need much inductance past about 1 K as the 2 pi F  starts to dominate the reactance formula,

so if this is a 400 Henry choke as noted on the schematic, that means they mean 400 H at 1 kHz with no DC, or 400 H at 100Hz with DC,
i am going to go with the "no DC"  spec at 1 kHz, Sowter also lists their V72 choke inductance as 750 H with no DC, so that makes sense.

Frequency Response looks good, using a gauge smaller wire drops the leakage C down as the size of the wire is the size of the capacitor plate,

ok , choke over, power transformer next,

 

[zebra50]

Nice work CJ!

By coincidence, here's a couple of shots of a V72 power transformer that I had on the bench this week....

And some measurements, CJ stylin...

BTW, if you have trouble with the heater supply on these units, check for a short circuit near the lamp / adjustment pot on the front panel. 

 

[CJ]

wow, a man of my heart!

exact same one over here.

if you need turns, stay tuned.

 

[zebra50]

Well, would be nice to see inside. This one is working now we found the short circuit, so I won't be getting out the hacksaw!

 

[CJ]

you might want to take it apart after you see what's inside this one, 

doing the math to see if i can wind on 625 EI in nickel

this would lower heat and keep 60 cycle flux inside the core for less hum,

BTW, that plate choke peaked out at 250 K Hz!

kind of  a bessel , not a bissell, that's a vacuum cleaner

or could be M derived, it's all lunch meat to me,

a whooping crane not tasted, is a whooping crane that's wasted,

some say save the baby seals, they are getting hurt,

i say save the baby seals, save them for desert, 

 

[nielsk]

If you ever decide to make a few of these, I am in need of a choke & output to finish my spirit of '72 project... maybe help fund your appetite for exotic waterfowl?

 

[CJ]

a C core version would be the way to go,

just 2 winds on 2 coils and you have  4^2=16 times less C instead of 9,\

plus it would be a hum-bucker,

and the gap is very stable if you clamp it right,

one weird thing about these high turn coils, you ohm them out on the core and you get 1 answer, then when you remove the core you get another answer, something to do with digital meters i believe.
also, symmetry causes leakage in coils, so you try to break symmetry by interleaving sections, or maybe reverse winding, or maybe fold-back winding,
so one minor tweak you can do is if you do fold-back winding, try doing the first half of the coil with the quick layer starting from the left, then half way thru, change the quick 5 turn layer so that it goes right to left, this creates an angle between the larger turn layers, and the smaller turn layers, the will cross over each other and maybe cancel some hummmm

 

[CJ]

spreadsheet for the V72 choke,

 

[gemini86]

CJ, getting ready to wind my choke here... couple questions:

1) what is the window size on the original lams? EI50 is something like .75 tongue length (the old oddball lams I have are .814), which seems way too small to fit that many turns in there

2) Is it even possible to wind one of these without a precise traverse guide?

I have enough copper to waste on some trial and error, but I just tried to wind one chamber and maxed out at about 6000 turns and my counter started skipping beats. starting to wonder if I should be spending more time rigging up a legit winder with a traverse cam setup (oh boy, can't wait  :)

 

[CJ]

you do not need a traverse to wind this choke,

travel the wire fast enough so you do not get crossovers, try for about 30 turns per layer, crossovers will lower your turns count per bobbin,

stock choke values are 250 H for the V72 and 300 H for the V72s,

so i need to adjust that spreadsheet up above, although i have tried 250 H and 400 H  and can not tell the difference,

you have to inject a 20 hz signal and take current measurements to actually see any difference between 300 and 400 Henries,

so if you can get 6,000 turns per chamber then you are right on track to get 18,000 turns total which adds up to about a 250 H choke just like stock if you use a 3 to 5 mil gap. when you butt stack the lams, you get a natural air gap of about 5 mils for small lams, the book lists 5/MPL x 10^3 as the formula for butt stacked lams of the smaller type like we use for the choke, so if 50 EI and 625 EI has a magnetic length of 3 and 3.5 inches, then you would have a natural gap of 16.6 mils and 13.3  mils,

these figures are probably worst case numbers as i get about half that on the work bench,

notice that the smaller the lam, the larger the natural gap as the imperfect surface to total surface area of the butt joint ratio gets bigger as the lam gets smaller,

a 400 H choke will extend your low end by about 10 hz over the stock 250 H choke.

this snap in lam is taller than a stock 50 EI lam, so a stock bobbin will not fit the original lam,


choke dims>