PDA

View Full Version : Core material BALUN MFJ 9982



PA1ZP
10-24-2012, 09:34 PM
Hi

Does anybody know what type of material the core of the balun in a MFJ 9982 is?
I m building a homebrew tuner in a manor that looks a lot like the design used in this MFJ 9982 tuner but need to know the core material.
i have seen this tuner on the inside and to me it seems to be a 3 time stacked FT240-43 ferrite core.

Maybe someone can help me with this.
Another help could be the uH of the 9 windings of this Balun having the uHenry would be good enough to figure out the materials used in the BALUN.

thanks in advance

Jos

WX7G
10-25-2012, 01:37 AM
The balun used in the MFJ-9982 has four cores and is available from MFJ as part number 10-109982.

Four stacked FT240-43 cores with nine turns works out to 350 uH at 1 kHz. But what is it at RF? This chart by G3TXQ shows what eight turns on a single FT240-31 core is. Four cores should provide four times the impedance.

http://www.karinya.net/g3txq/chokes/

PA1ZP
10-25-2012, 04:53 PM
Hi WX7G

Do you have a idea what material MFJ uses in there 4 x stacked cores?
Tnx for the reply, but as I have a good cheap source of Amidon cores both FT or T type, I would like to build the transformer myself.
Also the shipping US - Europe would cost a few bucks.
Also have good quality Teflon wires available here.
That is why the uH are in the question.
if I new the uH I could easely find out what materials I could use.
I also could then make the choise to wind the balun on say on 2 times T400-2A or so, because my tuner is a bit larger and uses far better plate caps and a real fine quality Russian ceramic army surplus roller-inductor.

But thanks a lot for the reply, I didn't have an idea that MFJ would have these parts for sale.
I knew that Ameritron has a lot of parts of there amps on the shelf.

73 Jos

G3TXQ
10-27-2012, 04:36 PM
Jos,

This is a real tough application for a balun!

First, consider the need to achieve reasonable current balance in the load: the load could be as high as 4000 Ohms in a doublet/ladderline application, possibly more. To achieve around 15dB current balance we would want the tuner input choke CM impedance to be about 10,000 Ohms across the operating range. On 80m that would require 454uH, or 80 turns on two stacked T400A-2 cores! Iron powder cores just don't give enough impedance/turn for this type of application.

A ferrite core is a better option because of the much higher permeability, but then we have to worry more about power dissipation because the CM impedance is likely to have a large resistive (loss) component: 1kW into a balanced 4000 Ohm load means a CM voltage of 2000v across the balun. If we limit the balun dissipation to 20W for a stack of four FT240 size toroids, we would require a CM impedance of 200,000 Ohms if it was wholly resistive !!!!!

In practice we might avoid the worst-case load impedances, and the mode duty-cycle might reduce the average power dissipation, but I'd be aiming for an absolute minimum of 10,000 Ohms for the balun CM impedance with a target of 20,000 Ohms.

As Dave said, you may find the chart on my website gives you some guidance. But be aware that the CM impedance is a function of both the winding reactance/resistance and the winding stray capacitance, so although simple scaling for multiple-toroid stacks is probably OK at frequencies away from the SRF it may not give accurate results close to SRF.

Hope some of that helps,
Steve G3TXQ

WX7G
10-27-2012, 05:11 PM
Since you want to duplicate the MFJ circuit what could they use but type 31, 43, or 61 cores? Looking at the fine G3TQX chart I suspect MFJ used type 43.

PA1ZP
10-27-2012, 09:01 PM
Hi Steve and Dave.

I do not want to duplicate the MFJ.
I want to use the balun at the 50 ohm side of the tuner.
For now I m using 2 x T300-2 cores with 2 x 20 windings silver plated and PFTE isolated wires.
It does the job but the MFJ tuner has a greater frequency reach and I suspected the cores would be Ferrite instead of iron-powder cores.

Thanks a lot for both answers and I already dug through your website Steve.
Compliments for all that info.

We will try the Ferrite materials, as I can reason it also could be 61 material as even the stacked T300-2 cores already are doing a decent job.

The tuner I am building and wich already is in use can be found at my QRZ listing.
It uses 2 of these caps, with welded plates on both the rotor and stator plates, and these caps are 6 inch wide 7 inch high and 10 inch deep.
These caps rotors are balanced with a 1/4 inch thick stainless steel plate because the rotor will fall down on its own weight. These things weigh 3 Kg + a piece

The variable inductor is a Russian made inductor without the cranking, rattling and jimmying and small contact wheel.
This inductor has all ceramic insulation and has bearings on its 8 mm shaft.
I can tune these caps and coil even when there is 1KW key down on the tuner and the caps and inductor won't mind a bit

The cabinet of my tuner is all brushed stainless steel and is 20 x 10 x 8 inch and the total weight of the tuner is 50 pounds or 23 kilo's
This tuner is a try to upgrade my 3KW CW (PA0FRI) S-match to a 3KW system that workes from 10 - 160 mtrs as the S-match only handles from 17 to 80 mtrs.

This "Soyuz" tuner is only made the this way because I can and it fits my paddles and Ahtonob noise-canceler.
The cabinet and knobs are of my own design and laser cut.
Even the brushed satinless steel and laser engraved gear driven inductor winding counter is my own design and building.
All hardware is stainless steel.
It is just Stainless steel because it can be laser engraved in black, with all text and scales and pics, you do not have to paint it you do not have to letter or write on it , it is all there.
If you want to clean it just take a soft cloth with WD40 oil and clean it.

PS

This tuner is a one off just like the noise-canceler and paddles or Energia 4-square.
All these things are totaly to expensive to even consider production.
And we own the firm GPM.nl in the Netherlands a big thank you for your help, they let me use all there high-tech CNC equipment in realising these projects.

This tuner is a bit over the top project , it will be time to pimp my AL572B as next project, maybe I will call it the Apollo project in honour of Neil Armstrong.
The Soyuz tuner is designed in honour of Andre Kuipers , the Dutch record holder cosmonaut of spending time in space.

The noise canceler has some Russian design in its guts.

73 and thanks a lot for both your replies.

108862

WX7G
10-27-2012, 09:17 PM
On the 50 ohm side of the tuner 50 ohm coax should be used for the balun, not 100 ohm (twisted pair line).

However, if you want to use the Teflon wire you can make 50 ohm twisted pair. Wind the twisted pair clockwise. Take two lengths of it and twist them together counter-clockwise. Connect the appropriate wires at each end and you have 50 ohm twisted pair (me and another guy at Micron "invented" or re-invented this a some years back).

G3TXQ
10-27-2012, 09:27 PM
I want to use the balun at the 50 ohm side of the tuner.
For now I m using 2 x T300-2 cores with 2 x 20 windings PFTE wires.
It does the job but the MFJ tuner hs a greater reach and I suspected the cores would be Ferrite instead of iron-powder cores.

Placing the balun at the input of that type of tuner doesn't alter the calculations on the required CM impedance. 20 turns on two stacked T300-2 cores will have an impedance of about 240 Ohms on 80m - that's way too low to achieve good balance with moderate to high impedance loads.

73,
Steve G3TXQ

G3TXQ
10-27-2012, 09:29 PM
On the 50 ohm side of the tuner 50 ohm coax should be used for the balun, not 100 ohm (twisted pair line).

Dave,

That's not a problem if the SWR is measured on the input side (rather than the tuner side) of the balun. Any impedance transformation will simply be "absorbed" in the tuner adjustment.

73,
Steve G3TXQ

WX7G
10-27-2012, 11:23 PM
I had considered that but suspect the bandwidth could be reduced by having a length of 100 ohm line between the tuner and the radio.

PA1ZP
10-28-2012, 07:39 AM
Hi Steve and David

Thanks for the replies.
The comment that 240 ohm is way to low, that are things I can work with.
I read on Steves website that his baluns that worked well were of impedances of 8 Kohm.

I didn't have a clue why that was neccesary.
We will go for the ferrite cores and will wind with PFTE wires like David described.
I do not use one flimsy wire I used 6 of them parallel.

I tried my tuner against the old Cubic (Swan) ST3 B
I used the ST3B on the 2 x 12 mtr wire dipole with open line feeders , this dipole has 2 x 8 mtrs spiderweb coils at each end of the 12 mtr legs.

I used the big guy on the 2 x 7 mtr rotatable dipole.
Both dipoles are fed with 11 mtr 600 ohm balanced lines with rose clips (wire is red copper braid 13 AWG wire).

On 3600 KHz now the sigs coming of both these antennas were about the same signal strength.
On both TX as on RX.
Then I took the Cubic ST3-B to the rotatable dipole and the big one to the wire dipole.
Now the sigs were about 20 dB in favour of the wire-dipole.
I did these experiments several times before with the S-match and the results were the same every time i tried.

Do not try to tune a short dipole and a very low Z at the end of the open wires with a T-match with a 1 : 4 balun at the open wire side.
The losses will be terrible and very high.
Even well known balanced ATU's like the palstar BT1500 double L match or even the PA0LL tuner will have troubles in loading an antenna with a R of 2.5 in the open-wire feed point.

That is the reason why we use 6 wires parallel in the balun that is why I use welded plate caps , and that is why I use red copper plate of 5/8 inch wide as wiring in the tuner.
That is why we use the big Russian roller inductor and not the ones with the flimsy wheel contact.
These inductors will be destroid in a few hours.

My friend PE1KQP blew the extra capacitors 3 times at his S-match due to high currents.
He is now on the look for a big ceramic 20 KV doorknob cap of 400 pF
He blew up his 600 ohm line on 160 mtrs when he used a short peace balanced line of 18 AWG red copper wires to get in his house.
On open wire systems currents can get very high.
Make sure the tuner can handle this.

I tried a MFJ 949E on the short dipoles and I didn't dare to use more then 5 watts on 80 mtrs.
At low Z's losses are made very quickly.
I have blown to many and seen to many blown parts in these little MFJ's hihi.
Melted down pillars of inductors melted down caps , blown balun cores melted L-switches , fried antenna switches.

The best design for a low loss balanced tuner on Low Z is a double C tuner with the L between the C's at the radio side.
But still losses can be about 50% trying to tune very low-Z like 2.5 ohms with -J 180, but even the 11 mtr open wire makes 5dB loss when using 13 AWG copper braid wires with a 600 ohm balanced line at the 2 x 7 mtrs dipole.
Double-L tuners will have about 75% losses against the double-C having "only" 50%.
But the trouble with these very well balanced tuners are that you do need to switch extra C at low bands parallel and getting the right big switches is a problem like the capacitors that can handle the high currents.
That is why I went for the floating T-network, I do not need switches or extra parallel caps.
And I still can tune from 10 - 80 mtrs

Thanks again for the explanation and help.
We will trie the stacked Ferrite FT240 cores with the twisted Teflon wires.
We will try both the 61 and 43 materials, and report back to you here on this forum.
It could take a few weeks though , because QRl is very buisy at this moment.

73 Jos

G3TXQ
10-29-2012, 07:33 PM
I had considered that but suspect the bandwidth could be reduced by having a length of 100 ohm line between the tuner and the radio.
Not for the sample load impedances I just tried the maths for ;)

Steve G3TXQ

PA1ZP
11-07-2012, 05:05 PM
Hi Steve and Dave

Wound a new balun on 4 stacked FT240-61.
I tried 2 x 15 windings, I tried 2 x 7 windings.
2 X 7 at 40 uH windings did a fine job.
2 X 15 windings was also good for balance but didn't alow tuning on several bands

The tuner is now easely tuned from 160-80 mtrs.
I haven't had time though to try very high power on 160 , I only tried it at 20, 40 and 80 meters .
Up to a KW CW no problem (I can not make more power hihi).

Thanks again for the help.
73 Jos

PS a few pics of the result.

109894109895109896109897

G3TXQ
11-10-2012, 09:04 AM
I just measured the impedance of 4 turns on various combinations of stacked FT240-31 cores at 5MHz. A small number of turns and a relatively low frequency were chosen to stay well below the SRF of the chokes:

1 core: Z=497 (337+j365)
2 core: Z=996 (697+j712)
3 core: Z=1535 (1119+j1051)
4 core: Z=2135 (1635+j1374)

So, the impedance ratios compared to the single core were:
2 cores: 2.004
3 cores: 3.088
4 cores: 4.296

Due to manufacturing tolerances there is a variation in μ' and μ'' between the 4 cores used, so we would not expect to measure the "ideal" ratios.

For practical choke construction, WX7G's advice is sound - the impedance is directly proportional to the number of stacked cores. However, note that this only applies at frequencies well below the SRF - as we approach the SRF things get much more complex.

Steve G3TXQ

WX7G
11-10-2012, 03:46 PM
Here is something from the Magnetics company:

How are properties affected when cores are stacked?

Stacking core will increase the cross section (Ae) by the multiple of the number of cores in the stack. The magnetic path length (le) will remain constant. The AL can be estimated by the same method as for single sets, where a leakage adjustment is estimated based on the ratio of window area (WA) to core area (Ae). Because that ratio decreases as cores are stacked, the AL of n stacked sets is slightly less than n times the AL of a single set.

WA8KJP, can you take some measurements to compare to those of G3TXQ?

G3TXQ
11-10-2012, 04:59 PM
One further measurement: I had a drawer with five FT240-52 cores, so tried those:

4 turns on a single core @ 5MHz: Z=200 Ohms (15+j199)
4 turns on five stacked cores @ 5MHz: Z=1004 (74.4+j1001)

Five core/single core impedance ratio = 5.02

Steve G3TXQ

G3TXQ
11-10-2012, 05:23 PM
Forcing air gaps between the five stacked cores by placing wooden matchsticks between them had the following effect:

4 turns on a single core: Z=200 Ohms (15+j199)
4 turns on five stacked cores tightly clamped with plastic tie-wraps: Z=1004 (74.4+j1001)
4 turns on five stacked cores separated by matchsticks: Z=1017 (75+j1015)

Steve G3TXQ

G3TXQ
11-10-2012, 07:50 PM
@WX7G:

Dave, the advice you gave in Post #2 is sound and was needlessly challenged - four FT240-31 cores would provide four times the impedance of one core at frequencies well below the SRF. I've checked that now by four different methods and get the same result.

If you look at Jim Brown's published charts you can see an extension from my 4 turns on 4 cores, to 4 turns on 6 cores. The charts are not easy to read, but if there is any drift from the Nx relationship it is to make the impedance slightly higher than expected with larger numbers of cores.
http://audiosystemsgroup.com/RFI-Ham.pdf

73,
Steve G3TXQ

WX7G
11-10-2012, 09:27 PM
Steve, thanks for taking measurements.

As you know, the reason spacing the cores apart makes no difference is because flux does not couple from core to core. The flux flows about the axis of each ferrite core and does not flow along the axis of the cores.

WA8KJP, if you can contribute some measurements it would provide evidence for or against the hypothesis that stacking cores increases the inductance by a factor equal to the number of cores.

If I recall, awhile ago you had the opportunity to provide some linear amplifier measurements to provide evidence for or against the hypothesis that decreasing the RF output power of a linear amplifier (after tuning for maximum power) by decreasing the RF drive level reduces the DC-RF efficiency and does not substantially reduce the amplifier plate, collector, or drain dissipation. Perhaps you can take this data on your transceiver or amplifier and post it.

WX7G
11-11-2012, 04:33 PM
I am after the the truth, it is as simple as that. I suspect that Steve is also after the truth. The truth furthers the art of amateur radio and can save someone time, money and effort. An untruth can cost someone time, money and effort.

The argument here concerns stacking cores. It was said that the inductance is not proportional to the number of cores and it was said that an air gap between cores decreases the inductance. The scientific method is used to provide evidence for a hypothesis or to refute a hypothesis. This is done by stating a hypothesis, designing experiments that are able to disprove the hypothesis if it is false, and performing them. The experiments are described so that others can duplicate them. Theory and references can be provided to further support or refute a hypothesis. The scientific method is best applied dispassionately, without personal attacks and with humility.

Hypothesis: Inductance is not linearly proportional to the number of stacked cores:
I provided text from a Ferrite core manufacturer to support the hypothesis that the inductance is linearly proportional to the number of cores. Steve designed a lab experiment and reported the results and the details of his experiment. The hypothesis is disproved.

Hypothesis: Inductance is reduced by air gaps between cores because the flux flowing between cores is altered. Steve designed a lab experiment and reported the results and details of his experiment. I explained why flux does not flow between cores. The hypothesis is disproved.

This has been a valuable effort that and invites further discussion and raises new questions. Two questions I have are:
1. What is the relationship between stacking cores and power handling?
2. How does stacking cores affect the SRF (Self Resonant Frequency)?

After this I would like to see Steve write an article for Eham.net. This knowledge can then be used to the advantage of amateur radio operators.

G3TXQ
11-11-2012, 05:42 PM
This has been a valuable effort that and invites further discussion and raises new questions. Two questions I have are:
1. What is the relationship between stacking cores and power handling?
2. How does stacking cores affect the SRF (Self Resonant Frequency)?
Dave,

Interesting questions! Some initial thoughts:

1. The power handling likely does not scale directly with the number of toroids: for the duty-cycles encountered in typical amateur operation the thermal limit is reached well before core saturation with the typical ferrite mixes we use in baluns; and that thermal limit is a function of the total surface area of the toroid stack. In a vertical, tight, stack of N toroids we have N times more inside and outside "edge" area, but only the same top and bottom surface area as for a single toroid. The issue is complicated because there is also a contribution from the winding copper loss, but I reckon a stack of N toroids wont handle N times the power that a single toroid would.

2. The effect of toroid stacking on the SRF is complex. I think we now understand that the inductance scales proportionately to the number of toroids; but at the same time, the inter-winding capacitance increases if we wind the same number of turns on a larger stack. The net result is a somewhat larger drop in SRF than is explained solely by the increased inductance.

I think there is some evidence for that view in Fig 43 of Jim Brown's paper: for example the drop in SRF from 5 turns on 4 cores, to 5 turns on 6 cores, is more than we would expect from the plotted increase in inductance.

With regard to an article: for some time I've been meaning to add some balun pages to my web site - maybe the time has come ;)

73,
Steve G3TXQ

WX7G
11-12-2012, 11:39 AM
Steve (and anyone else who cares to comment) I have some balun thoughts I'd like you to comment on and correct me if I'm wrong.

SRF: Let's say we have a balun wound with X turns on one core. We then wind a balun with X/2 turns on four cores to obtain the same inductance.

Assuming the core cross sectional area is square (height Y) the wire length for the one core balun is 4YX. We stack four cores and halve the number of turns. The wire length is 10YX/2 = 5YX. The capacitance is proportional to the wire length and so the capacitance has increased by 5/4.

The SRF of the four core balun is (4/5)^0.5 that of the one core balun.

G3TXQ
11-12-2012, 01:48 PM
Dave

I follow your reasoning, but there's a couple of problems I see:

1. The core permeability is frequency dependent, so the inductance is not a constant
2. I wonder if the total stray capacitance really is directly proportional to wire length.

However, I just tried measuring some chokes. I used FT240-61 cores because they are relatively high-Q and produce much more clearly defined SRF characteristics. Here's what I got:

8 turns on a single core: SRF = 46MHz, wire length approx. 19.25 inches
4 turns on a 4 core stack: SRF = 36.3MHz, wire length approx 21 inches

From data on the #61 mix I can estimate the difference in choke inductance at 46MHz compared to 36.3MHz, and from that estimate the change in parallel capacitance as 1.52pF to 1.38pF. However, from that we need to subtract my test jig stray capacitance measured as 0.2pF. That suggests the choke stray capacitance changed from 1.32pF to 1.18pF; that's a ratio of 1.12:1 compared to a ratio of wire length of 1.09:1 - very close!

These very small values of stray capacitance make the measurements fraught with difficulty, and I'd want to repeat them for larger numbers of turns before coming to firm conclusions; but the one measurement set seems to support your analysis.

73,
Steve G3TXQ

WX7G
11-12-2012, 02:57 PM
Steve, thank you for the measurements!

Now for a balun question I'm not as certain on.

Power Handling:
A one core choke balun exhibits a temp rise of T degrees for W power. All the loss is core loss.
Wind a four core choke balun having half the number of turns. The inductance is the same.
The total flux is the same but the flux density is 1/4.
Assume that core loss is proportional to the square of flux density.
The loss per core is 1/16 that of the one core current balun.
The total balun loss is 1/4 that of the one core current balun
The surface area (square core cross section) is 10/4 of the one core current balun.
The power handling has increased by 4 x 10/4 = 10 times.

G3TXQ
11-12-2012, 04:51 PM
Dave,

I don't think that's correct, at step 3!

Assuming the same Inductance/Impedance for the two chokes, and the same voltage applied and current flowing, the total flux must double in the 4-core stack because we have halved the number of turns [Faraday's Law]. Work that through your remaining steps and the total power loss becomes the same in the two cases. So, taking surface area into account the power handling has increased by 2.5 times.

Another (simpler) way of looking at it:

If the impedance of the single core choke is Z=R+jX, and we reproduce that on an N-core stack by adjusting the number of turns, we must still have the same R+jX because the loss tangent remains the same. So, for a given CM voltage across the choke and CM current through it, the power loss must be the same in the two cases.

73,
Steve G3TXQ

WX7G
11-12-2012, 04:58 PM
Steve, thank you. That is the step where I suspected I was wrong. This thread and the comments from WA8KJP (that brought up more questions and answers) have really furthered my understanding of current baluns.

ad: ki5dx-1