1:1 RF Transformer Design SWR problems

[ChristianFletcher]

Hello all, Many thanks for any help and advice in advance.

I am trying to build a 1:1 RF transformer to break the coaxial screen of the feeder from my internal radio shack to an antenna system that will have an earth stake. So this is basically the same as the isolation transformer used to prevent ground loops in audio circuits but this will be used at radio frequencies.

I have had a go at building a few of these using various core materials and winding configurations but these have not been satisfactory as they give a high SWR.

Unfortunately I have found the subject of RF transformer design somewhat complex (very) as the more I read online the less I understand it.

Requirement 1:1 Transformer 80 to 28m Feeder 50 Ohms Power 100W Core T200-2

1) For optimum transformer efficiency primary reactance should be 4-5 the impedance being transformed at the lowest frequency.

Impedance of the Primary will be same a secondary as the transformers 1:1 = 50 Ohm
Primary reactance = 4 x 50 = 200 OHM


2) So L = XL / (2*Pi*f).
If XL is 200 Ohms and f is 3.5MHz, L = 200 / (2*Pi*3.5*10^6), or 200/(2*Pi*3.5) uH.

L = 9.09uH.


3) A T200-2 dust iron core has an inductance factor of AL of 7.5uH per 25 turns.
N = 25*sqrt(L/AL) N = 25*sqrt(9.09/7.5) = 27.5 turns. 27 or28 turns.

4) I built the above using on what I think is a T200-2 Core bought from a radio rally so this core could be anything in fact. But at least the its red so that’s ok then  I have measured the following Primary characteristics using LCR meter at measurement frequency of 200khz. The analyser gives this information:

L = 9.2uH RES = 0.04.

The measured inductance looks close what I had calculated for the number of turns. Primary and secondary windings read the same.


5) I first tested the coil using the spectrum analyser to look for losses. I have included the print and a picture of my test set up. The output of the tracking generator was set a -10dbm and the display normalised. I think this looks good.


6) I then look at the SWR using a mini VNA. The SWR is horrible. The secondary is terminated using a good quality 50ohm pad. I also tried directly connecting the coil to the VNA to remove the coax but the SWR is unchanged.

This is where I become very unstuck to understand what wrong here or what I need to do to attain a good flat SWR over the lower HF bands. I assume the SWR graph is telling me that the Independence must not be a match to 50 ohms but I don’t understand why not.

Please could I have some advice about how to debug this circuit or is my underlying design just wrong ?

Very Best regard Chris 73

[Ian - G4JQT]

If you post the question on the RSGB Tech forum (https://groups.yahoo.com/neo/groups/rsgbtech/info - you don't need to be a member I don't think) or the specialist Yahoo group Ham-Antennas (https://groups.yahoo.com/neo/groups/...tions/messages) there are loads of people who will be able to help.

Not saying no one will be able to help here, but it may be worth a try.

Ian

[ChristianFletcher]

Thanks Ian

I will if I get no response. But I have been a not very active member here for a many years and I like this forum. Its a friendly place to be and a nice community. I perhaps feel more comfortable demonstrating my ignorance here than to the wider world.

I know there are some RF black magic wizards on here

Chris

[G6Tanuki]

That looks a hell of a lot of turns!

When I've used similar 1:1 transformers I've usually used about 12 turns (bifilar wound) and that's been good for 3-30MHz.

[ChristianFletcher]

Thanks - Well I could try with less turns and do the measurements again. I was trying to follow the formula rather than just trial and error. I'm more interested in why my transformer does not work and understanding out to fix from the data.

Chris

Use a common mode choke i.e. a fair amount of co-ax wound round ferrite. DC wise it is still connected but any RF will be isolated solving your RF ground loop problem.

[Radio Wrangler]

T200-2 is a dust-iron core, not ferrite. But it's probably better for this application.

Making a transformer which works efficiently over the full MF-HF amateur bands is a serious undertaking. Making a transformer which works over the range a lot of antennae span is an even larger undertaking. Making a transformer which does both at once is a very large task.

One way out is to use an unbalanced-to-balanced ATU. Something like the Z-match or KW107/109. They will also drive an unbal antenna.

David

[Jon_G4MDC]

If you check here http://www.amidoncorp.com/t200-2/ - I have one of these in an amidon balun kit and that picture is very close to reality.

I think you will see that the red of the core is quite a bit darker than appears in your photo so I suspect you do not have a genuine T200-2.

Your sweep shows some return loss at the low end 1-2MHz. I suspect some low frequency ferrite in that core.

The inductance you measured came out near what you expected at 200kHz - perhaps that was just luck?

You could go for this and be sure of what you have. http://www.amidoncorp.com/ab200-10-kit-1/

[ChristianFletcher]

Thanks David and Merlin and Jon

Ok so maybe Im doomed to failure with this one. This is off topic but explains what I'm trying to achieve.

Just for clarity there must be no DC connection at all. Im trying to isolate the outside earth stake from a fault condition when the neutral becomes open circuit feeding a PME wired installation. Under this condition the fault current for my house would try to return to earth via my antenna feeder.

It been suggest to capacitively couple the feeder to the earth stake and this will have a high impedance at 50 hz but low at rf frequencies. I think this will be the easier solution but I want to explore the RF transformer idea and learn a bit more about transformer design. etc.

Jon you may be correct about the core as is was bought from a radio rally stand so who knows. The one I have is darker but it got a wrap of red PVC tape so I don't scratch the insulation of the wire. The inductance was however spot on for the number of turns that I calculated. As you say maybe I got luck. At least if I buy a core from a known supplier I will be sure. Thanks

Regards Chris

Would it be easier to isolate the incoming supply? Or capacitors and a common mode choke?

[Jon_G4MDC]

thanks for explaining about the tape! I hadn't allowed for that

If I can find mine I might be able to see how it looks used as you intend. I do understand what you are trying to achieve. I wonder where it is....

[Ian - G4JQT]

(Not sure how relevant to the OP, but anyway...)

I used a 4:1 voltage balun (12 bifilar turns on T200 toroid) for my comudipole (pdf here, pages 15.11 and 15.15). That's a sort of G5RV, but although symmetrical, all the horizontal and vertical dimensions are just whatever you can fit into your plot.

The balun went at the bottom of the balanced line to the coax. In addition I used a 1:1 current balun over the coax to reduce any current in the coax outer. This 'balun' is just 12 ferrite toroids threaded over the coax where it leaves the 4:1 balun. You can buy them for the G5RV and similar antennas.

In effect the comudiople is a doublet. But not many shacks are placed at the centre of the doublet for the ladder to elegantly enter the balanced ATU. Hence one of the reasons the G5RV and similar ZS6BKW antennas are so popular - despite their shortcomings - because the balanced line is conveniently fed with coax which can meander for some convenient distance.

(This arrangement of mine is due to be published in the next edition of SPRAT.)

[ChristianFletcher]

Thanks Ian - Thats a good article you linked.

Thanks

[ChristianFletcher]

Originally Posted by merlinmaxwell

Would it be easier to isolate the incoming supply? Or capacitors and a common mode choke?

Well easier said than done. I could feed via a mains isolating transformer. But it all gets a bit iffy with multiple earth paths etc. There must be a about 20 cables running out the back of the computer and to the radio and all my other gizmos.

A few years ago I thought it may have been a good idea to bond to the radiator copper piping in my shack. This year my house flooded as the mice ate through the pipes - All plastic under floor. So much for that idea !

[G0HZU_JMR]

Originally Posted by ChristianFletcher

This is where I become very unstuck to understand what wrong here or what I need to do to attain a good flat SWR over the lower HF bands. I assume the SWR graph is telling me that the Independence must not be a match to 50 ohms but I don’t understand why not.

Please could I have some advice about how to debug this circuit or is my underlying design just wrong ?

My first impression is that your thru loss and your VSWR tests are not comparable because you aren't controlling the tests very well. It looks to me like your thru loss test is really just testing the transformer as a badly grounded transmission line. A bit like a 1:1 current balun but you have swapped the ground ends of the short coax cables. I'm assuming here that the ground on the Tgen and the RF input on the analyser are connected to a common metal chassis.

But your VSWR test is very different. Here you have fully floated the load end and I think your VSWR plot reflects this change. Note also that a 5:1 VSWR at 20MHz doesn't tally with a low insertion loss so this is also a useful clue that something is wrong between your two tests. You aren't comparing apples with apples because you changed the grounding between the two tests?

I would agree with Merlin and recommend you adopt the classic 1:1 Guanella or current balun as it is often named. You could get a DC block with a capacitor inline?

[ChristianFletcher]

Hi Jeremy

Thanks for you comments I had wondered about the difference between the loss and the SWR graph but your comment has highlighted this. I was really patting myself on the back when I did the loss plot. Looks like you have found the smoking gun.

I am sure that the input and TG port of the analyser must share a common chassis ground. I do often install an additional 50ohm attenuator to help flatten off the output from the TG but it sounds like this would not have helped in this case

Do you know if there is a better settup I could employ that does not in itself require some form of special transformer to issolate the ports on the analyser.

I think I may have read somewhere about when testing 4:1 current baluns doing this as a pair. I assume that this was just to provide a 50 Ohm match to the measuring instrument but perhaps this would help.

The question is from the data I how do I determine where the problems is. Is it the material or the number of turns etc. I note there is a transformer availabe for less cost than I have so far spent on the cores I do wonder if it works any better than my efforts.

If I had gone out and bought one I would not have learned anything. Even if the lession is it wont work.

Many Thanks

Chris

Originally Posted by G0HZU_JMR

I would agree with Merlin and recommend you adopt the classic 1:1 Guanella or current balun as it is often named. You could get a DC block with a capacitor inline?

Yes indeed, it is the inside of the co-ax that transmits the power and determines the impedance, you can chuck it through anything you like and it won't change (1:1 VSWR), an air cored coil to reject UHF/VHF, round some ferrite for HF/MF, round an iron core for LF/VLF, all in series if need be. And via some capacitors to block DC (as said) small ceramic multilayer types could even be made in the right dimensions to be a 50 ohm line if you need UHF and above.

[ChristianFletcher]

Thanks Merlin - Thanks everyone for the replies

I think I will end up grounding my remote ATU via a capacitive shunt to some well placed radials and a ground stake. Im not quite ready to give up on the 1:1 transformer design as yet until I understand what causing the problem and why it wont work. There is a commercially transformer available quite cheaply and I am tempted to buy this to find out how they did it assuming this device works.

[G8HQP Dave]

Did you use bifilar winding? The photos seem to suggest you did not - perhaps to get good voltage isolation?

Separate windings mean weaker coupling, which won't necessarily cause loss but will cause impedance mismatch.

Have you considered trying bifilar winding, but using wires with mains-level insulation rather than enamel?

[G0HZU_JMR]

My interpretation of Chris' design goal is that he wants to wind a transformer with terminations like the image below. This seems a bit unconventional in that I'd expect most people to opt for the classic Guanella/current balun topology instead.

But I think this will not work well with the transformer he has wound with the terminations arranged as the image below. This is because you can't neglect the transmission line behaviour of such a big transformer with such long windings. I'm going to guess that there's nearly 1.5m of wire in that transformer. If I guess the Zo of the transmission line to be up around 80 ohm I'd expect the VSWR at the source to be in the order of 4:1 at 25MHz with the load floating as per the image below.

So I think it's dead already, even before you ask questions about how well calibrated the VNA is in terms of a SOL calibration.

I think you could begin to improve the performance by reducing the Zo of the transmission line either by twisting the bifilar wires or by using two bifilar windings in parallel. But if you parallel up the windings to reduce Zo you will start to lose the LF performance.

Normally, you would try and use a ferrite core with high permeability allowing you to minimise the winding length so you get good LF to HF/VHF performance. But you have to be wary of the power handling of the core itself.

By contrast, one big bonus of winding a classic Guanella/current balun (with shorter wires and a ferrite core) is that the power handling can be very good even with a modest sized ferrite core