开云体育

I did a few systematic experiments using my nanoVNA to measure the properties of 49:1 transformers. It helped me to understand these beasts better... Write-up here:

Cheers,
Gary AF7NX

An excellent piece of work Gary.

thats all very interesting ... I'll need to read it a few times to 'digest' all your findings. thanks.

Very nice writeup Gary!

I may have missed it, but curious about how using two FT240 cores stacked
will behave vs total turns.

I built one of these transformers with two FT240 type 43 cores stacked,
with 2:14 turns. Do you think that will behave as well in the low-mid
frequency ranges vs. a single core with 3:21 turns?

And based on the "two-edged sword" aspect of the compensation capacitor,
would you recommend not using it, if antenna won't be used above 20m band?

I'm mainly only interested in using antenna on 20m, 40m and 80m right now,
with max power of 100W, mainly for CW and SSB.

Hi Bryon,
If you look further down to the "core size" experiments you will see that I did indeed test 2 FT240-43 cores with 2:14 turns. It is a reasonable choice, and with the large volume and surface area of ferrite, you should be able to handle 100W without any trouble. As long as the transformer doesn't get too hot on 10m, the compensation cap is probably a good idea because otherwise you may not be able to get a good SWR and not be able to drive the band very well.

Good luck!
Gary AF7NX

Gary - very interesting!
I would urge you to also post notice of your work on QRZ.com. There is active discussion of EFHW 49:1 technology in the "Antennas, Feedlines, Towers & Rotors" forum, with significant contributions by retired EE professor Mike Mladejovsky, WA7ARK.
* This should provide a link to the Forum:
* This should provide a summary of posts from WA7ARK:

Have you yet had opportunity to evaluate FT-240-52 mix cores? It's my understanding a triple stack of 52 cores offers better performance (versus 43 mix) for higher power HF applications - reference this 2019 YouTube video from Steve Ellington:

It would be very interesting to specifically compare FT-240-43 versus 52 mix cores for HF use with EFHW transformers.

Would you please take multiple pictures out of the box so we
can see the winding technique and how things are connected!
It appears the secondary is center tapped. Please - out of the bos!

Dave - W ?LEV

On Thu, Jul 1, 2021 at 8:29 PM Fred, W9KEY <fred.schwierske@...>
wrote:

Gary - very interesting!
I would urge you to also post notice of your work on QRZ.com. There is
active discussion of EFHW 49:1 technology in the "Antennas, Feedlines,
Towers & Rotors" forum, with significant contributions by retired EE
professor Mike Mladejovsky, WA7ARK.
* This should provide a link to the Forum:

* This should provide a summary of posts from WA7ARK:


Have you yet had opportunity to evaluate FT-240-52 mix cores? It's my
understanding a triple stack of 52 cores offers better performance (versus
43 mix) for higher power HF applications - reference this 2019 YouTube
video from Steve Ellington:

It would be very interesting to specifically compare FT-240-43 versus 52
mix cores for HF use with EFHW transformers.

--
*Dave - W?LEV*
*Just Let Darwin Work*

Hi Dave,

Yes, the winding is tapped as an autotransformer and not a bifilar separate primary. I didn't think it would make much difference. But after seeing a couple of comments, I measured things both bifilar and as autotransformer, and indeed, the bifilar winding seems better.

Can you explain why that should be?

Cheers,
Gary

Gary
Have you considered what the feed point impedance of an EFHW is against the equivalent source impedance the 1:49 transformer has created? Is there a reasonable conjugate match? I have measured the feed point Z of a 66 or 67 ft wire on 40 meters (all I can use with my small lot) and recently I have worked with 4NEC2 and there is a fair agreement to the measured impedance. I have raised these question to others and get absolutely no responses. Many of us have been using a resistance of 49*50 for a test load as we both have as a starting point but the real antenna is not a simple R.
Pete

Interesting article, thanks! I think the best way to measure/model the transformer is to take a proper full two port model using the nanovna. This is a bit fiddly to do with a TR VNA but worth the effort I think. I then allows a much more detailed analysis of the transformer on a linear simulator.

I may have missed something but the insertion loss of the dual FT240 transformer looks to be very low below 2MHz. Is this taken with a 50R source and 2450R load? Or is it matched for lowest loss at 2MHz?

I did look at one of these transformers a while back using 2 x FT240 cores but I didn't wind it as an autotransformer. I wound it as a regular 2:14 transformer. I got the best performance up at 28MHz with the turns well squished together into just part of the core. With a matching cap and a 2450R load I think I achieved about 0.5dB loss at 28MHz. This setup isn't ideal for power handling but the insertion loss was impressively low across 2-30MHz. I think it was less than 0.7dB everywhere but the loss at 28MHz was optimised at just 0.5dB.

On Fri, Jul 2, 2021 at 12:35 AM, WB2UAQ wrote:

Many of us have been using a resistance of 49*50 for a test load as we both
have as a starting point but the real antenna is not a simple R.
Pete

I think that is one of the reasons the transformer should ideally be measured as a full 2 port device using a full two port VNA. This can also be done with a TR VNA like the nanovna with a little patience. Once this is done the s2p model can be loaded into a simulator and the load presented to port 2 can be changed to a complex load at will. To make things more realistic, a powerful RF simulator can define port 2 as an s1p model taken from a real antenna for example.

Pete, I've modeled with 4NEC2 an end-fed wire with various source impedances and counterpoise lengths. The 2450 ohm source impedance seems close to optimum. You have to choose some step-up ratio. Once you do, then the match to 50 ohms becomes whatever it does and provides a good starting point for determining the properties of the transformer. Certainly the SWR will vary a lot depending on the antenna load. But one thing at a time :)

Cheers,
Gary AF7NX

JMR,
Indeed I was using the two port S11, S21 provided by the nanoVNA. If you go through the Appendix in my post you can see how I generated the loss plots I presented. Once I established a calibrated system and a consistent method of doing the tests, the runs were quite repeatable and gave reasonably self-consistent data. It is what it is! You should be able to repeat the same experiment from what I've laid out, I would hope, if you have lingering doubts...

Gary AF7NX

Hi Fred,
I appreciate the feedback. I will eventually post a link on some other forums.
I don't happen to have any mix 52 in my collection or I would have tested it. Looking at its complex permeability, I can see why it might be favored. The lower permeability allows there to be more turns on the primary, which will imply a lower B field for the same primary inductance -- That is probably a good thing for less loss. The competing issues would be larger leakage inductance and more winding capacitance...
Cheers,
Gary AF7NX

I take it back... Nothing magic about bifilar. Best is a soldered tap to the desired wire. Next best is a twisted wire loop tap. Third best is bifilar and worst is two separate windings. This assumes both windings share the same ground. This can all be explained with keeping the leakage inductance as low as possible. (see attached plots).
Gary AF7NX

Maybe of interest: