A serious effort at getting decent data to help understanding. Very well presented and detailed enough for the average Joe like me to get understanding. One of the best, if not the best, U-Tube presentation of this subject. Thank you. !!!!!
Howdy Bob! I tell you for a 14 minute video this one took some prep work, winding those roids was a real bugger... I had to solder and everything... lol.
Awesome work. Thank you. We can go even further in this quest. I’ve measured my 42ft random wire antenna impedance without any transformer from 7mhz to 28mhz. The impedance varies a lot and it’s very high on the lower bands (1270+324j on 7mhz) and starts going down as you increase frequency (790-1380j at 10mhz, 127-310j at 14mhz, 330-144j at 18mhz, 209-577j at 21mhz and 122-230j at 28mhz). So the 450ohm is a very simplistic assumption and the fact that a unun is not constant across the whole spectrum is not necessarily a bad thing depending on the impedance your antenna exhibits. Looking forward to see more videos of this kind! 73 de F4JAW
Many thanks for making these measurements. The measurements made here are one way to start, but we might reach quite different conclusions regarding the relative merits of the three items if the items were tested more fully. Vector network analyzers are wonderful tools, but we must be careful in interpreting results. This first test with two ununs connected “back-to-back” reveals just how an unun will perform (at least for small signals) when it is connected to another unun. This point is obvious, but we must not assume that this tells us how the unun will perform when connected to a different load. The use of the S21 measurement tells us the ratio of the output wave to the incident wave, but by itself it tells us nothing directly about the power loss or efficiency of the device or devices under test. For instance, suppose we test a lossless device that presents a bad mismatch at its input. We will see a low magnitude of S21 but that is due to a high magnitude of S11. A high S11 tells us that incident power is being reflected back at the input rather than being lost as heat inside the device. We have to distinguish between the power incident on the device and the power that that actually goes into the device. The formula to calculate the efficiency of the device, expressed as a ratio of the net power that actually goes into the device at the input port to the power that comes out the output port, is S21 magnitude squared / (1 - S11 magnitude squared), where the magnitudes are expressed as ratios rather than dB. The other test looking at only S11, when the device is connected to a resistive load, shows us that there is transformer action going on assuming the loss in the device is low, but it doesn’t tell us anything about the relative merits of these three devices when used in an actual system that will in general exhibit some other impedance. To determine by measurement of s-parameters how the unun (or any device for that matter) will perform with other loads, we need to measure all its s-parameters as complex numbers and apply a set of formulas that include the actual complex load impedance that will be connected to the device. It looks like the measurement loop may have an inductance of roughly 250 nanohenries due to the wire lengths and loop area. At 30 MHz, that would be an inductive reactance of about 47 ohms, or about 10% of the intended 450-ohm impedance level. Thus, the wire length is probably having some noticeable effect on the measurements at 30 MHz. At the low frequencies, there would be proportionally less effect. We also need to keep in mind that the answers we get when testing with small signals may not be representative of what will happen when we apply substantial rf power from a transmitter. For instance, the magnetic core may exhibit enough nonlinear behavior to throw things off, and the device may be heated enough that the wire resistance changes and the core loss factor changes. Given the tests that were done, plus assumptions that there will not be too much power dissipation or nonlinear behavior when used in our system, we can say that all three are candidates for use. We can’t say which would be absolute best in our system, but since all the results are similar, it seems likely that we wouldn’t notice any difference in on-the-air contacts and we would be okay with any of these as long as the impedance presented to our transmitter is within acceptable boundaries.
Loved watching this thorough and methodical analysis of each number of windings. it's also quite cool to witness someone having all the right test gear.
This is a cool video for sure…I made one of these using a ft37-43 core, just much smaller. A 470 ohm on the input showed the nearly 1-1 SWR, but not over the full hf band on mine. Interesting way to do it using two of these back to back for the insertion loss and match z on the second port on the nanovna.
Thanks for the demonstration. Lower number of turns would be good for higher frequencies because of increase in inter-winding capacitance and higher number of turns would be good for lower frequencies. Ni-Zn ferrites like #43 (ui850), #52(ui250),#61(ui125) and the "K" material can be tested for the specific frequency band/s. Higher permeability cores can be used for lower end of the HF spectrum and lower permeability cores can be used for the higher end of the HF spectrum. But then, the magnitude of power handling becomes an important factor too. Higher permeability cores would get saturated quickly when operated on the lower end of the HF bands; so, you have to select bigger, bulkier cores. For the higher end of the HF spectrum you can use lower permeability cores and with lower number of turns; you get the required inductance and at the same time the inter-winding capacitance is also taken care of. De VU2RZA
@@steveshelton5884 yes, sure. You can use lower permeability cores like #61,#52 or "K" ferrites. Ferrite mix #61 is usually my choice. It has a high Curie temperature, lower saturation and hence can withstand higher flux, and the likes. I am using it in my 1.2 KW SSPA (HF) (using 4×SD2933 MOSFET devices) for the broadband output transformer. And I have two such modules using hybrid splitter and combiner to put out 2.4 KW without any problem. The cores remain as cool as cucumber. De VU2RZA
Hi Ape, really enjoyed the video. It's great to see things outside of the box. I have not tested my 9 to 1's for loss, but have found that a 9-1 made with a FT-240-43 has better SWR in the low band, 20-160m and one made with a T-200 powder core has better SWR in the higher bands, 6-20. Twenty meters appears to be the point that both are somewhat close in SWR. I've finally broke down and bought a VNA, so I will be doing more testing in the future. My unun's are all 9 wraps. Have not tried anything different in that regard. I always enjoy you technical videos. Keep up the good work.
Hey John. The Ferrites generally are more broad-banded than the powdered iron cores. When you want a do it all solution you do end up making some compromises. Congrats on getting the VNA, it’s going to open up a new world for you 👍
@TheSmokinApe 100%. If I had to choose, it would be the 43 mix. I made one of each I can take into the field. That way, depending on the band, it gives me options. I know, a little anal, but I love making antennas and unun's. In the end, not a huge difference either way.
Spot on Ape. Trying to improve those losses is pointless as to what you would gain. I honestly thought the loss would be more. I have those 2 little cores here set up to do that test. I think there will be a difference with materials and geometries but not a massive amount. 61 material may be a good shout for 40-10 as Evil Lair has demonstrated.
I was using this method to test a 1:49 toroid wind and was scrambling to find just a few resistors to out in series for a load…when it hit me that I had an inexpensive resistor decade box. Dialed the 2,450 ohms in and tested. I think I paid < $20 on Amazon.
Great to see the differences in performance - but there's another test you could do. Its not a test the NanoVNA can do, however. Using an actual transmitter, vary the power level in and compare to the power level out. At some point, the core will saturate and you start losing power as heat. You should probably do this with T140 AND T240 cores, to help users decide when a bigger core is a good idea.
nice Ape explains why everyone says 9:1 :) btw to get 450 ohm, try a 270 and 180 in series, or if you have an 820 and a 1k, use them in parallel (but thats closer to 451ohms)
Maybe throw a 49:1 in there maybe?? that is some very interesting data. Great video! If I still had the gear to test this myself I"d try it in a second. It would be very interesting to know if I'm using the wrong unun!
One is to convert the signal from 50 to 450 ohm and the second is to bring it back from 450 to 50, the systems impedance of the meter (NanoVNA) is 50 ohms.
Howdy Don, the Ares 9:1 comes with a 140-43 and we suggest a 9x3 wrapping but it’s sold as a kit to make your own. I’ll have to remember how to use the Libre 👀
Check out my Chokes, Baluns and Ununs Playlist: ua-cam.com/play/PLm8ROkpFeqoqkGQtBYfcT9Y-2GCKxBuuk.html
A serious effort at getting decent data to help understanding. Very well presented and detailed enough for the average Joe like me to get understanding. One of the best, if not the best, U-Tube presentation of this subject. Thank you. !!!!!
Thanks Dave, glad you liked the video 👍
TheSmokinApe doing all the hard thinking for the rest of us. Thanks Ape.
Howdy Bob! I tell you for a 14 minute video this one took some prep work, winding those roids was a real bugger... I had to solder and everything... lol.
@@TheSmokinApe When I saw the first 3 I thought "that took some time".... Then you said you had a second set! Dedication to your craft!
LOL, I somewhat "enjoyed" it 👍
Awesome work. Thank you. We can go even further in this quest. I’ve measured my 42ft random wire antenna impedance without any transformer from 7mhz to 28mhz. The impedance varies a lot and it’s very high on the lower bands (1270+324j on 7mhz) and starts going down as you increase frequency (790-1380j at 10mhz, 127-310j at 14mhz, 330-144j at 18mhz, 209-577j at 21mhz and 122-230j at 28mhz). So the 450ohm is a very simplistic assumption and the fact that a unun is not constant across the whole spectrum is not necessarily a bad thing depending on the impedance your antenna exhibits. Looking forward to see more videos of this kind! 73 de F4JAW
Correct, you are adding the unun to the circuit and impedance with change with frequency due to reactance 👍
Many thanks for making these measurements. The measurements made here are one way to start, but we might reach quite different conclusions regarding the relative merits of the three items if the items were tested more fully. Vector network analyzers are wonderful tools, but we must be careful in interpreting results.
This first test with two ununs connected “back-to-back” reveals just how an unun will perform (at least for small signals) when it is connected to another unun. This point is obvious, but we must not assume that this tells us how the unun will perform when connected to a different load.
The use of the S21 measurement tells us the ratio of the output wave to the incident wave, but by itself it tells us nothing directly about the power loss or efficiency of the device or devices under test. For instance, suppose we test a lossless device that presents a bad mismatch at its input. We will see a low magnitude of S21 but that is due to a high magnitude of S11. A high S11 tells us that incident power is being reflected back at the input rather than being lost as heat inside the device. We have to distinguish between the power incident on the device and the power that that actually goes into the device.
The formula to calculate the efficiency of the device, expressed as a ratio of the net power that actually goes into the device at the input port to the power that comes out the output port, is S21 magnitude squared / (1 - S11 magnitude squared), where the magnitudes are expressed as ratios rather than dB.
The other test looking at only S11, when the device is connected to a resistive load, shows us that there is transformer action going on assuming the loss in the device is low, but it doesn’t tell us anything about the relative merits of these three devices when used in an actual system that will in general exhibit some other impedance. To determine by measurement of s-parameters how the unun (or any device for that matter) will perform with other loads, we need to measure all its s-parameters as complex numbers and apply a set of formulas that include the actual complex load impedance that will be connected to the device.
It looks like the measurement loop may have an inductance of roughly 250 nanohenries due to the wire lengths and loop area. At 30 MHz, that would be an inductive reactance of about 47 ohms, or about 10% of the intended 450-ohm impedance level. Thus, the wire length is probably having some noticeable effect on the measurements at 30 MHz. At the low frequencies, there would be proportionally less effect.
We also need to keep in mind that the answers we get when testing with small signals may not be representative of what will happen when we apply substantial rf power from a transmitter. For instance, the magnetic core may exhibit enough nonlinear behavior to throw things off, and the device may be heated enough that the wire resistance changes and the core loss factor changes.
Given the tests that were done, plus assumptions that there will not be too much power dissipation or nonlinear behavior when used in our system, we can say that all three are candidates for use. We can’t say which would be absolute best in our system, but since all the results are similar, it seems likely that we wouldn’t notice any difference in on-the-air contacts and we would be okay with any of these as long as the impedance presented to our transmitter is within acceptable boundaries.
Thanks for the detailed post, it's much to consider. I appreciate it 👍
I think my head exploded. Excellent work and very methodical testing.
Lol, thanks for watching Todd 👍
Appreciate the lesson Professor!
Thanks for watching Shane 👍
Loved watching this thorough and methodical analysis of each number of windings. it's also quite cool to witness someone having all the right test gear.
Thanks Dennis, glad you liked it 👍
This is a cool video for sure…I made one of these using a ft37-43 core, just much smaller. A 470 ohm on the input showed the nearly 1-1 SWR, but not over the full hf band on mine. Interesting way to do it using two of these back to back for the insertion loss and match z on the second port on the nanovna.
Thanks Walt, glad you liked it. This resistors can sometimes introduce inductance and higher frequencies and raise the SWR 👍
Thanks for the demonstration.
Lower number of turns would be good for higher frequencies because of increase in inter-winding capacitance and higher number of turns would be good for lower frequencies.
Ni-Zn ferrites like #43 (ui850), #52(ui250),#61(ui125) and the "K" material can be tested for the specific frequency band/s.
Higher permeability cores can be used for lower end of the HF spectrum and lower permeability cores can be used for the higher end of the HF spectrum.
But then, the magnitude of power handling becomes an important factor too.
Higher permeability cores would get saturated quickly when operated on the lower end of the HF bands; so, you have to select bigger, bulkier cores.
For the higher end of the HF spectrum you can use lower permeability cores and with lower number of turns; you get the required inductance and at the same time the inter-winding capacitance is also taken care of.
De VU2RZA
Any thoughts on what to use for 2m and 70cm bands?
@@steveshelton5884 yes, sure. You can use lower permeability cores like #61,#52 or "K" ferrites.
Ferrite mix #61 is usually my choice.
It has a high Curie temperature, lower saturation and hence can withstand higher flux, and the likes.
I am using it in my 1.2 KW SSPA (HF) (using 4×SD2933 MOSFET devices) for the broadband output transformer.
And I have two such modules using hybrid splitter and combiner to put out 2.4 KW without any problem.
The cores remain as cool as cucumber.
De VU2RZA
Thanks for the detailed information SR, much appreciated. I will continue these tests with other core mixes 👍
Hi Ape, really enjoyed the video. It's great to see things outside of the box. I have not tested my 9 to 1's for loss, but have found that a 9-1 made with a FT-240-43 has better SWR in the low band, 20-160m and one made with a T-200 powder core has better SWR in the higher bands, 6-20. Twenty meters appears to be the point that both are somewhat close in SWR. I've finally broke down and bought a VNA, so I will be doing more testing in the future. My unun's are all 9 wraps. Have not tried anything different in that regard. I always enjoy you technical videos. Keep up the good work.
Hey John. The Ferrites generally are more broad-banded than the powdered iron cores. When you want a do it all solution you do end up making some compromises. Congrats on getting the VNA, it’s going to open up a new world for you 👍
@TheSmokinApe 100%. If I had to choose, it would be the 43 mix. I made one of each I can take into the field. That way, depending on the band, it gives me options. I know, a little anal, but I love making antennas and unun's. In the end, not a huge difference either way.
Very good explanation of a trifiler . Thanks.
Thanks Marvin, I appreciate you checking it out 👍
Spot on Ape. Trying to improve those losses is pointless as to what you would gain. I honestly thought the loss would be more. I have those 2 little cores here set up to do that test. I think there will be a difference with materials and geometries but not a massive amount. 61 material may be a good shout for 40-10 as Evil Lair has demonstrated.
Hey Colin! Your work has inspired me, thank you. I was pretty surprised at these results as well, I’ll be repeating with mix 61 in the future 👍
I was using this method to test a 1:49 toroid wind and was scrambling to find just a few resistors to out in series for a load…when it hit me that I had an inexpensive resistor decade box. Dialed the 2,450 ohms in and tested. I think I paid < $20 on Amazon.
Hey Frank, that’s a great idea 💡
Great to see the differences in performance - but there's another test you could do. Its not a test the NanoVNA can do, however.
Using an actual transmitter, vary the power level in and compare to the power level out. At some point, the core will saturate and you start losing power as heat.
You should probably do this with T140 AND T240 cores, to help users decide when a bigger core is a good idea.
Hey Alan, thanks for the info. I do plan on doing more video with various cores 👍
@@TheSmokinApe some baluns use 2 stacked cores. Need to cover that, too.
nice Ape explains why everyone says 9:1 :)
btw to get 450 ohm, try a 270 and 180 in series, or if you have an 820 and a 1k, use them in parallel (but thats closer to 451ohms)
That’s a great tip on the resistance, thanks 👍
Very good job, thank you.
Thanks, glad you liked it 👍
Thank you Mr Ape
Thanks for watching LC 👍
140 size toroids are good for about 25 watts. I use 240-43 for 100 watts and mostly 2 stacked together.
Thanks for the info 👍
Nicely done Ape! I look forward to the "different mixes" comparisons as well Sir!
73 - KF6IF
Thanks Phil, much appreciated 👍
Man this is good stuff
Thanks Intellistan, glad you liked it 👍
And that's a wrap! Thanks for showing us how it's done, Ape.
Thanks for checking it out Mike 👍
Maybe throw a 49:1 in there maybe?? that is some very interesting data. Great video! If I still had the gear to test this myself I"d try it in a second. It would be very interesting to know if I'm using the wrong unun!
Hey Steven, I’m planning to do a similar video just for 49:1 👍
That's great. Did you use a standard or non-inducting resistor? If it was standard, wonder when it's important to use a non-inducting style?
I’m going to say it’s a standard but I’m not 100%
Why do you use 2 ununs for insertion loss test?
Couldn't you use one?
Love the video gotta get me a nano vna....
One is to convert the signal from 50 to 450 ohm and the second is to bring it back from 450 to 50, the systems impedance of the meter (NanoVNA) is 50 ohms.
So if this was a 1:1 you would use one core if I'm understanding correctly.
Chameleon now makes several UnUns, 9:1, 6:1, 4:1, 1:1, selling for 115 bucks. Wondering if you will be testing the 9:1?
Hey Blackheat, I doubt I will test theirs but you never know…. Thanks for watching 👍
I only could fit 8 turns on my core. It works so far. I’ll have to test it out in rig expert. Wonder if odd turns vs even turns matters?
It doesn’t, you should be fine with eight, looking forward to hearing how it tests out 👍
Thanks Ape. Which one is on the CaHR antennas? BTW, when are you gonna break out the LibreVNA?
Howdy Don, the Ares 9:1 comes with a 140-43 and we suggest a 9x3 wrapping but it’s sold as a kit to make your own. I’ll have to remember how to use the Libre 👀
@@TheSmokinApe Damn son, that Libre was some serious cash. You should do more videos with it.
Where is this explained in the Rothammel’s book? I got lost in there.
Love me some toe-roids! :) Also, more uses for my NanoVNA... I like it.
Hey Robert, thanks for checking it out 👍
will stacking 2 ft140-43 handle more power?
yes
Would 11 or 12 wraps make it moar better?
I’m not sure it looks like 7-9 is probably the sweet spot for this mix 👍
Now I know a 7:1 is better on higher bands, who knew? Lol
Thats alot of work for one video, especially a blind Ape.😂 Go get some spectacles Son.🍻🤓👍
I do need magnification of some sort 👍
More cat videos!
I’m working on it 👍
tri--filer which means three filers 😀
@@hazer72 right