Transmission Line Transformer Testing

Back to back 16:1 Guanella TLTs

Jerry Sevick, W2FMI (SK) called this the “Soak Test.” He used it to run real power through the TLT network to check efficiency. It’s a pretty simple test whereby you simply run significant power through the TLTs and measure the temperature rise in the cores. If they get hot, they are not as efficient as they could be.

I’ve found that the soak test can be used for efficiency testing, and it also can be used for general TLT network design and test, as well as compensation capacitor testing.

In the photo above, we have two back to back 16:1 Guanella TLTs (parallel inputs, series outputs, no bootstrap). The 16:1 is realized with two back to back 4:1 Guanella TLTs.

The low impedance sides are connected in the center, so you now have 50 ohms on each end. In this arrangement, you can attach one side to a 50 ohm dummy load (the Bird) and the other side to a transmitter or VNA, or a good ole Antenna Analyzer as shown.

Both 16:1 TLT networks need to be identical and and changes (tests) done to either side needs to be replicated on the other.

You can quickly check all frequencies from 0-60Mhz and see how the TLT network behaves.

Some say if you use a low power source, like a Antenna Analyzer, you then also need to test this with at least 10W of real power. I figured this might be true due to getting the necessary level of flux in the cores at the 1Mhz end, but I see zero difference, even all the way up to and beyond 1500W.

The SWR measurements are exactly the same (1500W vs Antenna Analyzer), to the tenth!

For those adventurous designers that want to do more than simply duplicate existing designs, or maybe even find a better TLT design, I’ve found this setup invaluable.

It’s really the only way to test both TLT network designs and compensation capacitor schemes…without zillion dollar equipment.

BTW, here’s Sevick’s soak test description.

Updated comment from 12/10/22

Its hard to exactly duplicate how TLTs behave in-circuit. I think thats because of the variable capacitances seen at the LDMOS, its basically a varactor. Think of that as a compensation cap whose value is always changing.
Also, in-circuit, TLTs are typically feeding a BALUN, then a complex impedance (antenna or LPF). Not a resistor. But regardless, TLTs work awesome and when happy, and can transform impedances at great power while staying ice cold.

Comments

7 responses to “Transmission Line Transformer Testing”

  1. Kjell Karlsen, LA2NI Avatar

    Hi Rob.

    I have done this test in another way than you. Instead of using 2 pairs of the tandem connected 1:4 TLT´s in series, I used the pallets PCB as a test bead and installed SMD resistors and capacitors along the Drain Pads of the PA transistors to emulate the capacitance and Load resistance of the PA transistors. Then I installed the two 1:4 TLT´s and measured from the RF output using my VNWA.
    The TLT´s I use are wound with low impedance coax (6,25 Ohm) for the one transforming from 3.125 Ohm to 12.5 Ohm and 25 Ohm coax fort the other.

    The best result I got after the final compensating is amazingly good! On 1.5 MHz the RL is 22dB (SWR=1.17). On 14 MHz, the RL is 28dB (SWR=1.08). On 50 MHz RL is 33.5 dB (SWR=1.04) and even on 72 MHz the RL is 25.5 dB (SWR=1.11)

    But if I try to use the found compensating values “in real world”, the results are very disappointing! So the dynamic parameters of the LDMOS transistors are so different from the static values found in the data sheet that they can not be used at all. I have tried to start without any compensating capacitors and add capacitors one by one (or two by two) but still it is very difficult to get full power out on frequencies above 30 MHz.

    I am going to try to use a transformer similar to the one that ICOM, Yaesu and nearly all the other manufacturers use in the 100W PA´s with RD100HHF1. I am using it in Munin II and Munin III and the power output is more than 100W on all bands even 50 MHz. The transformer is wound using 4 pieces of low impedance coax were the primary is the braid connected in parallel and the centre is in series for the secondary.

    If you want I can send you recordings of my measurements and also photos of the setup.

    73, Kjell

    1. N4GA - Rob Avatar
      N4GA – Rob

      Kjell,

      Thank you for the comment. I also tried this with a 3.1 ohm resistor for the low impedance side. I did not add a capacitor because the LDMOS drain capacitance value varies greatly with both power, frequency and load.

      Like you, I obtained extremely good results with this simulated source impedance (resistor) and then this did NOT translate to the real world. It’s too bad because the testing would be quite simple if it worked!

      Yes, at 100W you can use all sorts of TLTs and or regular magnetic flux transformers. You can use the tried and true 9:1 binocular transformer.

      It’s at higher power and ratios of 16:1 where you really need to use TLTs for efficiency and bandwidth concerns.

      Here’s my test setup (that does not work) using the resistor. 🙂

  2. Pali HA7HQ Avatar

    Here, thank you for all the shared information!
    Great help in building.
    73!

    1. N4GA - Rob Avatar
      N4GA – Rob

      My pleasure!

  3. Donald R Solberg Avatar
    Donald R Solberg

    Hi Rob,

    I just finished testing seven different output transformer designs. When my goal is to compare one transformer to another, and not to attempt to determine what capacitance is needed in the actual amplifier for leakage inductance compensation, I simply put a resistor that is equivalent to the MOSFET output impedance across the drain tabs. I then feed the 50 ohm output with my single port VNA and perform panoramic scans.

    I know this is not entirely accurate but over the years I have found that this method will confirm that I wound the transformer correctly and lets me compare one transformer to another.

    I started out the test with tube and sleeve transformers wound simply with teflon wire in 9:1 and 16:1, I expected poor performance and that was what I got. Very poor matching bandwidth.

    I then tested two different cores and both 10 ohm and 50 ohm coax. These were wired as a “modified tube and sleeve” as described by Helge Granberg in his engineering bulletins.

    The best performance was with #61 mix cores and HC-12 (10 ohm coax), with the coax braids soldered to the brass tubes.

    I also built a TLT similar to yours. This was two 4:1 TLT’s back to back. By far this was the best performing output transformer. With that said, I am still going to use the modified tube and sleeve transformer as the PCB that I plan on using was made for this type of transformer and I already have the copper heat spreader drilled and tapped. I will save the TLT for another project.

    In looking at the various commercial legal limit amps: Elecraft uses the (2) TLT’s, Expert appears to use a modified tube and sleeve (I have only been able to see one photo of the inside of their amp), and RF KIT is using the tube and sleeve with teflon wire.

    If I was starting from scratch, I would do what you have done and use the TLT.

    73,

    Don
    K9AQ

    1. N4GA - Rob Avatar
      N4GA – Rob

      Don – good set of experiments. I’ve found that attempting to substitute a 3 ohm resistor for the LDMOS and look backwards through the TLTs with a VNA does not work for me.

      I’ve designed dead flat 1:1 SWR TLTs with compensation caps using this method and they did not work at all in a real amp. Others have had the same experience. I think it has to do with the output capacitance of the LDMOS as it changes widely across different Po and frequencies.

      I wish it worked then this would be quite simple!

    2. Timothy S. Kraus Avatar
      Timothy S. Kraus

      Don,

      My present transformer is just as you described, plain wire wrapped four times through the #61 dual copper tube core. I guess that’s why I don’t see 2,500 watts. I see just 1,400 watts. No reflected power. I probably just have poor coupling.

      My intent is brass tubing, #61 core and TC-12 coaxial. Just waiting for TC-12 from Communication Concepts in Ohio.

      Thanks for listening.

      Kraus/KC4ZGP

Leave a Reply

Your email address will not be published. Required fields are marked *