The AM Forum

There are excellent papers, books, and commnetary on ferrite transformers. However, for some reason, I could never find details of construction accompanied by actual measurements. The APP notes from Mot and Sevicks book, all great, but still need more detail. So, find attached some... 2 slides that are .jpg with pixs and the measured results. I hope to build a small library as time permits of others. Please, if you have similar documentation, it would be great to see. On the 16:1 BB xmfr on slide 2, it would be neat to see how well balanced this thing really is. We all know they are used in push pull output of SS amps. However it would be nice to see their frequency response. The same goes for the simple 2:1 xmfr and the n:1 turns ratio winding on powdered iron or ferrite. How broad in bandwidth?? I would love to see an ideal n:1 xmfr that spans 1 MHz-to- 1GHz :) The 4:1 uses 25 ohm coax, however, you could use two 50 ohm lines in parallel, does it compare as well? And so on...

Please post more of this material. It's very enlightening and interesting!

Transmission line transformers do a much better job. I modified an mri amp that works 160M through 6M with type 61 sleeves in the output.

When I build wideband binocular core transformers I use enamel wire in Teflon sleeving for both windings. It minimizes the coupling C, maximizes primary to secondary isolation, and broadens the response.

For 2 wire Beverages a transformer is a big improvement over the transmission line design when isolation is more important than efficiencey. In an amplifier circuit the opposite may be the prime goal.

Carl

Good stuff! Thanks for sharing Alan.

Thank you and you are all welcome. As I build others for whatever reason, I'll post. A brief explanation of the measurement(s) is in order. The network analyzer is calibrated for ONE port measurements using a CALIBRATION kit appropriate for the SMA connector used in the measurement. Of course, at the low HF frequencies, this is not a big deal. However if we want to see how the xmfr deviates from ideal, calibration is essential. The measurements are conducted with the xmfr terminated in 50 ohms and then we measure the single port Z at the opposite end. So... 3.12 ohms would be the desired response and with NO REACTANCE. The "classic" 16:1 popularized by many functions amazingly well! The reactance that is present may be compensated by adding C and this was well documented by Philips and Amperex (and others), but again give me some details. Like how bad was that xmfr prior to compensation. As you can see, the 16:1 functions very well. The single turn is obtained by the copper tape or tube (use of copper pipe is popular) and the 4 turns are teflon wire threaded from the opposite side. This xmfr therefore provides ~ 3 ohm termination (do I get 1.5 ohm each side of CT??) and so for a single device operating at 12 V, should support ~ Vcc^2/2RL or 24 W and 48 W push-pull.

Carls response is interesting and if indeed the case, it would be great to see the frequency response. There are many tricks to the trade associated with the design and construction of these xmfr's.... however, nothing beats actually building one and then running off to the measurment test set to see what it actually does. If the unit is just dopped into a final circuit and the circuit works fine.... then it begs the question.... Could I have used something simplier?

Carl,
I like to use teflon covered wire wrap wire for transformers. This raises the impedance though and may require 4 conductors(parallel pairs) to get the impedance lower if you are working around 50 ohms.
Twisted enamal with about 5 twists per inch is around 100 ohms if I remamber.

Im not using transmission line transformers Frank since Im only interested in bandwidth and isolation between the Beverage in either antenna or transmission line mode and the 75 Ohm output. A little extra loss isnt of any consequence as it would often be in a power amp.

Here is a wideband TLT (transmission line transfromer) that has an additional connection for an equal delay. The test case for this was a 4:1 xmfr however the connection for the 4:1 is made by a second equal length of tline which provides compensation (for the phase delay). The resulting impedance measure vs. frequency is quite good, transforming 50 ohms to 12.5 ohms over a wideband, 20 MHz-1.2 GHz. See the 3-slides attached. The origin of this technique dates back to the late 40's, with Guanella and his work. It would be interesting to see what step UP could be accomplished. For the case here, the step down was such that 25 ohm line was used. The perm of the material was 125 and the whole affair is but 1.5 inches. If anyone has worked with these EQUAL DELAY BB XMFRs, great! Please comment. These have obvious applications in SS amps.

2 goes to 4 and 4 goes to 2 in the Erb MRI amps for equal delay and balanced on the low Z side facing the FET drains. Series DC blocking caps on the High Z side.

Thanks! Could you please clarify? Perhaps a sketch. Are you saying that the output points of the two lines are in parallel? 

I apologize, I see what you are saying and yes that would work as well! In fact, small coincidence, I was looking at an old CTC solid state device amplifier notebook, and they highlight the exact circuit you mention. They also claim in the same text, (C) 1973, that they originated the name and could not identify the origin  of the original concept!

I ran into the Erb amp designer who said they started with Motorola app notes and improved on them. I think Mot bought CTC at some point. I've built a few CCI kits and found their transformer design lacking below about 3 MHz.
Temperature rise of the transformers a bit much.