transmission line transformers on a board

It's always funny when something has been blurred out. They think they're hiding proprietary stuff, but really they're just hiding their embarassment at something that's "obvious to those skilled in the art". :-)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

I assume that my competitors will somehow get an early unit and open it up for inspection; I do that to theirs [1]. But there's no point in giving them a head start.

We do a few things that are not obvious, namely that nobody else appears to do. Some of my competitors make amazingly klunky stuff. I have a couple dozen circuits that I'd rather keep private as much as possible.

In SED, most people are secretive about everything. I share most of my design ideas, like this transmission line thing.

I am reminded of Phil's bootstrap cascode photodiode amp. He has given it away, but people keep making and selling stuff that could, with a modest effort, be 100x better.

[1] Sometimes I have to special-order the right cal seal stickers, so I can rent a box, peek inside, and replace the stickers so the rental company doesn't get upset. If a box has been around a while, I can usually buy one from a broker or on ebay.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Yahbut with twisted quad each wire intracts with two neighbours so the impedance is halved. and you get 25 insted of 50 (once you parallel the wires)

AB BA 25 ohms. hmm, maybe this would give 50? AA BB

--
This email has not been checked by half-arsed antivirus software

I mean, I did that a decade ago; I have a date of 2006-06-30 on the file. So I probably did the experiment a week before that?

I suppose it's nice to remind others of methods they might not be aware of.

IIRC, he hadn't invented it, but that happened much earier, having been found in expired patent literature, or some magazine articles or something. Or if nothing else, it's simply a good realization of using well understood circuit blocks and techniques -- what might be "obvious to those skilled in the art" (though patent examiners aren't very skilled, it would seem). None of which is to say Phil didn't spend real design time creating or optimizing it, or to say that it's a bad circuit and there are better; well, while there's always better, it's pretty darn good as is, no matter when it was created.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

A bit less, because the fields still overlap. Wait, more, because they overlap the wrong way? I forget.

Asymmetrical configurations like

AA B

can be handy, too. If you don't mind about asymmetry, which is fine in the middle of a transformer when the winding is short enough not to matter much.

I did that here:

The two pot cores are TLTs. One is configured for a CT choke, to improve balance. The other is a 1:2 balun to match drain impedance to 50 ohms.

This shows the windup, sort of. Note L2 is the balun, and is drawn showing every strand, where two happen to be in parallel to get the right impedance.

Bandwidth is okay for what it is, but it runs out of steam at full power (because of R10 and R23 being relatively weak; alas, the complement for Q2 is rather rare), and you can't expect much more from Q3/Q7 anyway.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

Tim, thanks for sharing.

Just musing: could R10 and R23 have been replaced with active load as in White follower?

piglet

Sure, though I wonder if the loop delay would be a problem. (PSRR might also suffer.) It could also be a phase-splitter-and-totem-pole configuration (something like a TTL output stage, biased for analog operation), which might save on delay, but also necessitates matched delays in the two paths again. And, again, all those transistors have to be fast, and biased for class A or AB operation.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

50 MHz is 6 meters. The windings in those transformers must be a few inches long. So those are mostly "real" transformers. The twists keep the leakage inductance down.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Good simple ideas always appear obvious in hindsight. That's one of the main problems in patent litigation.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

My suggestion wasn't a circuit topology, it was just a cheap and easy way to physically construct the transformer on a surface-mount board. I'd never seen that anywhere else. My actual circuit is obviously different from yours.

I don't know that he claimed that he was the original inventor. He shared something fairly simple that wildly improves a photodiode amp, and lots of people keep doing it wrong anyhow.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

see

MODELING OF TRANSMISSION LINE TRANSFORMER shows modeling for standard 4:1 impedance transformer. Considers unbalanced coil mode currents and balance d transmission line currents. The more inductance in the coil mode, the lo wer the unbalanced current and the more accurate will be the transformation . Winding on ferrite increases coil inductance. Best wideband operation i s with the length of the line shorter than 1/4 wavelength and the character istic impedance 2x the load. Also, ideally, a delay line should be used fo r the feedback connection to keep the phases correct. MICROWAVES & RF 11/9

6 P73-80.

I once made a 50 to 3.125 Ohm transmission line transformer for a 1kW broadband application that consisted of two cascaded

4:1 Guanella transformers. They were made with standard lengths of connectorized coax. A nice twist was that both transformers were wound on the same core. It looked a bit messy with all those cables, but it worked well. Losses were well below 0.5dB over most of the bandwidth.

Jeroen Belleman

To which my answer was always "If it was so obvious, why didn't you (or anyone else) come up with it?".

The problem is that the USPTO doesn't know enough art to know if something is new. We went around and around with the IP lawyers on a bunch of this stuff during the RDRAM debacle. The solution is simple but it would put lawyers out of business.