Winding MHz 1:4 transformer

That's a wide frequency range, so I think you would use a "Transmission line transformer" ( good search term) and a high permeability core material.

Here's a starting point,

And a compilation of sites.

Mike

--
http://www.google.com/search?sourceid=navclient&ie=UTF-8&rls=GFRC,GFRC:2006-50,GFRC:en&q=winding+broadband+RF+transformers

JF

Impedance is important.

If you need broadband at an impedance much over 200 ohms you're going to have a hell of a time with stray capacitance. Getting it to work over more than two decades of frequency will be a challenge too.

If you're going to do much work in that frequency band I strongly suggest you get yourself a copy of the "Radio Amateurs Handbook" from the ARRL.

--
Tim Wescott
Control systems and communications consulting
http://www.wescottdesign.com

Need to learn how to apply control theory in your embedded system?
"Applied Control Theory for Embedded Systems" by Tim Wescott
Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html

Tell us the impedance levels this thing needs to work at. Is your ratio a voltage or an impedance ratio? It makes a difference! The latter is the square of the former. And what will be the power level?

Jeroen Belleman

4:1 impedance or turns ratio? What impedance levels? What power level? What degree of isolation do you need between the windings? (Can they be bifilar, or do they need to be wound to have low capacitance between them?)

Though not a perfect source of help, John Sevick's "Transmission Line Transformers" is a reasonable place to start for bifilar-wound RF transformers. You may find some helpful info on MiniCircuit's web site, since they supply transformers of the sort you are interested in. See for example

may also be interesting. Do Google (or similar) web searches for "RF transformer" and "transmission line transformer" and you'll get LOTS of hits, some of them even helpful.

A Spice model will quickly tell you that to get wide frequency response, you need tight magnetic coupling between the windings. Bifilar windings on appropriate ferrite cores give you that. MiniCircuits T4-6T, for example, will get you 0.1-100MHz at -1dB points. Put in a bit of leakage inductance (imperfect coupling) and the top end drops rapidly.

Cheers, Tom

Thanks to everyone for the replies.

The input voltage is around 10Vpp. I am looking for four times that on the secondary at 10 milliamps with 50-100 Ohms output impedance. I am not aware of any problem with using a bifilar winding as suggested below.

The 100KHz to 20MHz frequency range given in my OP was the ideal. If not practical, I could cut the bandwidth back or use two transformers.

Does anyone have any suggestion on a circuit for a good amplifier to drive the primary? Power op amp, video amp?

Mike Kendall

Look up a balun. Balanced to unbalanced. That automatically gives a 1:4 impedance ratio. The trick is to treat the windings as transmission lines. You can buy really nice bonded magnet wire that is already at 50 ohms. The LF response is determined by the inductance and the HF response by the coupled transmission line.

Steve.

--
Steven D. Swift, novatech@eskimo.com, http://www.novatech-instr.com
NOVATECH INSTRUMENTS, INC.      P.O. Box 55997
206.301.8986, fax 206.363.4367  Seattle, Washington 98155 USA

(snip)

40V @ 10mA is 4k-ohm, _not_ 50-100.

--
Tim Wescott
Control systems and communications consulting
http://www.wescottdesign.com

Need to learn how to apply control theory in your embedded system?
"Applied Control Theory for Embedded Systems" by Tim Wescott
Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html

As Tim noted, 40Vp-p or 14Vrms if it's a sinewave, isn't 50-100 ohms load if it's also 10mA in the load. Saying it another way, at 100 ohms, 14Vrms is 140mA. It's also about 2 watts. I'd suggest you just use one of the minicircuits transformers, but 2 watts is a little much for them, from my experience. But it really shouldn't be difficult to wind a little transformer that will handle that power level nicely, and cover the desired frequency range. You'd use 5 strands of magnet wire, twisted together (unless you can run it as an autotransformer and share one winding between the primary and secondary). 4:1 turns ratio is 16:1 impedance ratio, so you would be driving the primary as a 6.25 ohm load, assuming you load the secondary with 100 ohms. That's pretty low for a video amp, and a power op amp probably won't do a good job at the high frequency end. What sort of signals are you wanting to drive into it? And--let's get the impedance levels nailed down before going much further!

Cheers, Tom

Thank you, Tom, for your clear explanation. Upon reconsideration, I would most likely be driving the primary with a function generator. So the impedance would be 50 Ohms. The signal is a 100KHz to 20MHz sinewave AM with an audio sinewave.

Here's the tricky part. According to the characteristics of the load, the impedance seen by the secondary side of the step-up transformer varies in proportion to the frequency. This would be between 10K Ohms and 100K Ohms over the above-specified range.

Finding how to accommodate this in practical terms is part of the purpose of the experiment. If there is no way to make the system adjustable, I may just have to choose a median figure and live with it

What would be the limitations arising from that? What do you think is the most logical approach?

Mike Kendall

Ah, OK, that helps. Thanks.

Since you want to start with a 50 ohm generator and use a 4:1 turns ratio, to step the voltage up, that dictates that looking back into the secondary you'll see a (4^2)*50 ohm = 16*50 ohm = 800 ohm source impedance. At audio, you could get by with loading the secondary with

800 ohms or more -- or even somewhat less. But at RF, you're liable to have a rather non-flat frequency response if you don't load the secondary with reasonably close to 800 ohms. The good news is, it's easy. Just put a 1000 ohm (or 910 ohm would be even better) resistor in parallel with your load. Then the net load on the transformer will be about right over the whole frequency range.

I guess we're still talking 10V p-p into the primary, 40V p-p at the secondary, nominal. 10V p-p at 50 ohms is 1/4 watt, or +24dBm. Some signal generators will be able to output that directly: the HP3325 (or the much newer Agilent 33250) should be able to, I believe. Most generators that go to higher frequencies will have some trouble going that high. Most video amplifiers aren't designed to deliver so high a voltage, since standard video runs much lower than that. Modern high frequency op amps mostly run at power supply voltages that are too low.

But note, too, that most of the power is dissipated in the 1000 ohm resistor. If you wind a transformer with a higher turns ratio, you can get by with less power. The price you'll pay is that it will be harder to get reasonably flat frequency response over so wide a range. But it may be worth a try. For example, if you managed a 7:1 turns ratio, that's 49:1 impedance ratio, 50:2450 ohms. Then you only need about 6 volts p-p on the primary, and a little less than +20dBm to excite it.

If you have a generator that will go to 10Vp-p into 50 ohms, just stay with the 4:1. It will be a lot easier to wind a transformer to do what you want. The fixed load in parallel with your variable one should let it work well over the range you want. Oh, and at that power level, you could use one of the MiniCircuits transformers. Their T16-6T is spec'd for max 1dB loss over just the frequency range you want (and I can just about guarantee they know how to wind them better than you'll be able to do without a lot of practice and trial- and-error).

One more comment: if you wind your own, I'd suggest you shoot for a primary inductance such that the inductive reactance at the lowest frequency of interest is roughly the same as the generator source impedance: in other words, 50 ohms = Xl = 2*pi*f*L, so L = 80uH. In fact, if you drive it with a generator with lower source impedance (e.g. with an op amp through a low value resistor, say 10 ohms), you can get a wider bandwidth out of a transformer, as compared with driving it with a 50 ohm source. With 50 ohm source, 800 ohm load,

4:1 turns ratio, and 80uH primary inductance, you'll need a coefficient of coupling of about .9975 to get a bandwidth of 100kHz to 20MHz at 1dB loss.

Cheers, Tom

Thank you, Steve. I hadn't heard of that stuff before. Looks very handy for homebrew toroids. I googled and ordered some samples from the first supplier I checked -- MVS.

Now to find someone who sells it in 1/4 pound rolls at price compatible with a hobbyist's budget! :-)

Tom

You may be able to snag some on ebay. It's worth a try, anyway. I've gotten Litz wire there at a price far below what I'd have to pay buying it through distribution. But you can also just make your own twisted pair. The impedance won't be so closely controlled, but if you're winding on high-mu core material you're probably using far less than 1/4 wavelength of wire anyway, plus the core will change the impedance of the twisted-pair as a differential line, since the line is no longer in an all-air environment.

Was that MVS or MWS?

Cheers, Tom

Thanks, Tom. Yep, eBay was the second place I looked. O found some large gauge stuff, but nothing for tiny l'il toroids.

Yes, I've twisted wires before, and from the looks of it, will continue to do so. :-)

Many thanks,

Tom