2.4 GHz matching transformer - who makes it?

I need an impedance matching transformer that handles 1 W at 2.4 GHz (2400 MHz). It should be a self-contained unit with connectors such as BNC (or N or F or other coaxial). I’m matching 50 ohm to 75 ohm systems.

Who makes such a thing?


<Sunspace digs through his mental componentry junkyard>

What’s the application? Are you goung balanced to unbalanced? My impression is that at GHz frequencies you can pretty much put together what inductances and capacitances you need from regular wire, and enclose it all in a shielded box.

Calculating what you need, on the other hand, might be interesting. I haven’t done that kind of thing for a long time, so maybe an RF hardware type with fresher experience will jump into this thread.

I’m going unbalanced to unbalanced. How can you have balanced going through coax?

The application is sending an 802.11b signal down a long CATV cable. Say what? Lemme explain. I use satellite broadband internet access because it’s the only thing available here, but my house can’t see the southern sky. My barn, however, can. Years ago I buried two CATV 75 ohm cables under what’s now a landscaped yard, between the house and the barn, because I also have satellite TV service. There aren’t enough cables down there to support the satellite internet dish, but I could sacrifice one of the TV cables to let me extend my home network out to the barn, so I bought two Linksys WAP11 wireless access points that can also act as a wireless bridge (ethernet to 802.11b to ethernet). And rather than actually letting it be wireless, I connect their antenna jacks to either end of the 75 ohm cable. So nothing is literally wireless.

Trouble is, this barely works. As the weather got warmer, I guess the cable got lossier. It’s already operating at 3x or so the highest frequency I can find anybody says you’re supposed to use it for. Well, I can afford alot of loss. Antenna to antenna transmission typically loses dozens of dB. But I’m near the edge and often suffer slowdowns or interruptions until I fart around looking for some bit of cable I can improve somewhere at the two ends. I know the impedance mismatch costs me some attenuation, especially given that the cable is lossy (lossy cable plus mismatch is an especially bad combination).

So I’m trying to avoid other nastier alternatives, like another two day rental of a trenching machine with hundreds spent on cable and pipe and so forth. Or like putting up high gain antennae (there are no really convenient places to put them or relocate the WAP11’s).

I’ve thought about winding something but have never built radio equipment in GHz territory and haven’t found much info about it. Buying books etc - jeez, just another project in a direction I’m not interested in exploring - sure hope I can find a less time intensive answer here.

There seems to be a lot of this stuff out there, unless I’m misunderstanding you.

Here’s some links:

A Goodle search on “Impedance Matching Pad” turns up quite a few results, you might find one close by

I need a transformer in the sense of something with windings, or a length of coax with tapered conductors, or something similar that essentially conserves signal energy while transforming the impedance.

Matching pads are resistor networks that meet the requirement of different impedances at the input and output, but only by throwing away much of the signal. That is, they are impedance transformers and attenuators at the same time.

Since my problem is a weak signal to begin with, the attenuation of a matching pad makes things worse.

I actually have matching pads like those cited - my system stops working immediately when I include them.

PS I have Googled “impedance match” and “matching transformer” and “microwave” and “gigahertz” and “50” and “75” and a bunch of other terms in too many combinations to remember. Likewise with the Alta Vista search engine. Typically I get hundreds of hits, none of which are what I want. And I actually did turn up one unit that would work, but with a 20 W power limit and a corresponding price tag hundreds higher than I hoped for. It’s probably cheaper than the new trench option, but not by much. If anybody could help, it’d be the Dopers, I figured…

Have you considered 10Base2 Ethernet with 50 to 75 ohm matching transformers to fit the 50 ohm connections on the Ethernet cards to the 75 ohm cable? Overall, the cost of the maching transformers and the 10Base2 equipment will probably be less than 2.4 GHz matching transformers (if they exist).

You would need 2 T-adapters, 2 terminators, 2 matching transformers, and a pair of 10Base2 adapters of some kind, like Ethernet cards with 10Base2 connectors or hubs with 10Base2 ports.

What about building a couple of microwave-sized yagi antennas, and point them at each other? With some fiddling, you could get the feedpoint impedances to 50 Ohms. I’ve made a couple of UHF yagis using coathanger wire through a dowel rod for just the cost of the dowel rod. A microwave-band yagi would require even less material. There are also slinky beam antenna plans out there as well.


If your system is barely adequate now, it seems doubtful that simply matching the cable impedances will help all that much. Your loss from mismatched impedances is only about 2 db with the present setup. A matching transformer will probably have about 0.5 db insertion loss leaving you with a net of 1.5. If your voltage at present is 1 your new voltage would be 1.19 at most.

It looks to me tlike an amplifier with 10 db gain would be better. I have no idea if they are available either which isn’t a lot of help.

>Have you considered 10Base2 Ethernet with 50 to 75 ohm matching transformers to fit the 50 ohm connections on the Ethernet cards to the 75 ohm cable?

Wanted to do it that way first, but couldn’t find 10Base2 ethernet converters - the box between my Cat5e twisted pairs and the coax. That’s why I got the WAP’s. Can you point me to them?

>What about building a couple of microwave-sized yagi antennas, and point them at each other?
You mean, making it a real wireless bridge? That’s been an option from the start, using store bought antennas (they’re not so expensive).
But like I said, there are no really convenient places to put them. I might still go that route, but not without trying first.
>Your loss from mismatched impedances is only about 2 db with the present setup.
Wait, is that true? I understood there were two sources of loss, one from the reflection of signal from the point where the impedance changes, which would be like optical reflection from a surface where index of refraction changes. This would be the principle by which time domain reflectometry works. Isn’t this the 2 dB you are calculating? But I am probably troubled by the other loss, which I assumed was far the worse, namely that from the large voltages and currents that have to be circulating inside the 75 ohm portion to satisfy the voltage and current boundary constraints posed by the 50 ohm ends. You can ignore those circulating currents in small things like an antenna because ohmic and dielectric losses aren’t big (unless the voltages cause arcs or the currents melt something), but in a system that has significant ohmic and dielectric losses, they soak up enormous energy. This would be the mechanism for an automobile muffler making the exhaust so quiet, really an impressive feat of energy consumption. I thought that was the main problem I had to deal with.

Are you sure you haven’t toasted something? Receivers aren’t usually made to have transmitters coupled straight in to them. I think I’d see if the two units can still communicate with antennas before I go looking for another solution.

Irrespective of all else, the power to the load is I[sup]2[/sup]R[sub]L[/sub]. The current is the open circuit voltage divided by the total circuit resistance.

I = E/(R[sub]L[/sub] + R[sub]O[/sub])

In the matched case this is E/100 and in the mismatched E/125. The ratio of powers is (125/100)[sup]2[/sup] and this comes out to be 1.938 db.

It is possible that your equipment might be adversely affected by a large standing wave ratio on the line, but that is a different matter. It has been a loooong time since I had any truck with standing wave ratio but I could look it up. In any case, I don’t think your mismatch results in an excessively high SWR but that’s only a WAG.

And I just noticed that you say you are driving the line with a 50 ohm source. This changes the picture considerably and it gets to be time for Smith Charts, or the computer equivalent, because now we have no idea what load is being presented to the 50 ohm source or what the effective source impedance is as measured at the load.

I assumed a 75 ohm driver and a 75 ohm line but that isn’t what you have. Sorry about that.

OK, this is a little nuts, but it wouldn’t cost anything but your time to try.

You can eliminate reflections using a quarter wave transformer, which is a quarter wavelength section of transmission line with impedance sqrt(75*50) = 61 ohms.

The coaxial characteristic impedance is

Z0 = ln(b/a) * 376.73 / (2 * pi * sqrt(epsilon))

where b is the outer conductor diameter (measured from the inside surface), a is the inner conductor diameter, and epsilon is the relative dielectric constant of the filler inside the coax.

For the coaxial line I’ve worked with, the filler is foam, and squishy. If you take off the covering, you could squish the outer conductor, making the characteristic impedance smaller. If you make b smaller by a fraction S, the new characteristic impedance will be

Z1 = ln(b*S/a) * 377 / (2 * pi * sqrt(epsilonPrime))

where epsilonPrime is larger because the foam filler has been compressed. Roughly,

epsilonPrime = (epsilon-1) * (b^2 - a^2) / (b^2 * S^2 - a^2) + 1

You can work out the right amount of squish, S, to make Z1 be about 61 ohms. If you can squish a section of length 1.23 / sqrt(epsilonPrime) inches at the end of your 75 ohm line, then connect this to your 50 ohm line, this would knock down the impedance mismatch reflection.

I warned you it was a little nuts… :wink:

Yo, ZenBeam, that IS pleasantly nuts. You’re my kind of guy.

I wonder if there is 61 ohm line for sale anywhere. Of course, I’m the one who can’t find transformers for sale…

Actually I was toying with the idea of making my own coax by winding dielectric tape on a wire, slowly changing my wrap angle to taper the insulation diameter. Then I’d put foil and a braid over it to make my own transformer. You’re not very far from that idea. But, admittedly, your suggestion doesn’t have the problem of putting connectors on nonstandard OD cables.

Maybe I can get foam insulated 75 ohm cable and soak oil into the foam a little. Oh, God, I only wanted broadband… but all I ever use it for is to figure out how to keep it working…

>It is possible that your equipment might be adversely affected by a large standing wave ratio on the line…

That’s what I think I mean when I talk about the other, worse source of loss. “Standing wave”, that’s it. Maybe I have to crawl into bed with a Smith chart, which I never figured out how to use.

>Are you sure you haven’t toasted something? Receivers aren’t usually made to have transmitters coupled straight in to them. I think I’d see if the two units can still communicate with antennas before I go looking for another solution.

Oh, absolutely - I keep bringing the two back together and putting antennas into the jacks and verifying they still seem to work in the same fashion. And every time I try another cable idea, I put two 15 dB attenuators into the line and try it, then remove one and try again, then remove both and try again (I’d stop removing them if it started working). Whenever I think I might have toasted things I scream like three girls and then run around retesting them with antennas, but always they are OK. It makes me guess that manufacturers figured out they couldn’t assume much attenuation because occasionally users would position them with antennas almost touching, or indeed touching - so whatever huge AGC range they design, they have 0 dB attenuation included at the top of it. Whatever the case is, I know that all my versions have been within 15 dB of the bottom of the range, because none of them have worked with even one of the attenuators in the line (though this may not be true if I have a big standing wave loss that makes the supposedly 15 dB attenuator do more than that).
Christ on crutches, I must have the hardest house in the world to bring broadband to…

Well, you dragged me in. I got a Smith chart off the net and worked the damned thing out.

Your mismatch loss varies from 0 db if the line is an exact number of 1/2 wavelengths long to a maximum of 4.2 db.

One way you can try to match by trial and error is to adding little bits of 75 Ohm cable. If you can make the line close to an exact number of 1/2 wavelengths long the output impedance of the line will be equal to the source impedance.

If the load is connected directly to the 50 ohm source there is a match. On the Smith chart if you then move in a 50 ohm circle around the center point of a 75 ohm chart toward the load the source impedance is transformed into a resitive-capacitive impedance through the various line lengths and to a 75 ohm resistance at 1/4 wave down the line. Continuing toward the load the impedance becomes resistive-inductive and arrives back at 50 ohms resistive at 1/2 wavelength down the line. And of course this cycle repeats every exact 1/2 wavelength.

This is the resonant line match. Of course, if the dielectric constant of the line changes with age the wavelength on the line will change and you’ll lose the match. However, if your line is going to change it should already have done so since you say it’s pretty old.

Whops, I can’t even read my own Smith chart. The output impedance at 1/4 wavelength down the line from the source is 112.5 ohms, not 75. That’s what give the max loss of 4.2 db.

Try eBay. A hub like this one is probably your best bet.

Or you could get these… or the used version…