Waveguide Bend Tuning

[SteveFehr]

How do you tune swept waveguide bends to avoid standing waves when the frequency you're using differs significantly from the midband frequency of the waveguide? I've struggled with this question for months now- we certainly never hit on it in emag fields and waves in college, and no textbook I've seen addresses this directly. Worst of all, the official guidance put out for military waveguide (below) seems to contradict the physics as I understand it. This has been the standard reference for manufacturing military waveguide since the original 1963 document, and I can't believe it's wrong, but it certainly looks that way... Am I overanalyzing this, or have we been doing this wrong all these years?

The first question is- what wavelength should I be using for the calculations- the ?=c/f freespace wavelength (ala MIL-HDBK-660A) or the "waveguide wavelength" taking phase velocity into account, as seems to be a more appropriate course of action? I mean, that's how I'd do it for straight waveguide, I don't see why bending it would change this fundamental principal, yet there's the mil-hdbk right there saying to use ?=c/f...

The second question is what wavelength fractions should be tuned for- half wavelengths like MIL-HDBK-660A recommends, or not. Now, call me crazy, but it looks like tuning to ½? would create standing waves instead of canceling them out- wouldn't an odd ¼? or ¾? be the right wavelength to use. What's the right approach to take? And can we even use the simply mean length for these calculations or go more in depth analysis with the wavefront through the H-bend? I can't see it mattering much for an E-bend.

My final question is, does it really matter? Are my VSWR & losses in a 5? radius bend going to be negligible no matter how I design it? What about 1? or 2??


MIL-HDBK-660A:

http://combatindex.com/mil_docs/pdf/Hopper/MIL-HDBK/CI-660A-MH-4822-4932.pdf

 

[biff44]

Put a steel ball bearing inside, turn on the waveguide (lowpower only please), use a magnet to move the ball bearing around inside the guide, and when you have a place that the ball bearing tunes out the vswr, take a ball peen hammer and whack the waveguide to leave a dent. Repeat as needed.

 

[Higgler]

You could try a C clamp squeezing the bend for your "dent tuning". The waveguide will bounce back to nearly the same dimension if you don't squeeze too hard.
Make sure no dead bee's are inside, for each one you lose one dB I've been told.
kch

 

[SteveFehr]

Problem is, I'm an engineer... I'm doing the design in CAD and will probably never actually see the waveguide in person letalone have the opportunity to tune it. And I do *not* trust the pipefitters doing the installation- I know from experience they don't know the theory, they don't understand the tolerances and they plumb waveguide like they'd plumb a toilet. Worse, actually, as copper waveguide is so easy to coerce into position... So, I buy prefabricated pieces whenever possible, but for the custom fit stuff, I have to assume lousy quality bends and flanges but I can at least hope they can measure and cut to about 1/16" or so even if they can't cut to 1/64" like they're supposed to. Anything I can do design-time to dummy-proof the install will help.

Is there an optimal radius and bend length given a particular frequency or is it all something you've got to don your wizard hat and cape for and cackle wildly while dropping smoke bombs?

 

[biff44]

In my experience, it is the LIVE bees you have to watch out for!

 

[biff44]

One thing you can do to help with gorilla installation is to put in two alignment pins in the flange. It is easy for the company making the waveguide to position the two pins with relation to the centerline of the waveguide itself, and the line up will be much more precise than counting on the thru hole screws.

As far as VSWR, the bigger the bend radius, the smaller the VSWR--theory of small reflections helps you out.

 

[zappedagain]

SteveFehr,

While I'm a long wavelength guy (i.e. I'm moving up in frequency so I might be learning here), here-s my two cents on waveguides.

The 1/2 wavelength guideline seems to be semantics. I interpret their 'integral multiples of 1/2 wavelength' as meaning 'even multiples of 1/4 wavelength'. Avoiding quarter wavelength impedance transformations seems the goal in their guideline.

Paragraph 1.2.5 states that the concern is for two reflections - one at the far end of the bend and the next reflection at the source end of the bend. So traveling two paths (there and back) means you travel a full wavelength for minimum VSWR. If the length was a quarter wavelength multiple it would take four paths to get back in phase; with attenuation on each reflection that seems much more prone to VSWR.

Let me know if I'm all wet here - I love learning!