Connect with us

Bending rectangular copper waveguides

Discussion in 'Electronic Design' started by Rene Tschaggelar, Nov 14, 2008.

Scroll to continue with content
  1. During my work I came across several way to
    bend a rectangular waveguide. The copper is
    best softened before by a heat treatment.
    Then the choices are :
    - fill it with sand and heat while bending
    - fill it wit water, freeze it with LN2 and bend
    - fill it with Wood metal melting at 60 Celsius
    and bend at room temperature
    - fill with brass strips and bend at room
    temperature

    Short of trying all, what are the better choices?
    Yes, I'm aware there are rectangular cast bends
    to solder in. They are limited the rectangular
    though.

    Rene
     
  2. _

    _ Guest

    cross-posted to r.c.m. as the level of practical experience therein is
    likely to be high. r.c.m.'ers should note that for wave-guides maintaining
    the (x-section?) dimension following the bend may be quite important.
     
  3. Jim Wilkins

    Jim Wilkins Guest

    Avoid them if possible. Semirigid coax and SMA connectors work to at
    least 20 GHz.
     
  4. Guest

    HAVE YOU TRIED URI GELLER?
     
  5. - Hide quoted text -

    We use the WR28 at 35GHz and the attenuation
    in a coax is dramatic. 5dB for a single meter
    are common. Even if the plumbing appears crude
    and tiring, we prefer waveguide if possible.

    Rene
     
  6. Guest

    What are the dimensions and wall thickness?

    How long are the pieces you wish to bend, what's the angle, and what's
    the radius?

    Can your application withstand a few failed attempts (mismatch)
    without self-destructing?

    We use some flexible waveguide here at work (corrugated) for some low-
    power radar stuff. Is that an option?


    Dave
     
  7. Guest

  8. Guest

  9. Electrician's perspective: If it's a one-off (prototype or custom
    project) for a complex bend like an offset, or a J-bend...

    I'd consider taking sheet copper, cutting the four sides with all the
    flat curves needed for each face, cut the flanges, bend in all the
    required squiggles and turns with a forming roller, and then braze or
    TIG weld it together.

    You leave two opposing faces a bit wide so they overlap on the
    outside for easier assembly - make a fillet weld rather than acute on
    the outside corner - the weld penetration on an acute corner will mess
    up the profile of the inside corners of the waveguide.

    Or combine both - TIG to tack it together and make sure it fits,
    then silver-braze so you don't spend all day on sealing it all up -
    brazing temperatures don't reach the melting point of the parent
    metal, so they won't make the TIG tack welds come apart.

    And brazing metals can bridge gaps and fill grooves, almost like
    Bondo filler putty - just have to be delicate with the heat to stay in
    the plastic state.

    Another possibility is to section the tubing and make a segmented
    bend, expand one side slightly with a flanging tool to make a step
    overlap for mechanical strength, and then braze the pieces together.

    If it fits and works as expected /then/ you can tool up molds to
    cast them, or figure out another way to mass produce.

    --<< Bruce >>--
     
  10. tom koehler

    tom koehler Guest

    in any effort to bend a tube, the metal on the outside curve wants to
    stretch, the metal on the inside of the curve wants to compress. Any of the
    various things to fill a tube are attempting to keep the tube from
    collapsing. If the filling material is sufficiently non-compressible, the
    metal on the outside of the curve must stretch. If it is not ductile enough,
    it will tear.

    This is way out there, but could you make a dummy waveguide from solid
    plastic rod of the proper cross-section, and then plate a metal layer onto
    the plastic?

    tom koehler
     
  11. Ned Simmons

    Ned Simmons Guest

    I think the Wood's metal or brass strip methods would work best. The
    problem with sand is that it relies on the fact that the geometric
    shape with maximum area relative to its perimeter is a circle. If you
    pack a circular tube with sand and bend it, the cross section can't
    deviate from a circle because, assuming the circumference remains
    constant, any change would reduce the section's area and the sand is
    incompressible. A rectangular section can distort into a more circular
    shape while allowing the sand to become loose. I imagine the ice would
    behave somewhat like the sand as it fractures.

    The brass strips idea is clever, as long as you can pull the strips
    out after bending. It simulates the ID mandrels used for sharp bends
    in production.
    http://www.toolsforbending.com/mandrels.asp
     
  12. I think there's little doubt that using a solid metal
    filling is the best option. Wood's metal is expensive
    due to the indium content. That might not worry you,
    but if it does, there's another eutectic alloy I've
    been meaning to try; you should be able to make it
    yourself. Bismuth (perhaps buy bismuth shot?) with
    tin and lead in the right ratios will melt near 94 C,
    i.e. in boiling water. You'll have to look up the
    exact ratios, but I recall that the tin/lead ratio
    is just below 60/40 which is convenient; just add
    lead to standard solder. Check the purity of the
    bismuth shot, but as long as the ratios are close,
    the alloy should still melt below 120 degrees or so.

    Clifford Heath.
     
  13. NoSPAM

    NoSPAM Guest

    Many folks have suggested flexible waveguide, which is probably the simplest
    method. I'll present two other alternatives.

    You can design a mitered joint using only pieces of rectangular waveguide
    cut at the correct angles and soldered or brazed together. The design of
    both E-plane and H-plane corners of arbitrary angle is covered in
    Marcuviitz' "Waveguide Handbook". This was one of the volumes in the MIT
    Radiation Laboratory series published after WWII. The book was reprinted in
    1986 by the IEE.

    Finally you can mill out two halves with a CNC milling machine and bolt or
    braze the two pieces together. The seam does not have to be perfect if it
    is parallel to the E-field. Again see the "Waveguide Handbook" for a
    discussion of symmetrical slotted waveguide.

    Both of these methods are easy with rectangular waveguide. If you use
    ridged waveguide, they are still possibilities, but you would need a good
    machinist!

    Dr. Barry L. Ornitz
     
  14. NoSPAM

    NoSPAM Guest

    Indium? Wood's metal is approximately: tin, 12.5%, lead, 25%, bismuth.
    50%, and cadmium, 12.5%. And it is cheap - cheap enough to be used
    extensively in temperature controlled baths for organic chemistry. Of
    course the cadmium is quite toxic but the baths were always in fume hoods.
    For EE's and others not familiar with working with this stuff, I suggest
    reading the J/ T Baker (Mallinckrodt-Baker Inc.) MSDS at
    http://www.jtbaker.com/msds/englishhtml/w3500.htm.

    Dr. Barry L. Ornitz
     
  15. amdx

    amdx Guest

    Working at 35Ghz is REAL magic!

    Mike :)
     
  16. Guest

    snip

    That is how they do when building trumpets, the trick is they use soap
    water
    because it doesn't freeze as hard

    -Lasse
     
  17. Rich Grise

    Rich Grise Guest

    At what point do you switch to fiber optics? ;-)

    Thanks,
    Rich
     
  18. Rich Grise

    Rich Grise Guest

    I just went to McMaster and searched on "bismuth":
    http://www.mcmaster.com/#catalog/114/3532

    158F? That's not even BOILING! =:-O

    Hope This Helps!
    Rich
     
  19. Ahh, must be a different low-melting point alloy I was thinking of. Thanks.
     
  20. 7.xx by 3.5xx mm, with a wall about 1.2mm
    less than 90 degrees, otherwise we're flexible.
    Yes, it can. We also have a network analyzer to
    measure some parameters
    We have the corrugated stuff here too.

    Rene
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-