wiki/ news/ 2004-02-06 - Built two more v2 CPAs (Wifi, GPS) to test the "altitude effects dielectric thickness" hypothesis

The GPS CPA was built using the new "eyelet" technique for the plated through hole. Both the technique and the hypothesis seem to work.

Based on a discussion this last Tuesday on what causes the thickness differences in the antennas, Tim, Jared and I met to build two more v2 cylindrical patch antennas (CPAs).

We first built up a WiFi (2.422 GHz) antenna, with the same technique we used in the desert:

  1. Cut out the FR4 antenna
  2. Cut out a piece of the 0.005" copper to match the height of the antenna. Make it 5.25 * pi = 16.493 in. long.
  3. Wrap the Cu around one of the bottoms of the Al airframe blank (the "mandrel") and hold in place with little bits of kapton tape.
  4. Start wrapping the 0.75 in. double-sticky polyethylene foam tape in layers around the Cu, starting the from bottom. That way you can just spin the mandrel around and feed the tape off the roll on the table so it is automatically aligned with the bottom edge.
  5. Each successive strip should be snuggled up next to the preceding strip as closely as possible without air gaps but also without bunching the tape. Use an exacto knife to cut the tape at the exact right place, but remove carefully since it might stick to the other side in the process.
  6. Some of the tape will undoubtedly run off the Cu onto the Al. To keep the Al clean, use a wrap of 1 in. wide masking tape right above the Cu sheet.
  7. Remove all of the plastic off the tape.
  8. Carefully align the antenna with the Cu sheet, and with the antenna FR4 straight up and down on the table, slowly rotate the mandrel while smoothing down the FR4 onto the exposed tape.
  9. Remove the antenna off the mandrel. Take a staple, and push it through the feed hole in the FR4 to make a hole through the Cu. DON'T put your finger behind the whole. Doh!
  10. Cut out a square hole in the Cu ground plane from behind, just big enough to the let the teflon insulator on the center conductor pass through.
  11. Strip some coax: make about 0.25 - 0.5 in. diameter circle of ground braid by unbraiding it, and be sure to leave only a small amount - like ~ 1/16 in - of teflon insulation.
  12. Poke the center conductor through the hole (this is a giant pain) and solder it, using flux.
  13. Solder the braid on the back, with a 150W soldering gun, again using flux.

The antenna turned out to be 0.066 inches thick, which is consistent with, but surprisingly thinner than, the flat antennas built here in PDX.

We then built another GPS antenna (1.575 GHz). This time, we used the cool new "eyelet" technology to make the plated-through hole for the center feed. To do that:

  1. Cut off the itsy-bitsy rivet-like eyelets with a jewelers saw. Take a piece of material (like brass sheeting) of the thickness you want, drill a hole in it to fit the eyelet, and then cut off flush to that surface. We used 1/32 in. brass, and that seemed to work well.
  2. Drill out the hole to the right diameter. Make sure it doesn't have a raised lip from the drilling.
  3. Tin the copper microstrips.
  4. Press in the eyelet, and then wack it with a pick or pin or whatnot in order to splay out the straight edges (make them round outwards).
  5. Once the straight edges are sufficiently bent over or out, tap gently on them with a nice clean hammer. That make it all nice and flat and rivet-like. CHECK TO MAKE SURE THE EYELET IS FLUSH WITH THE FR4 EVERY FEW TAPS!
  6. Solder on the eyelet to the pre-tinned microstrip.
  7. Proceed to building the antenna just like above.

This antenna also turned out to be 0.066 inches thick.


This is good news; it looks like our hypothesis of the change in air pressure causes change in thickness is sitting pretty... although there could be other, more true, explanations! But the air pressure changes seem likely.

Note that in Portland, the tape thickness is ~ 0.55 in. In the BRD, it's 0.070 in. So where is it 0.062 inches, like it should be? Well heck, at about 2,000 ft MSL, right about where the 3M manufacturing plant is in Minnesota. Go figure! But this brings up a critical issue: 3M has many manufacturing plants, and according to R.S. Hughes, more than one produce the 4466. This means we need to make sure that we get the foam from the same manufacturing plant each time we order!!

Note that the thinner tape made the FR4 too long; it overlapped itself by a few millimeters. On the v3 antennas, we'll need to take this into account and make the circumference length slightly shorter.

The eyelet technology works great; as long as it doesn't screw up the S11, it's a winner. We should try and get the right sized rivets and something better to install them with... but in an emergency, we can just do it by hand.

Although things seem to be falling into place, we still don't know why the flat antennas give us a lower Er than the cylindrical antennas. This is a bit disturbing, more so now because the air pressure vs thickness arguments seems to point out that the flat antennas should have a higher Er than the cylindricals. So we think it might be a geometry issue... but if it is, why doesn't anyone know about it? The results from this coming Monday's test should at least tell us if we should stick with our cylindrical data or figure out Plan B, whatever that is.