The radome design and manufacturing progress is briefly reported here for internal use by the SCIGN group. For others who are reading this, please realize we're in the process of trying to decide about all of this. We'd welcome input from others, of course.

We have designed a dome and are now working on having this design manufactured to meet our needs in the SCIGN array. We want to have a strong dome that will resist vandalism, yet it needs to be thin in order to minimize the signal refraction, so we are making it from Lexan. Also, we are trying to minimize thickness variations in the dome, especially azimuthal variations. Little of the GPS signal will come in from elevations below 5 degrees or above 85 degrees. Also, it is felt that thickness variations with elevation angle are a less serious concern than the azimuthal variations (although those are also a concern).

The first two prototypes are described in the measurements table, then analyzed a bit in a statistical table, and then shown in a few plots (see below). [NOTE: if you prefer to download the data in X, Y, Z ascii 3-column table format, you can get these files for either Dome A or for Dome B ].

Both domes were manufactured using a process called thermoforming. The first prototype, called 'A,' was delivered to us in mid-September 1997. Following that, we forwarded along some information on how to improve the process. Our reference for this was "Handbook of Product Design for Manufacturing," by James G. Bralla. Evenness of heating the sheet before forming is important. For thermoformed plastics, Bralla quotes wall thickness tolerances of +/-30% for normal processes and +/-10% for close tolerance. After this information had been passed along, some refinements were made to the technique, and the second prototype, called 'B,' was delivered to us in mid-October 1997.

We are now considering whether further improvements can be made with this same process, or whether we need to change to a new process, for example casting, in order to meet our specifications for thickness variation. We would like to have no thickness variation at all, but we might be satisfied with thickness variations of +/-0.2 mm azimuthally. With thermoforming, unfortunately, it seems that for an improvement in azimuthal uniformity, there is a degradation in uniformity with elevation angle.

There are, of course, cost considerations if we are to go to a casting process. We are thinking about this now and would appreciate input from others on this. We are soliciting bids from a number of different companies. Injection molding or compression molding, a.k.a. casting, would allow us to make domes with less thickness variation, but we think these manufactuing approaches might be more costly. There is a fixed, up- front cost of machining the mold, but then the per-unit cost for production is low. Also, there is a rather long waiting period while the mold is machined.

We are trying to resolve all of this as quickly as we can since it is holding up some of our field work and it looks like we'll be needing to make repeated trips to some sites, and possibly repeated site ties, so it is making our field efforts less efficient.

Prelimary Analysis of the Two Radome Prototypes:

It is pretty clear to us that in going from dome 'A' to dome 'B,' an improvement was made in the thermoforming manufacturing process. That is, the azimuthal variation about an average value, for a given elevation on the dome, was lessened in dome 'B.' This is apparent in the statistics table, where one can see an overall reduction in standard deviation and variance when reading across nearly every line of the table. Improvement is also evident in the plots - there are fewer contour lines intersecting lines of equal 'latitude' (elevation) on the domes. However, there were important tradeoffs. The variation with elevation on the dome unfortunately became larger in dome 'B.' Again, we believe this was the right tradeoff to make. It seems most important to minimize the azimuthal variations in thickness of the dome. The azimuthal variations in dome 'B' are near or slightly better than the hoped-for +/-10% value indicated as a reasonable expectation for the thermoforming method by Bralla. With some additional care and experience on the part of the manufacturer, it seems we could obtain satisfactory domes for the SCIGN array by continuing to pursue this manufacturing process.

As a rough estimate, variation in dome thickness maps to apparent phase center change with about a 1:1 ratio.

At this time we are not drawing a definite conclusion from all this. We are reporting this to the SCIGN group and will continue to have some internal discussions of this. Meanwhile, we are pursuing the casting alternative and will keep the group posted on costs and schedule.

One additional issue raised by this excercise is whether or not it would be worthwhile or necessary to make such measurements of thickness for some or all of the domes that will be deployed in the SCIGN array. For thermoforming, this would probably need to be done. For casting, each dome will be nearly identical so such measurements would not likely be needed. It would probably suffice to measure a few, average these, and use such an average model in post-processing. This could be a strong advantage in the long run. However, since we haven't tried the casting process yet, we are still concerned that things could go wrong with it.

GIF image comparison plot 1 (110 kB)

PS file of comparison plot 1 (1.4 MB)

PDF file of comparison plot 1 (1.7 MB)

GIF image of comparison plot 2

PS file of comparison plot 2

PDF file of comparison plot 2

GIF image of comparison plot 3

PS file of comparison plot 3

PDF file of comparison plot 3

GIF image of comparison plot 4

PS file of comparison plot 4

PDF file of comparison plot 4

GIF image of comparison plot 5

PS file of comparison plot 5

PDF file of comparison plot 5

Return to USGS Pasadena - GPS Index