Great progress has been made in the last week as I was up at the University and put the 2nd gen cavity under test for the first time.
Quick Summary -> The cavity needs better probes.
After hooking the cavity up to the network analyzer, the measured cavity produced about six different resonant frequencies between 2 and 3Ghz as expected. We first zoomed into the most prominent resonance between 2 and 3 Ghz at 2.282Ghz as shown:
- The graph shows a resonance at 2.282Ghz, with a Q of 228 (calculated by using markers 2 and 3) and an insertion loss of 0.5dB (S2,1).
- Although our target frequency is 2.45Ghz, the cavity does not have a tuning plate yet which means the cavity is longer then it needs to be and the TE0,1 mode is lower in frequency.
The problem is that because we can not see the e-field configuration inside, we were not sure if this was the right mode and can only guess based on indirect evidence. For example, we can compare the measured S parameter results to the simulated S parameters and look for a mode, which, like the TE0,1 mode, should also have a low insertion loss (i.e. low S2,1).
Today, I went back to the simulations to confirm which resonance is the TE0,1 mode and the diameter of both probes. Previous simulated results used two different probe diameters, 28mm and 30mm. In early results, the 30mm probe diameter looked to produce the best result, but, just to confirm, I redid the simulations with a much higher tetrahedral resolution. Our previous simulations had ~ 250K tetrahedrons done under CST MWS 2009 and I redid the simulations with 440K tetrahedrons with a newer version, CST MWS 2010. (The larger simulation takes all day and pushes the limits of the computer even with 6 cores running at 3.2Ghz and 8GB ram)
First, the simulated results done before the cavity was built (note the marker at 2.45Ghz. showing an insertion loss of 0.766dB):
- As shown, with a 30mm probe and with a higher tetrahedron count, the insertion loss of the TE0,1 mode is 4db!! Yikes. The calculated Q was still 67K, which is great.
- The resonance we measured in the lab at 2.282Ghz is very likely the wide band resonance shown in our simulations at 2.4383Ghz and the mode looks like this:
- Note the E-field starts and ends on the walls and is not the circular mode of TE0,1. (again, I used a cutting plane to make the e-field clearer)
- When in the lab, the next resonance above 2.282Ghz was not very clear and the insertion loss was worse then simulated (6db or greater I think) – I will confirm this the next time I am in the the lab.
My next action is to do a simulation with a probe diameter of 28mm and change other variables (not the cavity dimensions of course) to see if the TE0,1 mode can be made to have a lower insertion loss and perhaps a wider bandwidth. If the modifications work, then I will match those conditions as closely as possible in the machined cavity.
Once I figure out if the probes fix the problem (or if the cavity is a write-off), then I will figure out a better way to make them. Either using some type of jig to mold the wire in, or getting them machined directly. The probes I made are not very accurate (shown below) and likely affecting the results (they should be round and exactly 30mm in diameter, of which they don’t look to be either, LOL):
Until next week!
Here is a shot from the documentary footage as we were preparing the cavity for testing.
(Kevin, behind and left of us, is calibrating the network analyzer as Dr. Karumudi, myself and two masters students, Kashish and Gary (helping), bolt on the copper end cap. You can see the 1st generation cavity in the foreground.)