I got the circular design yesterday and tried to match it. Two observations:
A) This one I couldn't get to 433 mhz with my set of tuning caps at all. I needed to add a 7k resistor to get it to work - obviously not good for efficiency. Part of the problem is that this larger loop only needs around 1.6 pF to match and at this magnitude granularity and tolerance are just bad in caps.
B) Return loss was around -8 db for a cap only match at around 440 mhz. So this antenna suffers from the via related loss as well. The simple loop works better than a two sided one with vias.
Given results so far I think there are basically two overall directions: (a) a low q loop which is pretty inefficient, but implementable with standard parts or (b) a high q loop which needs to be custom tuned.
The possibility (b) of a high q loop with efficiency close to a dipole is what originally drew me to this antenna design. I see three options to still get there:
1) Use a high q air piston trimmer like Johanson's 5801 (
http://www.farnell.com/datasheets/35551.pdf) and custom tune each Mote.
I think this should enable efficiencies higher than 50% but it's expensive, large and prone to detuning after environmental changes. I ordered a few of these trimmers to try out. This is probably more to measure potential rather than as viable volume circuit.
2) Use a high q ceramic capacitor with a parallel varactor tuner.
This might give you 30% efficiency. Varactors are voltage controlled capacitors. They don't have a high enough q by themselves. But if you pair them with a fixed high q cap it could be good enough.
Unfortunately this approach invites all kinds of complexity: need a stable voltage to control varactor -> regulator required. Varactors are very temperature dependent -> need to retune frequently and/or compensate for temp changes.
3) Use a high q ceramic capacitor and tune the gateway.
Motes could continue to be simple if the tuning responsibility can be shifted to the gateway. This is possible: my gateway has a dipole which has a fairly wide bandwidth. The gateway could slice time into rx windows dedicated separately to all Motes. It would measure out the best frequency for each Mote and listen during its window at that frequency. The Motes would need a 32khz crystal to keep track of time.
I think this works in the US on 915Mhz where there's enough bandwidth available. It doesn't work in Europe where the 433 and 868 bands are fairly narrow.
Unfortunately I don't really like any of these options very much. Anybody with better ideas?
Joe