Thanks for the feedback. I think I've cracked the math of how much of Kerbin's surface will receive coverage, so I hope to update the document in the next few days. And for good measure I'll see about doing the math for the resonant orbits - that's a pretty good idea!
DStaal - the problem with RemoteTech and highly elliptical orbits is that the RemoteTech omni-directional antennas don't have a massive reach (the best one is 5Mm). The directional antennas, obviously, have much better reach, but then you end up with too many antennas on too many relay satellites.
Drew - yes I saw that page. I believe it gets one quite significant thing wrong - it doesn't account for eclipses by the Mun and Minmus which occur when the satellite is above Kerbin's terminator. I believe those eclipses tend to last far longer than eclipses by Kerbin. Also, I like to specify my orbits by period and not altitude, as it's vital to match the period between all the satellites in a constellation.
What I do is put 3 satellites in equatorial orbit with Communotron 32's in a period of about 1h30m (I think that gives best coverage for the surface of Kerbin when paired against Communotron 16-S's - I'll know for sure when I've checked the math). I also put the Mun and Minmus antennas (HG-5 and DTS-M1) on those 3 satellites. That's the Kerbin system covered.
For further afield, I use a pair of relay satellites in highly elliptical polar orbits, and put an RA-2 on them pointed at Kerbin. I tend to aim for a period of 38 days - it's easily achievable, and works well. I then put 2 (for redundancy) satellites in Keostationary orbits a bit ahead of and behind KSC with RA-2's pointed at the two polar satellites, and a Communotron-32 to reach KSC.