How about using a 202-260MHz VCO and mixing it with 200MHz, then taking only the difference frequency?

2MHz-60MHz is too much range for a basic VCO.

I forgot to say, the application needs -70 dBc spurs and harmonics.  Even with a fundamental VCO I am not sure of how to easily suppress the 2nd and 3rd harmonics (some sort of tunable/switched LPF bank), but with a mixer and all of its products--not a prayer (especially at the lowest frequency output where they all pretty much fold over each other).

I have been hearing about tunable thin film ferrites, etc.  Was wondering if anyone had any wild ideas.

I don’t think you can even buy a manually tuned signal generator at 60MHz with -70dBc spurs and harmonics let alone a VCO.

Thinking again about your first post, it seems as though you are able to use a VCO which is discontinuous in its tuning capability, having to switch between ranges. This obviously allows more flexibility.

I am not sure why direct digital synthesis must necessarily consume 3x too much power. What power consumption are you hoping for and for what signal level?

So we could make the fundamental as an on-off square-wave switched power rail. Now we have odd harmonics to infinity. That means we need 61dB attenuation at the third harmonic to meet your requirements. Any sort of tuneable filter is bound to create its own distortion problems so switching in fixed filters is more likely to succeed. If we went in octave fundamental tuning bands we would have:
2-4MHz
4-8 MHz
8-16MHz
16-32MHz
32-64MHz
In other words 5 switched harmonic filters, which is not too bad. The trouble is they now require 61dB attenuation from double the fundamental to three times the fundamental, quite a steep filter characteristic.

How about also creating the third harmonic as a switched square wave, then summing with the fundamental to remove the third harmonic. This simplifies the filter characteristics and should be workable.

And how would you filter out the 2nd harmonic at 4 when the oscillator was at 2?  

I have maybe 35 mA at 3 V to work with.

I wasn’t planning on generating even harmonics at all, the Fourier series of a square wave being only odd harmonics. The second harmonics would therefore only be created by slight imbalances, which admittedly might not automatically be down at the -70dBc level. Maybe a trimmer would be needed to take out the last few dBs of second harmonic to achieve the -70dBc spec; that is down to experimentation.

The power available is one piece of data, but the signal level required is also important!

Hmmmm.  Let me screw around in the lab and see how far down the 2nd harmonic is.  I assumed it would be pretty high, but maybe it is not.