[bojan]

Last week I decided to try another solution - to keep the original handpad, modified to accept the 5x higher clock frequency (17.897725MHz instead of 3.579545MHz).

My DUO handpad actually has two independent stepper motor drivers for each axis (each with its own processor including crystal clock oscillator).

The new crystals would be the simplest option for this mod, but there was nothing available on ebay. Local manufacturer (Hy-Q) accepts any "strange" orders, but charges $45 per unit and lead time is 4-5 weeks.

Since the declination axis doesn't really require accurate timing for stepper motor, I simply used 16MHz crystal and solve half of the problem.

However, RA axis needs at least 50ppm accuracy (for 10 minutes exposure without guiding for 400mm FL and 4um pixel size).
Being RF engineer, I tried to force the original crystal to work in overtone mode, but without success.. Then I tried with external transistor oscillator, 3rd overtone was not the problem but the circuit didn't operate at 5th overtone reliably, so I decided to try with NB3N511 (Clock multiplier IC.. it is actually PLL, and it can be programmed for 5x output frequency. AU$7.70 at RS).
This worked OK - the same circuit can be used for other additional (timing belt) reduction ratios.. 2, 2.5, 3, 3.333, 4, 5, 5.333, 6, 8)
The modification is relatively simple (it is reversible in principle), it involves cutting two tracks (to disconnect crystal circuit from processor) and adding couple of wires to bring the new 5x faster clock to the microcontroller clock input.

With this mod my LXD75 has now much lower PE while tracking.
Adding the ST-4 port is still pending.

However, because the whole controller actually operates from single power rail (including motor drivers on PCB) via provided 6V regulator in the handpad enclosure, 32x slewing speed is now too high and motors are stalled at that rate.

Because of the lousy design (the designer didn't think much about the circuit... yes it does work, but... ), the solution to this problem is a tricky one - I need to rewire additional 16 outputs from processors in total to be able to have separate (and higher) power supply for motor drivers.
So I left this exercise for some other time.