Stepper performance

I've temporarily mounted an older (but new-in-box) 4.4v 3.5A 300 ozin
NEMA 34 stepper to the 8 tpi leadscrew on my 9" lathe with 4:1 belt
reduction. Drive is a G210 running off 25 volts with the motor wired
half-coil. It certainly works and will make the cuts I need, however
rapids leave much to be desired at only around 8 ipm for reliable
operation (ie, that's over two minutes to traverse the 18" working
length of the lathe) This isn't the end of the world as the fact that
I'm driving the leadscrew via the halfnuts means I can easily
reposition the carriage manually, but it would be nice to run faster.

The most obvious suspect is the low power supply voltage, which was
set by having a transformer that could only be wired up in
increments of 25 volts and an older cross slide drive that was limited
to 40 something volts. I just aquired a torroid with 24 volt AC
secondary which I believe will yield about 34 VDC. Switching both
drives to this may help, or maybe I'll leave the gecko on the present
supply, but rewired for 50 volts (75v on an unregulated supply strikes
me as pushing it too close?)

Another possibility is the 4:1 reduction. I did that because the
inefficiency of the acme leadscrew coupled with all the drag in the
rack and pinion of the carriage was quite a hefty load. 8 IPM with
4*8tpi=32 revs per inch works out to 256 rpm. Would I do better to
change the reduction so that the motor stays in its better torque
range?

I was trying to decide if one could learn anything about the
performance of a stepper/drive system by looking at the waveform with
a scope. In full stepping mode, it would seem that chopping duty
cycle would approach 100% as the reactance approaches the limit of the
supply voltage.

To ease the friction loads I'm probably also going to add
needle-roller thrust bearings to the leadscrew mount, and hanging a
cheap ballscrew on the back of the bed is within possibility -
especially as it would save wear on the leadscrew.

Chris