Feed Speeds - Turrett Mill

Kondia FV1 that had some use in industry before converting to CNC.
New ballscrews but original screw retained for Z.
There are 2 gas struts taking some of the weight of the knee.

Rpi 4/B with 2.8.4 Raspberry-OS/Debian and Mesa 7i96 (the easy iso).
Servo thread set at 1.5 mS, 1.500.000 nS, as 1 mS was reporting timing errors.
Latency without background software is Servo 77,000 nS, Base 66,000 nS.

What is general view on how fast a machine/software combination like this should feed ?

I have found that -Z freezes sometimes at 200 mm/min.
Seems +Z at 200 mm/min is Ok.
X and Y seem Ok at 200 mm/min.

I have set the machine at X, Y 200 mm/min, Z 100 mm/min.
This adequate for the work I do, steel and cast iron with 12 mm cutters.

Just looking for comments on the various trade-offs.

200mm/min seems slow. I run an Aciera F5 on ballscrews and have it set to 1000mm/min using 24V/5A stepper motors.

I have found that -Z freezes sometimes at 200 mm/min.

Do you mean the motor stalls? Maximum  feed rates would mostly depend on the motor and mechanics (ie. motor power vs mass/friction of an asix)

There are 2 gas struts taking some of the weight of the knee.

When I converted my Aciera F5 knee mill I opted for a pneumatic piston and a constant pressure valve. While this is bulkier and more expensive you get a constant lift force while with gas struts the upwards force will increase with more compression.

Yes, the -Z stalls.
I don't think friction is excessive, the Z can be moved by hand on the pulley.

For the X and Y, I was just being conservative,
This week, I will research the actual ratings for the stepper supply and motor ratings.

This mill operated for two years with a desktop and BOB, before the changeover to the RPi4.
I will go back through the history of completed jobs to see what actual feeds were in use.
As a new hobby CNC machinist, I think all my jobs were run conservatively (without coolant).

Of interest, what percentage of the knee/table weight was supported by the pneumatic piston ?

You can experiment with a little acceleration . How many nm is the motor? Screw pitch and gear ratio ?

Of interest, what percentage of the knee/table weight was supported by the pneumatic piston ?

It's a 100mm diameter zylinder so about 80 cm2 which gives 800N per bar. I estimate the knee/table to be about 300kg. I guess about 80% of the weight is compensated by the piston.

I have two gas struts, each one thought to be about 80 Kg, so not too much different to that reported by Aciera.
The test was done manually away from the machine by a friend putting his weight on it, so not a reliable number.

Records from previous tests with cast iron and HSS tooling had the X, Y running at 500 mm/min.
Previous jobs in steel with HSS were done at 100 mm/min.
All Z movements were about 25 mm/min.

X, Y pulleys 2:1, with screw pitch 10 mm = 400 steps/mm.
Appear to be Type 37 steppers (big ones).
Power supply 60 V. Leadshine drivers.

Is it possible the slow servo thread could be some influence here ?

If the previous configuration worked at 500mm/min I would try a little acceleration. You can also check the pulse length settings.
I would start by disconnecting the motor. Changing Driver Steps . I start from 800 steps per revolution of the motor.
With your configuration and this low speed, you will have a lot of vibration from the large stepper motor - you need additional software steps to smooth out the motion.

This is not a problem in the servo thread.

I think with the right configuration/setting the machine should work without any problem with 1000-1500mm/min 
Do you have an exact model of the stepper motor?  There should be a chracteristic curve... Maximum torque versus different rpm .