One thing, as mentioned before, we run a warm-up cycle in the morning, or when the machine is idling for a long time.
The best way we have found, is to also get the all the axes moving, not just warm up the spindles. So for example I will travel almost the full range of Z back and forth, and then go to around the middle of the machine (G53 Z-320) and do some XY movements, we do them in the middle just to make sure we do not hit anything because the tools change all the time... The nice thing is that you can write a relatively short program, and use the memory restart function to loop it for as long as you need. We also load and run this if the machine is idling for a prolonged time, just to keep it warm and moving.
Well still, that is a big deviation, I am really surprised. I also have one and it is sub 10 micron machine all day long. Which I consider really good given that this is the "cheap" Nak.
We regularly do some bronze bushings on it which have an +0.0005 to + 0.0010 inch OD tolerance, and after a proper warm-up we adjust the X offset just about 3-4 microns during the whole shift.
And at that point it somewhat hard to tell for me if this is thermal drift, or insert wear, since the inserts practically last the whole batch of 1000 pcs. We rarely change them, usually only if we cannot get the required Ra 0,8 surface roughness requirement.
Correct on the AS200 there is only one option for thermal compensation in the NT Nurse setting 32, which works based on coolant temperature. Still your mileage may vary according to the workload, we have this turned off, seems more stable that way.
On the higher-end Nak's with 30-series Fanuc, you can get the NT Thermo Navigator AI, but on the WT or MX it has not really been a be all end all solution to all of our problems at all times.
First, you really need to feed the system with a lot of data, if you don't, the compensation will be off. This means a lot of measuring and data input.
Second, since we do usually only do high-mix low volume, some parts are running just a couple of hours at a time, which is not a big enough sample size, to get the compensation working reliably, or I should rather say as good as on the parts where we have run over extended periods of time.
A third option, on the higher end Nak's you can also get scales, but I have never seen them anywhere in the wild, the machines seem to run well out of the box even without them. At around 16k for scales on XYZ, it might be an option for someone.
I also think you can order them just for the X axis for example at a reasonable price. However the difference between scales and no scales seems to be around the one micron mark, and I do not see a point in that for our workload.
One question regarding your AS200, what size is your coolant tank? I have noticed that for some reason, Methods delivers these with a "standard" coolant tank, which is quite small.
Over here, you can get one with an extended Coolant tank which is over 600l, and it has done wonders for thermal stability, compared to other small lathes we run with smaller tanks.
We have considered a high pressure system with coolant chiller for the AS200, but as it costs around quarter of the total machine cost, we have decided to stick with the standard Brinkmann 25bar pumps without coolant chiller for this machine, and it works really good as long as you have the extended coolant tank.