Greetings,

Really like the web site, great information.

1)I have a Crusader Series-M control (3axis) and investigating the AjaxCNC system as a possible route: has anyone on this forum done the change over and any caveats doint so? Big concern is accuracy, the old system has glass slides for feed back.... both x and y & z motors to ball screws are belt driven....

2)Using BobCAD ver19 any issues sending code to the AjaxCNC system? Does the AjaxCNC Mill system have Ethernet communications?

All feed back will much appreciated.

Thanks,
EF

Feedback is the key to the accuracy of ANY type of CNC system.

The glass scales SHOULD be telling your controller where the "table is" on your current system.

The encoder based servos used on the Ajax system tell the controller where the servo went - due to slop in table ways/gibbs, ball screws, motor drive, etc - your Ajax system won't be getting TRUE table position feedback.

If you dutifully go to the trouble to tune out ANY/ALL slop, you should end up with reasonable accuracy with your new system.

However, your "new" controller still won't/can't actually "know" where the table went because of it NOT having first hand knowledge of the TRUE table position with the new system.

As of present I would not recommend using the Centroid with a glass scale feedback. If machine is tight with little backlash would work OK. But control is so responsive that the lost motion between motor and scale can cause oscillations. They are working on a card that has extra feedback spots to accept encoder and glass scale. Also if scale doesnot have differential ouputs ( notted signals) you cant use it anyway, most scales dont. Just add encoders to back of existing motors and adjust backlash with parameter in control would work fine. If you were to look at 95% of the new CNC machines on market today thats how they all do the feedback,

That's a real problem when you go with dual feedback - the two can generate conflicting inputs and cause oscillation. Bridgeport ran into the same situation nearly 10 years ago on the Z axis of their Eztrak system. They finally resorted to glass scale f/b for the DRO and encoder for the 3D moves when they finally figured out what was going on. Nice to see that Centroid has reverified the phenomenon.

If you minimize slop and hysterisis, you can get comparable results (assuming adequate resolution and slop/hysterisis minimization) with either system. Trouble is that slop/backlash or slop/lack of axis stiffness cause problems that motor encoders can't deal with. The optical ones THEORETICALLY do a better job but there will be a lag/disconnect between the table and drive system due to unremoveable hysterisis/lack of stiffness. Backlash comp and only cover up so much - we spent weeks trying until we learned what was going on and what we could ultimately do.

You won't be able to achieve perfection, you'll only manage imperfection at best.