The Craftmans Shop > New from Old

The Sequel - Oh Blimey I bought a CNC Lathe (Beaver TC 20)

<< < (107/260) > >>

WeldingRod:
Ooooh!  Tool (cabinet) lust!  Love the swing out shelves!  My Hardinge does that for collets.  I _might_ have made my own custom "board" out of Aluminum...

Sent from my SAMSUNG-SM-G891A using Tapatalk

cnc-it:
Also on the Mentor drive there are pots for tuning..I used them for setting the rpm after I had my motor re built..worth a try?

PekkaNF:

--- Quote from: cnc-it on November 02, 2018, 12:16:33 AM ---Also on the Mentor drive there are pots for tuning..I used them for setting the rpm after I had my motor re built..worth a try?

--- End quote ---

So does that drive has cascade contoller:
1) Inner loop for curren or torque i.e. motor "native" possibly analog control loop
2) Outer loop in PLC for position control

Cascade controller are pretty sensitive that way. When I had to tune somewhat similar system, we did it shis way:
1) Make sure that inner loop HW stop works with limitswitces or such (would be drag to crash it)
2) Remove PLC control loop totally from inner loop, we used programmed step respose generator for inner loop SP.
3) Use step response to tune inner loop as tight as possile, but make sure that it has enough stability to live with wear, changing load etc.
4) After inner loop step response is good , connect inner loop to outer loop and adjust outer loop.

Gain ratio is critical, inner loop should be least 10x faster or controllers start fighting. If cascaded controller outer loop is too tight to inner controller it will cause bad behavour.

Now, if the outer loop gain has to be too high for confort (it will try to ocilate in certain ocasions) there are few paramaters to play with like dead band (When position is "good enough" and correction would send the controller to hunt near SP) and derivate type parameter (how small error you want the controller to react).

We also used "Forward" controller type: Depending the difference on measured value and setpoint, the controller would lock the outer loop AND shoot the inner loop with precalculated (massive) step to jerk the system "close" and when "nearly there" would release the outer loop controller that would do it's job gently to control exact position. This needes the system to be modelled or tables of mass and transfer distance - like you change the moving mass and then your step response to moving distace had to be different.

Hope I did not confuse anything further. We did not made machine tools, but machines that sometimes had a lot in common with machine tools. I.E. linear rails, torque/speed/position control loops with electro hydraulics or electrical servos.

Pekka

awemawson:
Thanks chaps for the suggestions.

The control loop seems to be pretty well confined to the inner workings of the Siemens 6FX1121-4BA01 'measuring card', which takes in the spindle encoder and produces the -10-0-+10 for the Mentor drive.  and parameter settings don't exactly conform to what I would expect for a 'three term control' . If I set the gain such that it gets close to set point but then stops not at set point, and manually turn the chuck from side to side, there is a wide 'dead band' either side of set point with no significant servo action, and I have to go through the set point very slowly for the measuring card to notice the set point position and confirm 'in position' to the 820T controller.

I've set myself two tasks today. Firstly to order up some 15 pin SUB-D plugs and sockets to make an interceptor cable for the encoder to check all it's bits are toggling (task complete !). Secondly to remove the analogue input to the Mentor spindle drive, and run it from a battery box to check it doesn't have any offset or deadband of it's own - (I have a little drive box somewhere made up when I built the 4th axis for the Partsmaster)

Meanwhile I've made up an Excel spreadsheet of the 17 parameters that I 'think' are involved so that I can keep track of what I've changed - I've been printing out a copy after each 'session' and time and data stamping it !

awemawson:
Monitoring the analogue drive to the Mentor DC Spindle Drive when being driven by the controller it seems believable, about 3.5 volts at 1000 rpm and 9.99 at 3500. 3500 is maximum speed so sensible that that is 10 volts ie full range. I've not tried yet driving it from my little servo battery box  as a last minute booking for the holidays cottages threw us into a bit of a panic  :bugeye: I have however managed to find the little box and check that it's batteries are still alive.

Before 'the phone call' I did however have another play adjusting the M19 parameters and have found a setting that achieves 'lock' in that it arrives at set point. By adjusting the gain and seeing if the spindle under shoots or over shoots I was able to arrive at a compromise setting, but it is rather 'soft' and I think it is drifting as things warm up (not surprising). The gain setting is dramatically lower than my original parameters (170 as opposed to 2000) so I'm sure something is adrift elsewhere.

One handy feature is that I can send it to a particular angle, then using the PLC inbuilt monitoring facility find what angle it has actually achieved to let me know if undershoot or over shoot.

Complications this week end due to friends staying so little will probably be done, but it does provide another set of muscles to turn those cupboard over even if he is a septuagenarian like me  :lol:

Navigation

[0] Message Index

[#] Next page

[*] Previous page