View Full Version : Turret style cylindrical cnc grinder build

16-12-2018, 09:15 PM

A quick rough draft of the grinder design (bear in mind all images are a work in progress, lot's of missing stuff).

Objective :
- Positioning of 0.003mm
- Repeatability of 0.001mm
- Grinding turret design (ie 4 grinding spindles on a indexable rotating turret)
- Mostly ID grinding wheels going from 15mm diameter up to 75mm diameter so a grinding spindle speed range of 6000 - 24000 RPM and OD grinding going from 100mm diameter up to 250mm diameter with a speed range of 1000 - 4000 RPM
- Servo driven headstock spindle to enable C axis grinding later, headstock to have a +90/-90 swiveling range.
- 315mm travel of headstock (along the Z axis)
- 315mm of travel for the turret (along the X axis)
- Linear HIWIN HGR25 with HIWIN GHG25-CA trucs, 2 640 mm rails per axis with 2 trucks per rails (might sound like overkill but grinders need a huge amount of dampening and rigidity)
- HIWIN 20mm by 5mm pitch ball screws
- DMM servos ? (I want to keep it stupid simple and reliable even if this implies a higher budget)

The base :
A single polymer concrete cast with aluminium insert to serve as bases where the linear rail will be bolted (I use aluminium because in the event of a distortion from the cold casting process it's dead easy to lap to correct)
To cast this the aluminium inserts will be set wrong way up on my surface plate (1650mm x 850mm DIN ) and the casting mold will be built around.
The grinder turret insert (X axis) are at Z0 (ie surface plate surface) but the headstock insert (Z axis) are Y+75 (various reasons for this offset in the build) so those will be set on precision machined riser blocks for the casting.
I'm hopping to get near perfect planeity of all insert on casting but will probably not have a perfect right angle, I'll correct that when setting the rails
Other than that it's really just a dumb polymer concrete block with countless cutouts for weight reduction and inserts for various bits to be screwed here and there, even so it's still a 310kg part bare...

The headstock :
The headstock rides on the Z axis and swivels on the B axis (and of course the spindle rotates on the C axis) , it has two basic parts :
- A carriage (a 300mm x 30xmm x 50mm low carbon steel machined part the trucks are bolted to and with the swivel bearing (B axis) on the other side
- The headstock which swivels on the carriage is a polymer concrete cast part holding the spindle bearings insert (also aluminium) and at 90 a support system for the grinding wheel diamond dressers

The headstock swivels manually and is clamped down pneumatically after swiveling to either allow tapper grinding or when swiveled +90 to dress the grinder wheels.
Swiveling angles will be determined by a indexing plate and pin system (the design is sufficiently "open" so that if later I want to have a servo drive for the B Axis it can be added, but for the moment I see no real need here).
The spindle is driven by a 750W servo, with a 3 to 1 reduction (standard timing belt) hence spindle speed range is 0 to 1000, the servo also makes it possible to do some C axis profile grinding in the future.

With a 200mm magnetic chuck the headstock assembly comes to about 115kg add a 35kg maximum part weight and design weight (to calculate servos) is 150Kg

The grinding turret :
The grinding turret rides on the X axis and swivels on the B axis, it has two basic parts :
- A carriage (a 300mm x 30xmm x 50mm low carbon steel machined part the trucks are bolted to and with the swivel bearing (B axis) on the other side
- The grinding turret which swivels on the carriage is a polymer concrete cast part holding 4 spindle support at 90 increments to align then with the Z axis.

Here too the swiveling angles will be determined by a indexing plate and pin system (and again the design is sufficiently "open" so that if later I want to have a servo drive for the B Axis it can be added)

With a standard assortment of grinding wheels (one 250mm x 15mm OD wheel) and 3 various ID wheels on various quills the turret assembly comes to about 125kg (weight does help with dampening so I am making no effort to keep it down)

The grinding spindles :
I plan to use some "cheap" 65mm x 188mm Chinese water cooled spindles (I'll call them motors from here on) in the 800W to 1.5kW range (800W being more likely) these can be the twin bearing design since they are only used as motors, the shaft (usually an ER11) will be machined to feature a simple flat drive key that will drive a specific spindle running in 2 BNH008TU (40x68x15) Nachi High Speed Angular Spindle Ball Bearing 15 degrees Contact Angle Abec-5 (or funds permitting 7) hence the Chinese motor bearings will not be solicited by the grinding forces.
Only one high quality VFD will be used to drive the grinding spindle since only one will be running at any given moment (of course so some kind of switching is going to have to be set up here).

However for the OD grinding where the speed range is much lower, I'm assuming running one of these Chinese motors below 100Hz is a big NO NO even if some of them claim a 0-400Hz range and that even if it were possible the avialable power would be close to nil so I plan to set up a transfer shaft (still to be designed) from one of the spindle to the opposing one with two belts drives each with a 1/3 ratio hence a total ratio of 1/6 which means running the Chinese motor in it's "optimum" 6000 - 24000 range gives me my 1000 - 4000 RPM grinding range with no extra motor or VFD

I'll start "simple" (euthanism I suppose) with only one spindle motor and set of bearings to start with to test the machine before investing in the 3 other spindles motors and bearings

Each spindle has a 10 50mm conical quill interface with a male M30x1 pull down screw (kind of like a Fischer UJD system) where the pull down screw retention is interchangeable between a right hand or left hand thread so as to accommodate whatever grinding rotation is needed CW or CCW

So that's the "mechanical / electrical" aspect of the thing now comes the part I'm clueless about.... the CNC
In terms of axis capabilities, a cnc grinder like this is similar to a turret lathe, but with specific types of cycles so I absolutely need to be able to set up my own macros, other than that were to go now ?

The drivers (steeper/servo):
I absolutely cannot allow for any type of missed steps or positioning error on such an application, if a cnc router goes to far or not far enough worse case scenario you break your end mill and your part, on a grinder you potentially have a 250mm grinder wheel running at 200 RPM
breaking and flying into pieces.... not good..
Hence I'm pretty stubborn set on using servos rather than stepper or even hybrid steppers (whatever they are).

For the grinder to be in anyway of practical use I "need" a repeatability of 0.001 mm so with a 5mm pitch that means a minimum 0.072 angular resolution on the screws hence at least a 5000 line par revolution encoder resolution if the servos are a direct mount (which I'm going for for the moment), I quickly calculated that 400W per axis is way good for the required acceleration and cutting forces so I was orienting myself towards some DMM servos (read a few good thing about them)

17-12-2018, 07:14 PM
Cant add much, but subscribed.

DMM is certainly worth a look.

Colin Barron
18-12-2018, 11:08 PM
It would probably be much cheaper and more likely to achieve an accuracy of 3 microns if you buy somthing like a Jones and Shipman cylindrical grinder and modify it to suit your needs. Most cylindrical grinders would be at least a tonne, there are a couple of Myfords mg12 on Ebay at the moment (255kg according to a website i checked). These mg12 machines can be fitted with a boring spindle which is mounted where the Myford plate is fitted.

19-12-2018, 02:00 AM
Well cheaper maybe (even thought in France these are expensive and hard to come by the last one I saw went for over 3500€) but having worked on some "old" grinders I know that getting one is something of an act of faith because wear is sometimes so high that a rebuild is needed and that's something I do not feel like attempting.
But the bigger issue is that the parts I need to grind are over the working envelope of a machine like an MG12