Advanced Search

feedback is a linear measuring scale for all axis, in case those remain, I can additionally take a card that would read them as an encoder

The EGB would add or subtract pulses to the table for helical gears based on hob height, as well as handling hob and table speed.

I bought the strongest driver of 7A, I will try as soon as I find some time to connect the existing motor in a bipolar connection with higher voltage and lower amperage, so we will see, if one works I will buy two more drivers and I think the problem would be solved, the controller in that case 7i96s for example.

The first CNC hob I worked with was a Pfauter PE 150 with an Allen Bradley 7300 and the Pfauter EGB (electronic gear box). 1983

thank you I will take a look at that

cheapo 1kw servo motor

motor model A2M-80SV03230
driver S2SVD20

It works well and was generally easy to get running, atleast without the whole switching between c axis and turning

The message is something that i introduced, and it means that the parport_pc kernel driver has not found the card. However that doesn't mean that the LinuxCNC driver won't work.
github.com/LinuxCNC/linuxcnc/commit/7bd5...dedfd986fb84f6a9c4a6
It's not easy to know what to do in this scenario. There needs to be some indication that there is not _necessarily_ a parport at the give address (but, also, there might be...)

You could set up a prox or mechanical switch in the saddle to detect the tailstock and abort homing. This might be easier to wire than a dedicated tailstock detector, given that you are probably already running a bunch of wires to the saddle.

My own lathe has resolvers, and I modified the Mesa resolver driver to work in conjunction with the LinuxCNC .var file to give me absolute homing (most of the time) That means that most of the time my lathe homes without moving at all.

wiki.linuxcnc.org/cgi-bin/wiki.pl?Psuedo-Absolute_Encoders

spumco wrote:

Why do I get the feeling this is not the first time you've looked at failed (or suspect) industrial components?  Spend any time involved with, or doing, NDT/NDE?

I spent 12 years at Sheffield University doing postdoctoral research on cracks of various sorts. 

andypugh wrote:

The moment is greater further from the shaft end, and a ball bearing isn't particulary stiff against torsion. Also the step that the bearing butts against, if machined poorly, might be a stress raiser. 

Why do I get the feeling this is not the first time you've looked at failed (or suspect) industrial components?  Spend any time involved with, or doing, NDT/NDE?

I have never found a use for encoder_ratio, at least partly because it relies on software-counting of the A and B phases.

If you are using a Mesa card then encoder_ratio won't help you.

But gear hobbing is nevetheless very possible. I do it regularly. I invented the idea of electronically geared hobbing, but I probably didn't invent it first

wiki.linuxcnc.org/cgi-bin/wiki.pl?Hobbing

Note that due to a problem with a certificate on the server that hosts the Wiki, https won't work. This means that to download the linked zip file you will need to copy the URL, edit out the "s" in "https" and then paste the URL back into the address bar. Sorry about this.

kevin_allein wrote:

How do other people deal with that ?

Most CNC lathes I've seen with a moveable tailstock I've seen fall in to the category of 'Operator Beware'.  Z-home is set to max positive travel, minus the tailstock length and bit for clearance.  If you've got the tailstock anywhere else and locked down you're in trouble.

I've seen one lathe - can't remember the brand - with a tailstock-in-place switch.  If the tailstock was parked in the usual location (right end of bed) the machine homed normally.  If the tailstock was not present at the switch homing wasn't permitted - an operator message "reset tailstock" was flashed on the screen.

I don't know what the procedure was if the tailstock were removed completely for long jobs... maybe an OEM-approved switch bypass device that indicates to the control the new do-not-exceed Z-pos value?

Cooking up a switch-based homing interlock shouldn't be too hard, but I'm not sure exactly how to go about it without writing a custom homing sequence.

spumco wrote:

Wouldn't misalignment between motor and driven component tend to break outboard of the bearing?  That's assuming the bearing didn't come with lots of negative preload from the factory, of course.

This is one of those cases where, given the final result, I can think of a plausible explanation. But that would be true in either case...
The moment is greater further from the shaft end, and a ball bearing isn't particulary stiff against torsion. Also the step that the bearing butts against, if machined poorly, might be a stress raiser. 

andypugh wrote:

I think we would need a better macro photo to tel if that was a fatigue failure or not, but it leads me to be rather suspicious of the motor-to-shaft alignment and mounting. Is there any type of flex-coupling there?

I couldn't make out any beach marks on the surface. A macro would be interesting, as would pulling the bearing shield/seal and checking for signs of heat or abnormal conditions.

Wouldn't misalignment between motor and driven component tend to break outboard of the bearing?  That's assuming the bearing didn't come with lots of negative preload from the factory, of course.

If you are looking for on site assistance, it might help to start with saying where you are located. There are Linuxcnc users all over the world. (And a few might even have spare time to pop in and give you a hand.)