Jump to content

Italian shed ornament - Nick's cnc router repower


NickJ

Recommended Posts

  • Replies 60
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

12 minutes ago, NickJ said:

VFD - Settings confirmed as all wrong, but not terribly so, but with "better" numbers the external tach is now all wrong, while I didn't get my head around the external output settings that feed it, for now I'll let it be until I can get my hands on a hand held tach, unless someone has a fancypants way to work out the number of poles if its not on the nameplate? Currently set at 2 because thats what it was and it turns........

Managed to spin up all axes again and get each one to home on the limit switches, good learning here, setting home as zero throws errors cos it trips the switch, moving home position to a nominal offset cured that. Also found a few of the limit switches not quite in position, so fixed them too.

IMG_1323.jpeg.b522d03156ceab0e3875a696816fec3c.jpeg

All Green! If I was so keen, the controller would accept and run a program :)
IMG_1322.jpeg.5b3c94d194f6a8aefb9a4e478b13589c.jpeg

In reality this puts me servo tuning away from using the machine, in true procrastination form I dug into the pneumatics...

Spilling the wires and tubes out from the head I set up a test manifold to find out which solenoids worked and which were NFG.

IMG_1317.jpeg.966ed69452b0d7033185b968a9e9474c.jpeg

Expected result, more bits dumped in the bin

IMG_1320.jpeg.948c5691bc9921f579c0d7aabb68fa0f.jpeg

Thankfully I was left with 7 operational 5/2 solenoids, a few other shuttle valves and a mountain of various fittings &hose, these were set up on a manifold and re-installed to work the tool change on the main spindle.

IMG_0908.jpeg.4dc33534777eeb6a9a5d2df5dbb79e41.jpeg

All of this^ is now this:

IMG_1340.jpeg.d9c0da4d3d6dd79de70b7a190146944b.jpeg

Not being one to leave a chance to test, a quick switchboard was thrown together linking the above solenoids

IMG_1329.jpeg.10f31b142e049e0256349454b2dd60e2.jpeg

In what I can only amount to a miracle, I somehow managed to get the polarity of the air lines correct allowing removal and insertion of a tool holder first time, while it was only manual, it gives me hope that if I can sit down with the ladder logic side, I should get tool changing on the functional list (Of course its not this easy, there are many interlock switches to monitor making sure everything is correct ie tool correctly loaded before 10,000rpm etc)
The tool change was also horrifically violent, must re-adjust the needle valves.

So, limit switches, homing and tool change. Time to sculpt?

No, theres still more distractions!

Enter the auto-greaser....

IMG_1331.jpeg.5bf0ba62a7e7b52317891ddd3482d8cb.jpeg

I personally hate these things, experience tarred from older, mistreated farm machinery where blockages cause dry joints and days disappear in feeble attempts to unblock lines or work out which orifice plate goes to which port.

But, when they work, life is pretty darn peachy. 

Using the valves assembled for the above air tests, I soon found the pump wasn't building pressure, a quick loosening of fittings and air purge had that raised to a healthy 200bar grease pressure (safety glasses on)

Cracking the furthest distribution block I soon had grease pouring out then followed by a MDF/grease goo flowing from the linear rails and ball screws.

IMG_1345.jpeg.40666e2e368623504f0ba1707ca84e17.jpeg

While the flow is pretty uneven, i'll give it a few cycles a day in the hope everything will even out, worst case the lines can be swapped for nipples.

Ok, so large phallic engravings now or what?

Nope, in the adventure of chasing the grease lines I found the X axis drive cover was actually split and therefore working out the actual reduction ratio is possible. Cover off locates, well, more grease.

IMG_1343.jpeg.6db1317ee8ddb89a02c72fcfdec2c150.jpeg

Grabbing the nearest doggy bag the goo was disposed of and wiped up

IMG_1346.jpeg.238aa30a112fc049a4b9aaeb7107a6be.jpeg

Fairly certain its just grease, the belt appears healthy but I might try locate a new one and replace it if the price is palatable.

And yes, this whole post has been a tease, I never managed to run a program, having a go at tuning the servos I realised I have no idea how to run the internal oscilloscope, back to the school of YouTube then I promise to cut something worthy.

In my limited 3 phase motor knowledge. I'd say it must be 2 pole to get some rpm.

6pole 920rpm

4pole 1440rpm

2pole 3000rpm

Link to comment
Share on other sites

15 minutes ago, rusty360 said:

In my limited 3 phase motor knowledge. I'd say it must be 2 pole to get some rpm.

6pole 920rpm

4pole 1440rpm

2pole 3000rpm

This is the nameplate:

1908202292_IMG_1313(1).jpeg.7f1d8d9728d7adfee5f58668a28317c8.jpeg

That aligns with our best guess of 2 pole.

4 pole would put everything out by a factor of two right?

Link to comment
Share on other sites

1 hour ago, ajg193 said:

I have a handheld soviet clockwork tach that is deadnuts accurate up to 10,000 rpm. Come over some time and grab it

Continuation of roadside sharns sounds like a good idea, i'll get in touch later this week.

Link to comment
Share on other sites

23 hours ago, rusty360 said:

Humm so 2pole motor at 200Hz be 12000rpm.

What's the max spindle speed?

According to the spec sheet for the machine 18,000 rpm, one explanation i’ve been given is 12,000rpm full load and 18,000rpm at reduced capacity.

 

  • Like 1
Link to comment
Share on other sites

43 minutes ago, ajg193 said:

What's the backlash like? Is it pretty much ready to throw back into mass cabinet production?

Whats backlash? ;)
 

Spent the day digging further into the greasing issues, pretty sure its now getting everywhere, lots of black gunk spewing out onto the rails and its much, much quieter!

still have quite a list to do before full operation, but yeah, could start smashing out cabinets if the will to CAD them existed.

  • Like 3
Link to comment
Share on other sites

  • 3 weeks later...

Servo tuning time....

 

So, how does a machine get from A to B? well, in basic terms you say hey, go to this location.

In greater detail, imagine you have a tennis ball in your hand and want to get it onto a shelf 10m up, in order to get in nicely, the ball will need to go slightly above and come back down, this is called overshoot, not really a problem until you want to machine an internal corner, that overshoot will eat into the corner and not leave a sharp edge. The ideal path is to just make it onto the shelf and stop.

There is an additional problem, in order to perfectly land on the shelf and minimise the overshoot, the max acceleration of the ball will be finite, limiting how quick you can get there.

If we add brakes to the ball, then we can throw it much harder, bringing the ball to a stop in just the right place, hit the brakes too hard and you'll need a harder throw, get it just right will minimise input energy which is an obvious goal for machine life etc.

For a machine, all these values are fixed in code so in order to find them we can either do really complex maths (which also needs many precise inputs) or we can force an input and look at the raw path it took to get there, iterating on initial acceleration and braking force until the error of reaching the desired position is minimised.

-Now before I get much further one can go way further down this path, the above is a simple explanation and my methods are not exact, but get the machine working to a level that the residual error is less than other errors, ie no need to roll in too much glitter.

How does the machine know where it is you ask? On the back of the servo motors that drive each axis is a rotary encoder that sends a signal back to the computer informing distance and direction, in this case, 8192 pulses per revolution, or one pulse for ever 2.5 micron of travel.

The inputs I will look at are the position error "joint.0.f-error" in halscope, a handy tool in linux cnc to look at raw inputs, this output provides us with a simple calculation of position expected minus position actual, if we adjust the input acceleration and braking forces, we can iterate around to get this to a minimum.

Starting values are hard to guess, thankfully I had the factory settings which where actually on the money for a conservative start.

Round one, request a 5mm movement:

IMG_1396.jpeg.6906a9a856ecd3a9f4fea6b2e08d9e8c.jpeg

That squiggly line is telling us we're close, but can it get better?

Adding some numbers, in real rough terms, P = the initial throw, D = the braking, I = A helping hand to hold you on the shelf FF0, FF1, FF2 are new to me but I was advised to leave I, D and FF0 alone, add small amounts of FF1 and even smaller amounts of FF2 which should get us there quick. Before FF1 and FF2 are added, the value of P was increased until the drive went into oscillation then backed off to stability.

IMG_1397.jpeg.28a2d0185321df77be7b8deb33ed2f07.jpeg

After some iteration where I double or half the input and react about that these are the numbers I settled on for the X axis, once the changes are in orders of magnitude I accept good is good, hence round numbers. The result:

IMG_1398.jpeg.4cb1b1a71a318f840e9a13e3fc18227f.jpeg

The output is now scaled 100x larger and the error is now visible in relation to each stepcount coming back from the encoders, if we view the square pulse as 2.5 micron, the overall error is in the order of 0.01mm, on a woodworking machine, no worries.

This was then repeated on the Y and Z axis to even better results:

Y axis:

IMG_1400.jpeg.06e7d2c58478f1b8c536a4c865990175.jpeg

Z axis was a bit weird, but I also had issues with the pneumatic counterbalance as it adjusted to grease in the raceways for the first time in 20 years, if it shows up in machining I may look at these again, i'm betting its not going to.....

IMG_1401.jpeg.1585c90dc8a72a58add7638dc9d1c9e4.jpeg

All the above tuning took me less than an hour, by far the hardest part was finding the correct inputs for the scope, searching google for information is a nightmare, so many Barries harping on for 45min+ on youtube but not distilling the required 2min of info or reams of data sheets written by software engineers (fuck know what methods they communicate in). Once the inputs were confirmed (Cheers Andrew, again) I could dial in on the signal, getting gain/zoom correct, moving forward.

  • Like 8
Link to comment
Share on other sites

14 hours ago, BlownCorona said:

will you be interested in running jobs though this once done? i have a miniature library to build for my wife (2 small walls floor to ceiling of shelves) that could be greatly more enjoyable if cnc cut. 

would also open up scope for more style too

Yeah definitely, while I don't want to spend my weekends shuffling jobs through, if its gunna take up half my shed it needs to be productive.

 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...