[rlockwood]

Its starting to look good, but man, that transfer arm looks wobbly. What's up with that?

[SCzEngrgGroup]

Its starting to look good, but man, that transfer arm looks wobbly. What's up with that?

Still some slop in the belt drive. I can't get rid of all of it, due to the backlash in the gearbox. Functionally, it no longer matters, since the position is only critical at the carousel and the spindle, and in those two positions, it is positioned by that pin sticking up from the end of the arm contacting the spindle or the tool receiver.

Besides, with TTS, there HAS to be a fair amount of compliance, or the tool shank will jam going into the spindle. 30-taper would've made this whole thing a lot easier.

Regards,
Ray L.

[engnerdan]

Have you thought of how your plan to lock BT30 tool holders into the pods yet? I am curious because a ATC is on the wish list for the future of my machine, and free time being in short supply I would rather buy then build.

[SCzEngrgGroup]

Have you thought of how your plan to lock BT30 tool holders into the pods yet? I am curious because a ATC is on the wish list for the future of my machine, and free time being in short supply I would rather buy then build.

I can see several ways to do it. Given that my PDB has no problem screwing and unscrewing the drawbar, I'd be tempted to just use a drawbar in the spindle, and "mini-drawbars" in each tool receiver. A miniature version of my PDB would then handle the tightening and loosening. The other option would just be to put a very simple ball-bearing "grabber" in each receiver.

Regards,
Ray L.

[SCzEngrgGroup]

Its starting to look good, but man, that transfer arm looks wobbly. What's up with that?

I've also decided to change the transfer arm drive configuration in the next iteration, and eliminate the belt drive. That will reduce the backlash at least some. Zero backlash is just not achievable, without going a an uber-expensive gearbox, which would completely blow the budget. As I said, the "wobble" is a cosmetic problem that has no real effect on functionality, so I'm not terribly concerned about it. A damper, similar to what I have on the carousel, might be enough to mitigate it.

Regards,
Ray L.

[rlockwood]

I've also decided to change the transfer arm drive configuration in the next iteration, and eliminate the belt drive. That will reduce the backlash at least some. Zero backlash is just not achievable, without going a an uber-expensive gearbox, which would completely blow the budget. As I said, the "wobble" is a cosmetic problem that has no real effect on functionality, so I'm not terribly concerned about it. A damper, similar to what I have on the carousel, might be enough to mitigate it.

Regards,
Ray L.

I think i'd read that previously and was mistakenly thinking this WAS the new iteration, and found myself thinking "that doesnt seem much better than the last time I saw it.."

It doesn't look THAT bad in the video, I just think it'd be off putting to buyers if the production version went without improvement in the area (regardless of its lack of effect on functionality.) Glad to hear you're already on the case.

[hayes]

Ray,
I've been fascinated following this thread. Are there any updates on progress for the ATC?

[SCzEngrgGroup]

Ray,
I've been fascinated following this thread. Are there any updates on progress for the ATC?

Sadly, not in a while. I've been up to my eyeballs building more PDBs, and dealing with a ridiculous number of non-work issues that have consumed a great deal of my time. Hopefully the worst is finally now behind me.

I did, just yesterday, finally have some time to make some forward progress on the re-design of the transfer arm drive, to eliminate the (troublesome) belt drive. I now have a nearly complete model of the new design. Once I complete the design, and build it, I should make some quick progress, as everything else is working very nicely. The previous design had too many problems, introduced by the belt drive, which should all be eliminated in the new design. The new design also simplifies position/speed sensing on the lift mechanism.

Regards,
Ray L.

[SCzEngrgGroup]

Below is the (almost complete) new transfer arm drive, with only a few minor features missing - the sealed housing for the encoder, the lift limit sensors, and the tool fork lock mechanism.. I've increased the lift leadscrew from 1/4"UNC to 1/2" ACME, and the "leadnut" will be Acetal, with the screw completely enclosed, which should make the lift smoother, quieter, and maintenance-free.

Although it looks almost exactly like the previous version, they actually do not share a single part. This one ends up being more compact, quite a bit simpler to build, cheaper, and lighter. The belt drive is completely eliminated, so backlash on the arm will be entirely determined by the gearbox. I would have preferred to be able to mount the encoder up top, but there's no way to do that, without it being on the "wrong" side of the gearbo backlash. This design is about 2" shorter than the previous one, so I will design the transfer arm with a step-down, to keep the drive assembly up high and out of the way. That will allow me to keep the existing mounting bracket as well.



Regards,
Ray L.

[SCzEngrgGroup]

Made some forward progress today - got most of the parts for the new transfer arm drive made, and I'm really pleased with how it looks, and the parts all fit together perfectly. I thik the new Acme leadscrew and Acetal nut will be a big improvement - smoother, faster, quieter, and more forgiving of misaslignment between the motor, screw and nut. .I think this new design will resolve all the problems I had with the old one, and it's considerably quicker and easier to build.

Along the way, I did several things today I've never done before:

1) I cut my first internal snap-ring groove, to fasten the new Acetal leadnut to the lower plate of the drive
2) Did probably the most aggressive HSM machining to date - cutting 1.2: DOC, 0.05 WOC, at 80 IPM, with a 1/2" HSS 2-flute. The spindle motor was running at least 80% power the while time.
3) Made my first piece ot stock! When you need a piece of 2"x2" 6061, and all you have it 1"x2" material....
4) Made my own 1/2-10 Acme tap, by "fluting" a piece of Acme rod. It actually works surprisingly well, though I wouldn't want to try cutting metal with it. But it works ok on the Acetal, and is a whole lot cheaper than buying one! I will buy one down the road, as I'm sure it would give a slightly better result, and cut more easily.

Regards,
Ray L.

[keithmcelhinney]

I would love to see a video of a 1.2 DOC HSM toolpath! Damn!

-Keith

[neilw20]

In Acetal, you will find the tap dulls surprising quickly, even a HSS one will quickly get a dulled edge.
I had a few hundred holes to do in abs with a 15/16" drill, and it only took about 40 holes to become really rounded edged.
Just pushed harder and sharpened it once.
Carbide rocks.

[SCzEngrgGroup]

Woo Hoo! Scored a lovely little Procunier 1E tapping head for only $100! These are MUCH better for smaller taps than the Tapmatic and Chinese ones, as they use a friction clutch, rather than a dog clutch, so they engage much softer when you reverse. One nice bonus is it's fairly easy to make custom collets, which sure beats the crazy prices Procunier charges for them. And the collets have a stop in them, so when you swap in a new tap, it will install at exactly the same length as the old one. One thing I think I'll try to do is make a quick-change collet chuck for it.

Regards,
Ray L.

[SCzEngrgGroup]

The Procunier tapping head is soooooo much nicer than the Chinese one I have! The clutches make it engage in both directions without any "jerk" - just a nice, smooth engagement, without the big "clunk" and jerk the Chinese one has when the dog clutch engages. I'm going to have to watch for more of these.

Almost have the new transfer arm drive ready to test - just one more part to make. The new leadscrew and nut are a HUGE improvement - MUCH faster, and quieter, with a lot less friction.

Regards,
Ray L.

[SCzEngrgGroup]

I'm happy to report the new transfer arm drive is now working nicely, after a short detour to build the first prototype of my custom Hall Effect rotary encoder, after the cheap POS Chinese encoder I had been using died. The list is now fast, and reasonably quiet. There's still a bit of noise when it's moving, which is caused by the disc coupling between the motor and leadscrew rattling around in the housing. I should be able to quiet that down with a minor design change. I also got the lift sensors on and working. These basically act like home/limit switches to tell me where the lift is on power-up, and to prevent the drive from trying to run into the limits of the leadscrew.travel. Positioning is now very good - settling quickly to within 2 counts, which is about 0.013" - MORE than good enough.

Next step is getting the drive mounted to the test fixture, then making at least one new part for the transfer arm, to provide a downward step, to keep the drive itself up as high as possible.

Another side-project I'm working on is cobbling together a set of quick-change tap holders for my new Procunier tapping head. I've got a concept that should be really trivial to build, as the "collet chuck" is a very inexpensive, off-the-shelf bit I already have several of.

Regards,
Ray L.

[SCzEngrgGroup]

Had another good day on the ATC front. I made a few new parts for the lock drive, and it's now working MUCH better. The transfer arm drive is working beautifully, except I need to swap out the output shaft on the gearbox - the one in there right now was damaged a long time ago, and it's now affecting performance. Tomorrow I hope to re-make the bracket for the two carousel sensors, and then things should be in pretty good shape on the hardware front.

Regards,
Ray L.

[SCzEngrgGroup]

Here's a short video showing where things are now. The old transfer arm drive bracket does not allow the arm to line up correctly with the carousel and spindle, so I can't put the fork on it right now, but all else is working nicely. The new lock mechanism and the new lift are both working really well, and I've improved the carousel positioning, so moves are pretty "crisp" now (single tool moves are kinda slow, but I'll fix that later). I swapped out the gearbox for an undamaged one (the last one was the one I used for the early PDB development, so it's had a pretty hard life....), and backlash is now quite acceptable.

The video shows a sequence of about a half dozen random tool changes, starting and ending with no tool in the spindle.

VID 20141016 134458 776 - YouTube

Regards,
Ray L.

[Hirudin]

Very nice!

[LeeWay]

Yep. It is coming together.

[SCzEngrgGroup]

The last several days have seen a lot of progress on the ATC. The new transfer arm drive assembly appears fully ready for prime time, except I still need to add the lower bearing to the output shaft, which will stiffen up the arm a bit more. The new lift is working perfectly, and now that it's broken in, it's nice and quiet. As a bonus, this new drive design is considerably quicker and easier to make and MUCH easier to assemble than the old one

I've made a lot of progress on cleaning up the firmware, and it's working remarkably well now. I see only one significant issue that needs addressing - I am seeing failures to unlock the tool from the carousel at an unacceptable rate. In fact, this is the ONLY error I've seen in the countless complete toolchange cycles I've run over the last few days! I know exactly what's causing this one error, and it will require a hardware change to the lock mechanism to correct. This change has been on the ToDo list for a while anyway, as the current mechanism, while it works well, creates a bit of a chicken and egg issue in system initialization, as there is a co-dependance between the lock "homing" cycles and the carousel "homing" cycle - the lock can't be homed unless the carousel is in the correct position, and the carousel can't be put in the correct position unless the lock is in the correct position. All works well as long as the carousel is in the correct position on power-up, but if not (i.e. - in the event of a power outage) things may not initialize correctly. The new lock mechanism will eliminate this co-dependence, making the two mechanisms completely independent of each other.

One thing that has really surprised me is how good the whole system is at detecting faults, and self-recovering after faults. I have yet to induce a failure that was not properly recognized, and simply re-running the failed operation is nearly always enough to get things operational again. For those rare cases where it's not able to recover automatically, there will be a very simple manual "reset" procedure to put everything back into a consistent state - basically just putting the tool back into the carousel, and re-homing all the drives, exactly as happens on power-up.

The electronics "bread-board" has gotten to the point I will have to think about making a real PCB before long. It's a giant rats-nest of wires at this point. Surprisingly, despite how ugly it is, it all works flawlessly - I haven't had a single unexplained glitch or crash, and not a single hardware failure of any kind - even on the motor drivers. The new Arduino has been a real boon - providing virtually unlimited resources. I'm nowhere near using all the bandwidth, memory and I/Os it provides, which makes the programming a lot easier.

Regards,
Ray L.