[SCzEngrgGroup]

Some forward progress - I found a pulley in my junk collection that I could modify to get the belt drive going.

But, that bit of good luck is balanced out by a bit of bad luck - there is too much backlash in the belt drive! It's a 3/8" wide 0.200" pitch XL belt. The belt is MORE than tight enough - enough so that it's not easy to assemble the drive. Yet there is MUCH more backlash than with the arm mounted directly to the gearbox. The belt actually moves within the pulleys, so either the teeth on the belt are too narrow, or the grooves on the pulleys are too wide. Either way, its not acceptable, so the belt drive will have to either use a different belt profile (like GT2), or be removed entirely, which is a bit of a complication.

Regards,
Ray L.

[SCzEngrgGroup]

It's now working, albeit with the sloppy belt drive. Ignoring the backlash, positioning is excellent, and, most of the time, good enough to work. With a small change (about 0.020") in one dimension, I can swap out the XL belts and pulleys for 3mm GT2 equivalents, which should eliminate the backlash in the belt drive, and I should then be all set. Back to the SDP-SI web site....

Regards,
Ray L.

[SCzEngrgGroup]

After a little tuning, I'm happy to say I'm completely satisfied with the servo performance on both the arm and carousel. With a little bit more damping on the carousel, and the belt/pulley backlash removed from the transfer arm drive, they will be basically done. Both reach position very quickly, and settle in under a second to within 1-2 encoder counts of correct position - more than good enough. Both drives are "stiff" enough I don't expect any significant issues when running with heavy tools.

Regards,
Ray L.

[chevalier]

Ray,
I have been following this thread for some time and really appreciate your explanation of the design stages. The development videos are great to watch too. I wish I had your electronics knowledge. Could it be adapted to retrofit other brands of machine? I am from Australia and have a Syil X7 (BT30 spindle, air draw bar) just crying out for a decent ATC. Indexing of the drive dogs (VSD on standard asynchronous motor) could be an issue, but I have seen a similar machine run the spindle at 50 RPM (next to zero torque) and drive the dog up to a mechanical stop. So, I would have no issues retrofitting a spindle orientation arrangement. Anyway, I understand if its only for Novakon machines. Looking forward to seeing the fully functioning video.

Thanks in advance
Greg

[SCzEngrgGroup]

Ray,
I have been following this thread for some time and really appreciate your explanation of the design stages. The development videos are great to watch too. I wish I had your electronics knowledge. Could it be adapted to retrofit other brands of machine? I am from Australia and have a Syil X7 (BT30 spindle, air draw bar) just crying out for a decent ATC. Indexing of the drive dogs (VSD on standard asynchronous motor) could be an issue, but I have seen a similar machine run the spindle at 50 RPM (next to zero torque) and drive the dog up to a mechanical stop. So, I would have no issues retrofitting a spindle orientation arrangement. Anyway, I understand if its only for Novakon machines. Looking forward to seeing the fully functioning video.

Thanks in advance
Greg

Greg,

The basic design was designed specifically so it would be easily adapted to almost any machine by changing just a few very simple parts. It was also designed so that it can easily support 30 taper tools rather than TTS. However, It won't be offered for anything but Novakons until the Novakon versions are in production and fully sorted out, which, realistically, will be late next year at best.

I should have the first full-function video in a few days, as soon as the new pulleys and belts show up. I could do one right now, but it's not very nice to look at, with the backlash on the arm.

Regards,
Ray L.

[chevalier]

Ray,
Thank you for the reply. Its good news and bad news isn't it, yes it will be able to handle BT30 and adaptable to my Syil but I have to wait 12 months. Well, I am a patient person and I feel the wait will be worth it. Please let me know when I am able to purchase. Thanks is advance. I guess another silver lining will be that it will be fully tested by many before I have the opportunity to retrofit one.

Thanks again. I will look forward to the finished video.

Regards,
Greg

[kvom]

Have you thought about the process needed for someone to retrofit it to a current mill rather than a new buy?

Based on what I've read in this thread so far and the need for precision in positioning, I'd guess that once the carousel is mounted to the mill, some adjustments will need to be made to get the tools positioned accurately. Would all such adjustments be mechanical, or would it need some software tweaking? I'm also assuming that the carousel would need to be mounted with some precision vertically relative to the spindle.

[SCzEngrgGroup]

Have you thought about the process needed for someone to retrofit it to a current mill rather than a new buy?

Based on what I've read in this thread so far and the need for precision in positioning, I'd guess that once the carousel is mounted to the mill, some adjustments will need to be made to get the tools positioned accurately. Would all such adjustments be mechanical, or would it need some software tweaking? I'm also assuming that the carousel would need to be mounted with some precision vertically relative to the spindle.

Mounting will require drilling some holes in the column and the head carriage. Location of the holes will not be terribly critical, as the brackets and transfer arm are all adjustable to get things lined up correctly. Using a drilling template, it should be quite simple.

Regards,
Ray L.

[MichaelHenry]

Greg,

The basic design was designed specifically so it would be easily adapted to almost any machine by changing just a few very simple parts. It was also designed so that it can easily support 30 taper tools rather than TTS. However, It won't be offered for anything but Novakons until the Novakon versions are in production and fully sorted out, which, realistically, will be late next year at best.

Ray, was that a typo or do you think it will really take 18 months before the Novakon version is in full production? Seems like you are pretty close right now.

Mike

[SCzEngrgGroup]

Ray, was that a typo or do you think it will really take 18 months before the Novakon version is in full production? Seems like you are pretty close right now.

Mike

Michael,

I expect it to be in production late this year, very early next. But the real fun begins when production starts. There are ALWAYS a number of minor adjustments and a lot of fine-tuning to be done once it gets out in the field. That's when I'll find out about all the machine-to-machine differences, learn some of the goofy things people will try to do that will cause hiccups, etc. And there is always several rounds of design refinements to improve consistency and make manufacturing faster, easier and more predictable. It took a good 6 months to shake out all the details on the PDB, which is a much simpler mechanism. There were wiring changes, firmware updates, and several mechanical design changes. Those 6 months kept me very busy, revising the units on-hand, making updated parts for ALL the units built to date, getting those updated parts to the customers, updating documentation, etc. In fact, even the new build of PDBs I'm doing right now has a number of small, but significant, design changes based on potential problems I saw, even though no customer has ever reported any problems in those areas. For the ATC, I think a year to get everything really well shaken out is reasonable, particularly since I will still have to do the 18 and 24-tool conversion kits.

You'd be surprised how much there still is to do. While the prototype is now very close to "working", I don't think there a single piece that will go into production in its current form. Every part will end up being significantly modified - some for purely aesthetic reasons, but most for functional enhancement. And there's all the production fixtures and test fixtures to be designed and built, firmware to completed, the interface to the PDB to be written and tested (which will require updated PDB firmware), and an Ethernet-based Mach3 plug-in to be written.

Regards,
Ray L.

[LeeWay]

On a simple product like my blade guards, the design certainly evolved to finally get where I wanted it to be. I had to overcome limitations with tooling. Then got into cnc machinery. Techniques and tooling for thermoforming polycarbonate tube had to be developed. Then expanded on that. Then outsourced some of the manufacturing. A weekly search for better pricing on materials and better bulk suppliers. Finally at the stage in product development that satisfies me. Counting the first prototype, this one is version 11. Several things changed with each version number. Some very minor. Some major. Been doing it since 03, so it has taken awhile. I doubt Ray's will take anywhere near that long, but I do expect a few version number changes to get to that sweet spot.

There are several things you simply cannot account for on the drawing board. Nothing beats field trials.

[SCzEngrgGroup]

A little more good news - I've been cleaning up the firmware, making some minor enhancements, and just now finally had a chance to profile the PID code. As I'd hoped, it is now taking well under 1 mSec to run each PID update - Average is about 650 uSec, worst case about 850uSec. That is at least 10X faster than with the original 16MHz MCU. This means I could run a single PID at a 1mSec update rate. However, I see no functional advantage to this, as it's now running very nicely at 2mSec. I am instead going to retain the 2mSec rate, and allow over-lapped operations. This means I'll be able, for example, to operate both the arm and the carousel at the same time, or operate the lift and the lock at the same time. This will measurably improve overall toolchange performance.

I'm now working in earnest on the state machine code to sequence the mechanism for real, using all the available feedback from the encoders and sensors. This is in place of the largely "blind" (i.e. - limited or no feedback or error handling) code I've been using to date. Now that all the low-level code is working and very stable, it should not take too terribly long to get code in place that will provide near full functionality, including error reporting, and even error handling. I've also added a new safety feature - "watchdog" timers in the FPGA to shut everything down if the firmware loses its mind and stops updating things in a reasonable manner. I hope to eventually integrate motor current limit shutdown capability into the FPGA as well, as, like in the PDB, some of the motors draw enough current at stall to damage the motor and/or fry the H-bridge if not shut down quickly. This feature should make the electronics extremely robust, and almost entirely immune to damage due to firmware failures.

One thing that does surprise me - the firmware executable is already almost 100K bytes in size! Yowza! It did grow considerably (>2X) going from the 8-bit MCU to the 32-bit MCU.

Regards,
Ray L.

[MichaelHenry]

I expect it to be in production late this year, very early next. But the real fun begins when production starts. There are ALWAYS a number of minor adjustments and a lot of fine-tuning to be done once it gets out in the field....

Thanks for clarifying. This thread will be a good reference for future Zoners that wonders why ATCs and similar accessories cost so much and take so long to bring to market. Years ago my wife made me a needlepoint sign that said "It's impossible to make things foolproof because fools are so ingenious" and I expect you've seen your share of that over the years.

Mike

[SCzEngrgGroup]

Here's a short video of the ATC prototype doing random, over-lapped moves on the arm and carousel. You can see the backlash in the arm, caused by the XL belt drive. I hope to have the GT2 belts and pulleys in the next few days, along with the correct servo-motor for the tool lock. By that time, I hope to have most/all of the control state machine coded up, and be almost ready to run the first true, fully-automated tool-change sequences.

VID 20140712 145621 987 - YouTube

Regards,
Ray L.

[hjl4]

Wow, progress is really coming along.
Do you have plans to have it 30 taper compatible soon after the release of the R8 version?
Also, why so fast? Do you think you can incorporate an adjustable speed setting so customers can potentially slow down the tool change speed?
To me, slowing it down 1/2 it's current speed, would make for an easier and smoother action. Personally, a 7~10 second tool change is acceptable. Especially for the fact that you can walk away from the mill until the job is done.
Maybe I'm just getting old?

[SCzEngrgGroup]

Wow, progress is really coming along.
Do you have plans to have it 30 taper compatible soon after the release of the R8 version?
Also, why so fast? Do you think you can incorporate an adjustable speed setting so customers can potentially slow down the tool change speed?
To me, slowing it down 1/2 it's current speed, would make for an easier and smoother action. Personally, a 7~10 second tool change is acceptable. Especially for the fact that you can walk away from the mill until the job is done.
Maybe I'm just getting old?

Right now there is no 30-taper spindle available. When/if there is, I will do a 30-taper version of the ATC. While 30-taper would be nice, from a users stand-point, there is really no advantage - TTS can handle anything the machine is capable of. 30-taper would make my job easier, but once I have a working, reliableTTS ATC, what's the advantage?

The speeds, along with about a billion other parameters, are programmable, though its unlikely I'll release the information on HOW to change them. The final chip-to-chip toolchange time will be 10-12 seconds.

Regards,
Ray L.

[hjl4]

The final chip-to-chip toolchange time will be 10-12 seconds.

This is more than acceptable.
As for the 30 Taper, well I have a bunch of tool holders of that size. I guess I was just getting my hopes up. But if no market for such, then I totally understand.
Good job either way.
Will be interesting to see it in action on a real job.

[SCzEngrgGroup]

Here's another short video, showing the auto-sequencing of operations:

https://www.youtube.com/watch?v=eOnk8SDMfKc

First thing that happens is the arm is moved to locate its "home" position, behind the spindle. In parallel with that, the carousel performs its homing sequence, which finds the "home" (i.e. - tool 1) position, figures out how many tool "slots" there are, and which ones currently have tools loaded. After that, it does four "tool load" sequences. This is the sequence that would be used to load a tool into the spindle when there is no tool currently loaded. The arm is lowered, to clear the spindle, it then rotates to a position just clear of the carousel, in parallel with the carousel rotating to the (randomly selected) target tool. The arm then goes back up, then into the carousel, to grab the tool from the carousel. The arm goes down, to withdraw the tool from the carousel, then rotates back to the spindle position. It then raises up to insert the tool into the spindle, then withdraws to its "park" position. The whole sequence repeats four times. Obviously, the tool lock and PDB motions are not yet included in this sequence.

Regards,
Ray L.

[SCzEngrgGroup]

<RANT>
I had a friend who had a saying I always like, and almost every day that passes brings me more proof that he was absolutely correct:

The cumulative IQ of the planet is a constant, and the population is increasing. Do the math.

There seems to be a steadily increasing number of truly ignorant, immature, and ill-informed people out there. The kind that not only haven't learned much, but have no interest in learning anything, and are, instead actually proud of their ignorance and immaturity, and wear them as some kind of warped badges of honor. I bring this up, because of something that's happened twice in the last 24 hours. I've posted two videos of the ATC on YouTube. In both cases, within a very short period of time after I posted them (only 2-3 minutes in the first case, about an hour for the second), some nitwit gave each video a "thumbs down". I'm quite sure I know who this particular nitwit is. It seems to me, he must have actually gone to the trouble of subscribing to my YouTube "channel" for the express purpose of giving a "thumbs down" to anything I post there. He did the same thing here, for the (thankfully) brief period of time CNCZone supported that idiotic "feature" - did it for every single post I made for about a week, until the feature was disabled. How pathetic is that?

For me, this raises several questions:

1) What kind of person is so demented, so juvenile, so totally devoid of imagination, and lives a life so pathetic that the best way he can find to entertain himself is to "anonymously cyber-harass" a complete stranger?
2) Why do web sites even see value in including such a moronic feature?
3) What kind of person is so vain, and so lacking in self-confidence that they even give a rats a$$ what a complete stranger with the emotional maturity of a three-year-old thinks of them, much less a stranger with such obvious mental and emotional problems?
4) IMHO, this, and the whole Facebook/Twitter/etc culture makes me think civilization is headed down an ever-steeper slope right to a very bad place.

It is truly sad what the world is coming to, and how the lowest common denominator is being pulled lower and lower by the day, by these knuckle-dragging, mouth-breathing morons....
</RANT>

Regards,
Ray L.

[Hack]

Ray,

Very well said....

Dan