[rlockwood]

Semi related to kvom's question.. Are you using dedicated/corresponding pot/tool numbers? Or random/closest empty pot association?

Sent from my Nexus 5 using Tapatalk

[roundman]

Hi Ray, I like those parts. On the tapping do you think the machine is rigid enough to do it? I have hand tapped many holes over the years and it is the preferred method when I do not want to risk breaking a tap using the machine. I would say I now do not hand tap but maybe 2% of the time. It is a good practice to take a sample piece and use your same program cycles for spot-drill-tap to prove out that your machine will do a good job. With any tapped holes it is critical to have sharp tools and to blow all chips out of the holes before tapping so you do not drag existing chips in with the tap. It is good practice to gage the hole diameter to see that you have the hole on the right diameter round and straight. In metal take care not to peck too much at one time unless the speed is just right as it will kind of oversize your hole thus making it where your tap hole does not gage correctly. Tweeking it here and there to be able to use the machine to do the work brings nice rewards. Be sure you have a good chamfer on the edge of the hole also.

[SCzEngrgGroup]

Are you compiling with the default optimizations for size or did you jury rig Arduino to compile for speed?

Just using the defaults.

Regards,
Ray L.

- - - Updated - - -

Semi related to kvom's question.. Are you using dedicated/corresponding pot/tool numbers? Or random/closest empty pot association?

Sent from my Nexus 5 using Tapatalk

Initially dedicated, but eventually it will support full logical-to-physical mapping, so any tool can be put into any slot.

Regards,
Ray L.

[SCzEngrgGroup]

Hi Ray, I like those parts. On the tapping do you think the machine is rigid enough to do it? I have hand tapped many holes over the years and it is the preferred method when I do not want to risk breaking a tap using the machine. I would say I now do not hand tap but maybe 2% of the time. It is a good practice to take a sample piece and use your same program cycles for spot-drill-tap to prove out that your machine will do a good job. With any tapped holes it is critical to have sharp tools and to blow all chips out of the holes before tapping so you do not drag existing chips in with the tap. It is good practice to gage the hole diameter to see that you have the hole on the right diameter round and straight. In metal take care not to peck too much at one time unless the speed is just right as it will kind of oversize your hole thus making it where your tap hole does not gage correctly. Tweeking it here and there to be able to use the machine to do the work brings nice rewards. Be sure you have a good chamfer on the edge of the hole also.

I'm struggling to understand the context here - Did you post to the wrong thread? What parts? What machine?

Regards,
Ray L.

[rlockwood]

Just using the defaults.

Regards,
Ray L.

- - - Updated - - -



Initially dedicated, but eventually it will support full logical-to-physical mapping, so any tool can be put into any slot.

Regards,
Ray L.

After posting I basically came to that conclusion. Definitely one of the benefits of using more open hardware. Most other MTB's steer very clear from firmware updates..

[SCzEngrgGroup]

After posting I basically came to that conclusion. Definitely one of the benefits of using more open hardware. Most other MTB's steer very clear from firmware updates..

Rob,

MTB??

Regards,
Ray L.

[rlockwood]

Sorry, Machine Tool Builders

[SCzEngrgGroup]

This will, very soon, be the latest ATC video. Servo is looking good now, just needs some improvement of the PID coefficients, and a more effective damper on the carousel.

VID 20140609 124049 266 - YouTube

Regards,
Ray L.

[SCzEngrgGroup]

I'm happy to report that things on the firmware front are looking very good at this point. I have all four servos up and running, and all working nicely. I've made a lot of progress on the power-up and calibration part of the firmware. The intent is to have the whole system highly self-calibrating, so it will continue to work properly even as the hardware characteristics change due to temperature changes, break-in and eventual wear. This means sensing how much torque it takes to move each actuator (i.e. - overcoming internal friction), where the critical positions are for each actuator, and, eventually auto-tuning of the carousel and transfer arm servos. The carousel and transfer arm homing and calibration routines (i,e, - locating the exact position of each tool receiver, spindle and carousel positions, etc) are all working nicely.

There have been questions about the start-up state of the ATC, and whether it will detect if there is a tool in the spindle. I'm happy to report I have figured out how to do that detection, so the ATC will now KNOW whether or not a tool is loaded in the spindle on start-up, and be able to deal with it automatically. Since it will also know which tool receivers are actually holding tools, this means it can now remove the tool from the spindle, and store it in an open slot on the carousel, and not have to ask the user to remove it from the spindle (unless the carousel is already full.

I've been able to do the first "toolchanges", albeit with a little manual assistance to overcome some of the minor hardware issues noted below. Next step is to re-make some of the hardware as follows:

1) The "lock" coupling needs to be re-designed. The current one has a bad habit of pushing the carousel sideways if too much torque is applied. It works fine as long as the lock arms in the tool receivers turn easily, but I will nonetheless revise the coupling to make it self-centering, regardless of applied torque.

2) The transfer arm fork, which is a left-over of a previous design, requires too much force to open the spring-loaded "fingers" that grab the toolholder "ATC groove", making it difficult for the transfer arm to completely lock on. This means I sometimes have to give the arm a push to get it securely locked onto the tool.

3) The transfer arm drive has the following problems that will be resolved with minor hardware changes:
a) The pulley spacing on the belt drive is too close (despite using the value provided by the SDP-SI belt calculator tool...), so there is currently a fair amount of backlash in the belt, making the transfer arm very "sloppy".
b) The transfer arm encoder needs to be moved from the drive motor shaft down to the transfer arm shaft. This is necessary to deal with the inevitable backlash in the drive gearbox and belt drive.
c) I need to add a radial support bearing on the transfer arm shaft, below the belt drive, to prevent the shaft from tilting sideways under heavy load
d) I need to add a damper to the transfer arm, similar to the one on the carousel.
e) It's time to finally add the positive lock to the transfer arm fork, so the tool CANNOT come out when the arm is in motion - Don't want to be tossing tools across the shop when the arm moves....

Once these changes are implemented, which should take a few days, I think the whole ATC will be ready for its first serious use-testing. If that goes well as I expect, we could be ready for the first beta testing in a few more months.

The electronics and firmware are proving very robust and reliable, which is nice to see at this early stage. The FPGA logic also has been working perfectly. I expect within a few days to have a new, MUCH faster processor (~10X) to drive the whole thing, which will enable further improvements in the servo performance, and allow me, for the first time, to run multiple PIDs simultaneously..

[GJeff]

Whenever I feel like I have a lot on my plate with my Tormach teardown/clean/rebuild, I come here to make myself feel better.

What's your method for detecting tools in the spindle?

[SCzEngrgGroup]

What's your method for detecting tools in the spindle?

That, I'm afraid, will remain my little secret....

Regards,
Ray L.

[LeeWay]

If the arm crashes, there is a tool in there.

[SCzEngrgGroup]

If the arm crashes, there is a tool in there.

That was the solution I rejected very early on.... Call me crazy, but I think crashing, and breaking things, is just poor design. One of several things I never liked about the Tormach ATC is that it was specifically designed to break in the event of a crash. I think it should be designed first and foremost, to virtually never crash or jam, and second to deal gracefully with a crash or jam, and to specifically NOT drop a fragile, expensive tool pointy-end-down onto the table.

Regards,
Ray L.

[LeeWay]

Amen!

[treyjugson]

what's a tormach atc? not to be picky but i've read this thread from the beginning and there have been many design changes and detecting a tool in the spindle is relatively new, it was suggested by other members but dismissed by you initially as not important. i'm getting a holier than thou vibe even though this has never made a tool change in the real world and more to the point by a customer who may not be the most savvy that has to install this himself. i'm just saying maybe hold off on the attitude until this has been proven in the field. best of luck.
trey

[kvom]

I am assuming that the ATC operates in tandem with the PDB. And I also assume that even with the carousel installed it will still be possible to manually mount a tool in a standard R8 collet in the spindle. If that's the case, the arm would not be able to remove the tool nor install it in an empty slot.

[bozidar22]

A proximity sensor on the tool carousel can be used to find full tool slots. One on the transfer arm would detect a tool in the spindle.

Most H-bridges have a motor current draw function that would detect if something is jammed. Modern car power window work this way.

[hjl4]

Actually, it could be much simpler than that. Being that he designed the PDB, he just has to check and see if it is (locked) or whatever . No biggie.

[SCzEngrgGroup]

Made a little progress today. Machined the first of the new lock motor couplings, and it seems to work perfectly - it no longer tries to rotate the carousel, no matter how much torque the motor applies. And, it's very easy to adjust the amount of friction in the lock cam, to ensure the locks won't un-lock due to vibration. Those will be both be nice improvements. Now I just need to make 11 more of them....

I hope perhaps tomorrow I will also get the new transfer arm fork made. Still working on the design changes to the transfer arm drive, so that will be a few more days. One nice thing - as the lift drive wears in it's getting both quieter, and smoother. It's now down to a not unpleasant "whirring" sound - Soooooo much nicer than listening to an air cylinder and solenoid. Between that, and the new 10kHz PWMs on all the motors, this whole thing now operates almost silently.

Regards,
Ray L.

[tmarks11]

...i'm just saying maybe hold off on the attitude until this has been proven in the field. best of luck

How about you show us the ATC you made that has been proven in the field? No?

Ray's thread, on Ray's creation, I sense no attitude problems (on his part). He doesn't want to give away his trade secrets to nonpaying customers? Welcome to the real world. Some times you have to pay for things.