ShopBot Controller - Next Generation - What would you do?

[bleeth]

I'll be there at 7 on the 7th of July. Can I get paid by the sheet? Can I bring my own software and CNC? Can I make coffee?

Works for me!!

[ed_lang]

I like the Ethernet link between the control computer and the SB controller suggestion Dana makes.

The idea of having a dedicated CPU inside of the control box with a USB connector(s) on the box is one I have voiced several times. I would like to give ShopBot (the company, not the tool) total control over the system so everything is just like they designed and configured it. Then there are no issues with the parts that the customer supplies.

I hope everyone knows that they are asking for the price to go up again!

[richards]

Let's take the mystery out of the operation of a CNC router. Each move is made up of straight line segments. Each motor moves its axis in a straight line. A circle is cut by sending at least two motors a whole bunch of coordinated straight line moves. The various sensors are really nothing more than Go/No Go switches. If a limit switch is active, then there is a No Go flag. If a limit switch is not active, then there is a Go flag.

My idea would be to have a "master" PC computer disassemble each instruction into its component straight line moves. Each axis would have a dedicated computer, perhaps a low cost Z-world module or even a PIC controller with lots of memory. The master computer would fill the queues of the dedicated computers and a master 'clock' computer would 'tick' when each line in the queue was to be executed. Three Z-world modules could cost about $100 to $200 total. The interface between the computers would be ethernet. The Z-World Rabbit Core modules have plenty of I/O to monitor actual movement vs commanded movement.

There is no reason to have a Windows PC perform low-level functions. That's where the COMM errors come from. A PC can only handle so many interrupts per second, so why not off-load all real-time code onto dedicated computers that were designed to run real-time programs?

I designed master/slave Z80 computers in the early 1980s that could perform those simple tasks (albeit without ethernet). I wrote multitasking software that time-sliced all functions expected of the slave computers. That was thirty-years ago. With Z-World rabbit modules, a system could be developed at minimal expense and in minimal time.

[dana_swift]

I like the idea of direct support for 3D Bezier curvers, or 3D-NURBS. Not a series of straight lines beyond the individual pulses for the steppers.

Curves should be definable in 3D, or a possibility of compound motion, where X&Y are driven in a true circle, while Z plunges linearly. Then "drilling" would become an elegant machine operation instead of millions of micro move commands.

Ramps should be replaced with acceleration terms in G's, in the mathematical sense. The velocity and position with respect to time should be entirely predictable. Then pre-computing the time to make a cut should become quite accurate instead of a wild guess as it is now. I can imagine a minimum velocity where ramping would not be required, and then acceleration would be used to accomplish the transition to the steady velocity of the current move command.

The control language should support modern flow control, such as "for", "while", and "do" loops.

I have written about this elsewhere, but still believe it is the direction for the future.

Perhaps a future control board should be based on a high-bandwidth DSP chip instead of a uController based on the 8051.

If anything I am amazed at what Ted Et. Al. have been able to extract from the controller they are using. What could they accomplish if it had some real horsepower?

How about support for a water jet cutting head? Plasma? or Laser? A composite machine.. hmmm

The mind boggles.. its a question of what upgrades would justify buying an new bot just for the improvements. I think this is a good start at the list I would want to see.

Its not about the costs, when the benefits are substantial enough.

D

[ironsides]

George...
Take a .wav file you like that is loud and copy it to your C:\SbParts folder.
Add this to your post right before the end:
PLAY C:\SbParts\{filename}
PAUSE {number of seconds you want file to play}
END (should be there)

Gary,
Thanks for the tip about adding a .wav file at the end of my cut files, it works great.

George

[dlcw]

It would be nice to eliminate the comm errors. This would, I assume, necessitate putting a "PC" inside the controller. Use a USB key or a network cable to download the files to the controller and let her rip.

The PC inside the controller should be comprised of one of the mini mother boards available. This way as technology changes, the board can be upgraded as the technology gets better. The controller could have the necessary plugs to accept keyboards (wireless preferred), mouse and monitor.

The ability to remove much of the two-way communications issues would be a major step up. I would even pay to upgrade my controller to a new one with these features.

[Gary Campbell]

George...
Glad it worked for you.

Don...
I am with you. Micro ATX board inside the control box. Get a list of the basic required XP services, kill the rest. I would suggest Linux as an OS, but then my Link wouldnt work!

Now that the "masses" have the easy button, how about a "machinists version"? No previewers, in fact no preview mode, no editors, toys, toolbars, or setup, virtually no commands except run and exit. All work can be done on the design (other) computer and tranferred via thumb drive. A version of SB4 (sic) that has the axis position boxes, an integral spindle display (as the 5th or 6th axis box)

You know, an industrial type display for cutting only, no frills, not geared for the hobbyist or occasional user. Instead of a 3D colored toolpath, I prefer to know my offsets, tool numbers and axis 4-5 move and jog speeds. Throw in a few goodies for the indexer. And the big one..... max penetration into the spoilboard. I bet I am not the only one that would like that setting.

SB3 is superior to most of the gcode control software in features, programmability and ease of use. Not so for dependability, reliability or reputation. If OpenSBP ever hopes to become "an industry standard" it has to appear to be more "professional" than hobbyist and more bulletproof than buggy. And a higher comm rate that allows high speed and high resolution might be on the list.

Oh yea, and some of those things that those scary smart guys like Dana and Mike want too!

[curtiss]

It would seem the bot could run off a fairly small cpu/screen that would import toolpaths off a usb drive. This could either be a touch screen or a mouse interface for those who have grubby fingers most of the time.
A cell phone screen would be a bit small, but a 7 x 5 might be about right.

The mouse might be a closed system wireless that would work from say 75 feet away.

A clean bluetooth system would broadcast the screen to another computer for those who want to be out of the room...which can have its downfalls.

A small gooseneck camera on the cpu would show the bot movements and an E stop could be hit from the other room...

[michael_schwartz]

If the actual controller were integrated into the shopbot hardware all you would really need for an interface would be a very basic PC such as a netbook. Keeping the user interface separate from the controller would also add a layer of redundancy. The more complicated GUI, would be separated completely from the simple and hopefully reliable controller.

In theory as well a single PC could then be used as an interface for multible shopbot's.

I could see the controller itself having a very basic interface accessed perhaps with a keypad with a small LCD. The keypad would have the ability to load files that are stored on the controller, view basic information, and start/stop the machine. However a $175.00 net-book might in fact be cheaper when its all said and done than developing such a device.

[frank134]

All I want is to not see the words (LOSS COMM PORT). Did I pay $25,000.00 TO LEARN THE WORDS LOSS COMM PORT?

Is this really that hard to do?

That my only complaint.