Gantry Printer -> Router retrofit.

Hello All,

New poster here. I've acquired a large 48" x 100" Gantry style ink jet printer (Leggett & Pratt VU98).
Video of one just like it.

Here's a quick description:
-It is a gantry style machine with total weight of around 12000lbs. It's only two axis currently & both are controlled by linear servos.
-The Y axis has two motor (one on each end).
-The X axis has just the one motor spanning the entire 100" left -right travel of the machine.
-The current Z axis has a nema 43 motor mount and ball screw. However, it's only attached with a handwheel as for the printer, it only needs to be set once of course.
-The servo drivers are ServoStar CD series (don't have model number handy) but are rated at 3.5kw.
-The pc driving the system has Parker Acroloop motion cards.
- glass scales on both sides of the Y-axis. (one / linear motor) Unsure of resolution at this point.
- unknown linear style encoder on the 100" X-axis.


My goals:
- Convert to Router. Primarily wood, but simple machining of aluminum plate would be nice for making brackets and such.
- Use for drilling pin/probe holes in 1/4" FR4 (fiberglass printed circuit board material) & light milling. This is for making text fixtures for a Teradyne ICT if there are any other electo-heads reading.
- Accuracy of .001 is fine for everything I would want to do with this.
- Come up to speed & learn more Linux awareness.

My Background:
- Electrical engineer & PCB board designer.
- Own a contract electronic design & manufacturing company. (we design/make pcb assemblies)
- Embedded systems programmer with a lot of experience in design/mnf of same.
- LImited Linux experience, but will shortly gain more I'm sure.

My thoughts/questions: (happy with any feedback others can give)
- Linear motors seem quite nifty, but you can't gear them up to gain force over speed so the amount of force you get is how much they have to start with. I don't have the model number off of them yet, but guessing from the size of the servo amps driving them (3.5kw), my gut instinct says these are plenty hefty enough for doing the light machining I am thinking of. Obviously the feed rate can be slowed, but I don't think linear motors are that common for CNC and that the industry has gravitated to servo/ballscrew. Should I be concerned about these? The print head it was moving around weighed about 200lbs, and the Y-Gantry weighs over 1000lbs I'm sure. It's 5"x12" x 10'. Boxed steel construction. 3/8" plate.

- Should I try to use the acroloop motion cards? I think they are a nice card from what I have read, but it doesn't appear that linuxcnc has support already for them. I write a lot of embedded code & probably could write an interface for them if needed, but I'd rather play with the router wiring/hardware & I think the basic concept of LCNC is for the pc to do the loop, not dedicated hardware. Plus I would rather not come up to speed on these as they are expensive and long term when I do another cnc I don't want to buy more of them. I think it would be better to ebay them and use the money to buy more open source or LCNC friendly hardware. Thoughts? Are these too nice to just sell of and start fresh?

- Assuming that I don't use the Acroloop motion cards, what would you suggest? I am just now learning about servo control as I have had limited experience in that area. I'm assuming that LCNC would drive the servos directly (step/dir or current mode or other???) and then use the glass scales as the feedback. I'm not sure if the linear motors have a quadrature encoder built into them or not. I think they do, but haven't looked at them in detail to verify that yet. If so, do I use them for feedback as opposed to the scales? Are there any motion cards that are already work with LCNC? I like the concept of offloading the loop to dedicated hardware. Many of the projects we have worked on do just that, offload the work to a micro and use the pc just for the GUI. I just don't know if that is the LCNC friendly way of doing things.

- Z needs either a stepper or servo on it as well. It's a nema 43 mount. I have a spare servo amp so am I crazy for even considering a stepper.

Thanks for reading so far, and any comments/advice would be greatly appreciated. I will snap some pics and post once I get the pieces moved into place.

Len

Andy,

Thanks for the information. I have some more information on the motors. They are rated at 169 lbs peak, 17lbs holding , 216 in/sec max. As I suspected, they do have quadrature encoders on them. I'm not sure if these quadrature outputs or the glass scales are feeding back to the acroloop motion card, but it looks like the acroloop is outputting +/-10V signals to the servo drives.

I think now that the best way may be to replace the Acroloop with one of the Mesa FPGA cards. If I understand the block diagram, the Mesa card will take the encoder feedback (and/or possibly the glass scale as well) & then drive my Servo Amps with +/-10V analog. Will the mesa cards be doing the motion control (via a downloaded motion control block) or would it simply be an easy interface to connect the LinuxCNC motion control to my servos? In addition, I'm assuming that there are plenty of I/O on the mesa cards to read in limit switchs & such.

As a side note, we do a lot of Altera FPGA development vs. the Xylinx. We do program Xylinx FPGAs customers products, just never designed them into anything as we have all the Altera tools & are familiar with that tool chain. Sadly, none of our existing customers are in the motion control industry. We did just pick up a new customer that builds servos though, but not any of the control electronics, just the motor itself, so the boards are simple encoder type boards. Soon, we will have boards we manufacture in some guided missiles and possibly up in space on a satellite.

Lenny

Excellent. I think I can sell the Parker Acroloop for a good sum on Ebay. I haven't looked at which exact cards are in the system, but they all seem to bring a good value.

As to how I like to interface to the hardware. I've designed a few systems in the past with embedded PCs at the primary control & they all seem to fall into plug in cards with an Altera FPGA target that breaks out the I/O and analog. I say card, but it's always a custom motherboard that an off the shelf CPU module plugs into. One was an electron beam controller with 12/12 analog in/out and about the same with 24v io, It used a 586 class ISA plug in card. Another was similar in design only using PCI instead of ISA and employed a Kontron ETX module for the cpu. It was for a 911 call system, so more audio channels & eithernet & far less I/O. Still it was a FPGA hanging off the PCI bus as the interface. If I rolled one from scratch for this, I'd probably use PCIe & breakout from the cards in some handy manner. However, the Mesa boards utilize the exact same design philosophy I would so I will just use them....or one of the others you mentioned. I'll do some research, but from what I have seen of the Mesa, it looks like an easy quick solution. We currently are working on a medical design that I can use the base board/cpu (Intel Atom based) combo and plug in Mesa (or similar) style cards for a nice system I think.

My more immediate problem is physical space. I just measured the gantry beam. It's 14' total in length. Because of the print head being so wide, they needed a lot of overhang to get a full 100" of printable area. I can squeeze it in my metal shop but only barely, and it will interfere an area that I try to keep open to work on stuff. I may have to move it to my woodworking shop. It's bigger, but more full of heavy machines which require more 4000lb tetris games. I could also chop off 12" from either side of the gantry & gain back a little room.

Thanks again for the pointers to some hardware.

Len