RE: [CAD_CAM_EDM_DRO] USB CNC (was Interface. Gauging interest.)

HI John,

Hopefully MACH and EMC would get "options" to use USB. And TurboCNC could
gain from it as well!

Of course, Bert of USBCNC has already essentially ported EMC over to Windoz
using USB to his firmware (USB PIC). I don't know how "open" this
architecture is or may become.

I'd even like to try switching my STEP4 to USB. It would certainly be
interesting to try!

> Thanks. So would I. Maybe after the ELS....but I'm not interested in the
> G-Code interpreter or MACH clone. OTOH, a standardized GNU type interface
> for this with a DLL library for basic motion would be useful for more than
> just CNC....

The DLL library approach sounds interesting. Then others could build their
own Gcode or interactive "front end" DLL.

Having several different levels of "USB box" at the end of the wire
available would make putting systems together quite modular and interesting.

>
> Simple protocol.
>
> Host sends Packet of 250 x 40uS step images, Packet#, IOPort Image (Pumps
> etc).
>
> PIC sends back Packet# received, Switch Image, Error Status which would
> include Packet # and Byte # when error occurred.
>
> Host is held off sending more packets until response received. Host also
> times out and sends Info Request packet which is smaller than the move
> packets.
>
> Have I missed anything?

Doesn't sound like it!

> >
> > You mention using the FTDI device, could the USB PIC do that
> > as well, or
> > would the USB core need to share interrupts with the PIC core,
> > and would they
> > "get in the way"?
>
> A quick glance at the device overview shows 1K of dual port USB ram. That
> would imply at least two of the 250+ bytes ping pong buffers. Still that
> might limit this type of unit to only 4 axis as opposed to a full 6 axis
> unit. Using an external FTDI device with larger buffer offloads a lot of
> things at the expense of a few more dollars. As in the ELS project, when
> is enough and when does it cost too much?

A PIC-based USB-parallel board like you describe could be quite compact; it
could easily be built into a driver/PS box. Almost like some are doing with
breakout boards.

In fact, this USB-parallel board could be quite generic, and then different
PICs with firmware could be available. I'm wanting to try USBCNC and
CNCdudeZ PICs (although they DO use different pin assignments for interface
functions...)

The guys at CNCdudeZ have commented that they can compile their early
(non-coordinated) code to use different pins. That might work! How about
it guys? (and if Bert at USBCNC wouldn't mind). Then we can use the little
boards for more projects!

> I could see a simple box that has 4 micro-stepping drivers based on the
> LMD18245 devices. That would do the Sherline size mills I guess.
> Alternatively drive the big guys with external step/dir lines.
>
> John Dammeyer
>

Exactly! I've just made room in my "Baby PCB Router" driver/PS box for a
USB-CNC-parallel board. The output will be a DB-25 connector (also to be on
the back wall of the driver/PS box), so that a simple short 25-pin DB jumper
cable can "daisy-chain" into the opto-isolator board (EAS). The opto board
will drive 4 LMD18245 stepper driver cards (EAS), and a SSS or two.

USBCNC has posted a schematic of just such a "USB-CNC-parallel" board for
use with his Windoz CNC program and PIC firmware. Or you could use a proto
board he recommends. It looks like a PICDEM FS USB demo board would work
also; although that's an expensive way to do it.

http://www.usbcnc.com/

OH, that does bring up an idea that I had. The PICDEM board has both USB
and RS-232 connectors, so like a PIC programmer I've seen, the
"USB-CNC-parallel" board could have Rs-232 also! Does that make it a
"USB-serial-CNC-parallel" board? ;>) That WOULD be flexible!

OF COURSE, I need to get the Baby PCB Router done first, and borrow a
driver/PS box from the Sherline lathe so that I can mill some PCBs! Well,
that's the plan, anyway!

Alan KM6VV