Lots of good suggestions in these posts - thanks.
First, I've called my project CirCut (~circuit) as opposed to Cricut (one of many CNC paper cutters available at a "hobby" price today). Sorry for the confusion.
The Cricut I have (Expression) but am not using now uses full or half step on the stepper motors, not microstepping (I believe). You can usually tell by the sound as I've worked with steppers many years and have begun to learn their sounds w and w/o microstepping. I can put a scope on it to tell for sure. Microstepping (should) gives you more fine control (and is more "musical" than half or full stepping) over the step size, velocity, etc. The Sillhouette Cameo I am using now I believe uses microstepping.
That said, if I were to suggest a modification where I take over the motors and software, an obvious solution (other than find the direct step&direction link from the machine's USB port to its motors) is to buy a few microstepping drivers and go with LinuxCNC, Mach3, etc and do it from scratch. Check out Sparkfun for some decent cheap modules to do this. If you check out PCB-Code (Yahoo Groups) there's a free Eagle-to-G-Code converter so you're almost there.
I hope you appreciate my route in this thread - to follow this method until either it's determined it's a failure and shouldn't be pursued further or it's a success and competes favorably with present methods which are - mask-WetEtch-drill, cut&apply copper tape, or mill&drill.
With that in mind, I do agree that the mask-WetEtch-drill and mill&drill methods do produce a better PCB...but at what cost and are the trade-offs worth it? The drag knife method has several advantages over others (IMHO):
1. (So far) requires essentially no modifications to hardware or software. This is a biggie IMHO as a lot of PCB makers don't want to bother with hacking h/w and/or s/w. BTW, one small mod I implemented is to support the weight of the drill accessory handle with a spring.
2. The cost is relatively low.
3. The speed to finished product is the quickest (I believe).
4. No mess with chemical etchants.
5. The knife blade kerf width is as small or smaller than a milling bit can make consistently (at least comparing it to our Accurate360 $15k machine here).
6. Board flattness doesn't seem to be an issue (a popular discussion on the PCB-Code Yahoo forum, a mechanical mill etching discussion group.
7. The berm left by the knife appears to act as an unexpected solder mask due to cupping (jury still out on this).
8. The cutter (at least my Silhouette) comes with PC controllability via USB, built-in cutter compensation, excellent built in auto tracing, SVG compatibility (with the Pro update) and auto registration via index marks.
9. Post-etch checking for shorts and trace continuity appear to indicate this method requires less "fixing" than our Accurate360 mechanical mill etcher.
10. Less noise than mill etcher - cutting is fairly quiet, drilling is noisy.
The downsides (so far) are:
1. This is a new technique so bugs are being found. The up side is solutions are being suggested, discovered and applied.
2. The berm of copper left on both sides of the knife cut could cause a later short. I'm looking at ways to mitigate this.
3. Carbide blades wear out - but about the same "mileage" we get with bits on our Accureate360.
4. The Y-axis board movement relies on friction but, so far, doesn't seem to be a problem.
Left to be done (known so far):
1. Reduce minimum trace width.
2. Discover method of front2back registration for double sided PCBs.
3. Automate the workflow.
4. Minimize risk of trace shorting.
Thanks again for comments & suggestions.
Mike
--- In Homebrew_PCBs@yahoogroups.com, David Mitchell <gossiphog@...> wrote:
>
> Getting the Cricut machine to talk HPGL is a project I was working on
> for several months last year. I based my work on an extension of the
> "Freecut" firmware written for the Cricut Personal.
>
> (see https://github.com/Arlet/Freecut)
>
> I've ported it to the Cricut Create, and it does indeed speak a few
> HPGL-like commands, but the place where I've gotten stuck is with
> flow-control. My plan was to support the commands that other vinyl
> cutters support (like the Black Cat, Roland, or US Cutter). This
> firmware does require re-flashing the unit via ICSP programmer
> connected through the cartridge port, which also irretrievably
> overwrites the original firmware.
>
> The Cricut has a FTDI chip on it providing the USB interface to the PC
> which (from my understand) transparently converts XON/XOFF signals
> from the onboard processor into USB flow control signals that the PC
> obeys.
>
> Unfortunately, the serial library I used (Fleury) does not support
> software flow control, so unless you type your commands very slowly
> into a terminal window, the machine skips most of them. The hardware
> flow control lines on the FTDI are not connected to the Cricut's
> processor, so this is not an option.
>
> I started a private google group a while back with the purpose of
> collaborating on it, but unfortunately noone else has taken up the
> project since I got bored of it and moved on. I would love to find
> someone with more motivation and a little better C skill than I to
> finally solve the flow control problem. (see
> https://groups.google.com/d/forum/ccfwhack).
>
> Another idea I had that might be simpler to implement is a PC
> parallel-port adapter to Cricut cartridge adapter cable, sending
> encoded stepper control signals directly from PC to cricut via 1 mbps
> SPI, and a thin firmware on the cricut that would obey those signals.
> The PC would run something like LinuxCNC, using a custom configuration
> for the parallel port driver.
>
> I don't believe the USB interface is capable of fast enough throughput
> to handle direct stepper phase control, but step-direction commands
> might work. (flow control would still be an issue).
>
> Here's a page linking to photos of the Cricut Personal, Expression,
> and cartridge hardware. They control the stepper windings directly
> rather than with dedicated stepper driver hardware. Cutting pressure
> is via PWM to a voice coil in the cutter solenoid.
>
> http://www.built-to-spec.com/cricutwiki/index.php?title=Main_Page
>
> I have several Expressions as well, which have the same processor as
> the Create, but will need some additional code changes to support the
> larger screen, additional keyboard buttons, and larger cutting area.
>
> Anyhow, if you're seriously interested, join the list above. I can
> potentially provide some dev hardware as well.
>
> -David
>
>
> On Fri, Jun 22, 2012 at 7:43 AM, jeremy youngs <jcyoungs76@...> wrote:
> > linuxcnc outputs step/dir directly and is configurable and needs no signal
> > conditioning before a l293/97 etc. conditioning or coding is only necessary
> > for microprocessors. i have three sla7086 printer stepper setting on my
> > bench now that run just fine
> >
> > On Fri, Jun 22, 2012 at 7:36 AM, Tony Smith <ajsmith1968@...> wrote:
> >
> >> ∗∗
> >>
> >>
> >> > LinuxCNC (or anything else for that matter) won't talk to the CriCut,
> >> you'd have
> >> > to add your drivers and bypass all the electronics. Most of these things
> >> use some
> >> > form of HPGL, or mangled a bit to provide security (presumably that's how
> >> it
> >> > was reverse engineered).
> >> >
> >> > not necessarily true some printers run a simpl l293, 278 or 298 that will
> >> directly
> >> > communicate with lcnc, i know ive built them, however if your machine is
> >> > operated via microprocessor without the source code it will not work as
> >> you
> >> > say. This is why i said the pinout was necessary as if it has enable
> >> input
> >> , or bus
> >> > line it probably will not work, if it has simple step and direction
> >> inputs
> >> as per the
> >> > ic mebtioned above I could make it work. I say try it the only thing he
> >> has to lose
> >> > is some time.
> >>
> >> Well, what I meant was there will be some electronics before the stepper
> >> drivers (293 etc), that converts the HPGL (or whatever) to Step &
> >> Direction.
> >> Some use weird driver chips simply replacing the lot is easier than
> >> figuring
> >> it out.
> >>
> >> You can't just plug it in as it and expect it to work.
> >>
> >> Tony
> >>
> >>
> >>
> >
> >
> >
> > --
> > jeremy youngs
> >
> >
> > [Non-text portions of this message have been removed]
> >
> >
> >
> > ------------------------------------
> >
> > Be sure to visit the group home and check for new Links, Files, and Photos:
> > http://groups.yahoo.com/group/Homebrew_PCBsYahoo! Groups Links
> >
> >
> >
>