[Coco] [coco] Coco CNC

[Gene Heskett]

On Sunday 10 February 2008, George Ramsower wrote:
>From: "Gene Heskett"
>
>> On Saturday 09 February 2008, George Ramsower wrote:
>>>I did my first fully automated CNC machining this morning. I finished
>>>cutting the groove for the X axis lead screw into the X axis table. It
>>>worked but with a problem with (I think) the Dremmel. Twice, the end mill
>>>began dropping down into the part..slowly.
>>
>> I've had several occasions where a bit would ever so slowly walk out of
>> the
>> dremel's collet.  They could stand to have a better tightening mechanism.
>>
>>>However, I don't think it is
>>>moving inside the chuck. I think perhaps the dremmel is moving in the
>>>clamps
>>>I made, since they are simply pressed onto the dremmel. It might be also
>>>the
>>>output shaft on the dremmel that is moving. I'll have to look into this.
>>>It
>>>didn't hurt anything as I'm only making room for the lead screw and
>>>nothing
>>>is supposed to come in contact with this groove. It only makes room for
>>>the
>>>screw.
>>> But this issue will need to be dealt with before actually trying to make
>>>parts.
>
> I was thinking of taking the chuck off of it and putting a collet on to see
>if that may help.

IMO that chuck is a disaster with a ticket to happen.  Been there, done that, 
had to find the bit where it flew off into the grass while carving wood with 
it and one of those blue colored carbide rasp bits Lowes has.

>But as I thought on it, I realized that it would not then 
>be able to reach the slot I'm working on. The chuck does extend the length
>of the dremmel.

Humm, can the Dremel be lowered in the clamps?

>Other than that, this machine worked as well as I expected. 
>The only weak point is the dremmel, as I knew this when I began this
>project. This is why I said in an earlier post that the dremmel may become
>only the drive motor for a new mill head. The work I did with it on the lead
>screw slot turned out pretty good.
> The side to side milling had issues, because the Z axis was extended far
>beyond safe limits because it was actally milling below the slide surface,
>which in turn began 1/4 inch below the designed work surface. So flexing in
>the whole affair is considerable. I'll have to go really slow and make small
>cuts to make the pocket for the lead nut.
> When the Z axis began to fall, I discovered that the machine can cut a lot
>deeper and with more tool pressure than I thought it could do. It actually
>made a pretty decent looking cut. Hmmmm.
> The B09 program is working quite well. However, I do need to get the thing
>to use an external file to tell it what to do. This would reduce the amount
>of disk space used quite a bit. I expect once the thing is totally working
>as intended, the program itself may push the limits of the available memory
>for the window. Passing params to smaller processes  are an option, but
>right now, I'm avoiding it. Even keypresses halt the process and this short
>halt can be seen in the part while the machine is cutting.

The passing of params is how emc does it.  Its main loop, running under the 
rtai kernel extensions, runs every 30 or so microseconds on my machine, 
driven by nmi timers.  The real time module itself takes over the hardware, 
and runs linux with the leftover time, not vice versa.  This timing loop must 
be steady as a clock crystal else the steps issued to the motor have so much 
timing jitter that the motor can lose steps.  Very important when the motor 
is moving fast enough to be a ballistic object.

The coco can't run that fast of course as it probably couldn't add a motor 
output stage to the hsync IRQ, not enough time and the vsync IRQ is too slow 
but would do for development of the routine.  I believe there is a timer in 
the gime that could be used for this however. I believe it can count either e 
clocks, or hsyncs to set its speed.  If it could count say 50 hsyncs and fire 
off your the motor output loop it might be more usable.  Get that working to 
drive the motor steadily, in any combination of simultainious motions and 
then reduce that count to gain maximum speed until the keyboard response or 
what ever video tally for position you might cobble up starts to get 
noticeably sluggish.

With emc, I have a real time display of both the axis positions, and a graphic 
showing a rotatable wire frame of the working part and bit position, updated 
about 10x per second.  Nice.

The disadvantage of this method is that at the maximum speed of the motor, the 
speed changes are discreet, and obvious, as adding another IRQ period to slow 
the motor is a large change in speed.  Adding another count to the countdown 
when the count is several hundred already is not an obvious change in speed 
so the slow stuff can be very precisely done.  I can turn motors so slow you 
have to feel them to verify its moving. 0.0001" a minute? Yes. Useless unless 
grinding mirrors or??

Unforch, at low counts, the speed change gets radical just when the motor is 
running out of torque due to its inductance.  This is why we use external 
step/dir type drivers that chopper regulate the 1 or 2 volt rated motor to 
say 2.2 amps, while running on a much higher voltage supply, in my case about 
27 volts.  Many years ago, before such chopper stabilized drivers were 
available, the supply voltages to drive the motors were often in the 80-120 
volt range and the currents were limited by big multi-kilowatt banks of 
resistors.  Big machinery can move fast, 2500 ipm on a few of them, but for 
us and our toys, I can move at 10 ipm with very high power, or at 30 ipm if 
everything is free, but it cannot do that over the full range of my x table 
as its gibs bind when the motor end (heavy) is hanging way out.  So I limit 
it in the sw to about 17 ipm.  That's still fast for you, I have an xp-1400 
running at 1600mhz so I have a lot more cpu cycles to play with.

You can simplify your driver considerably by going to a step/dir 2 bit output 
per motor, and there are several usable, free except for the parts, designs 
available on the internet.  I played with one such low cost kit but couldn't 
get the regulator chopper to work right, it chopped at about 100 khz and that 
overheated the L-298, should have been 20khz, so I gave up on that particular 
vendors kit and solved my driver, and puny motor problems with a 3 
axis 'xylotex' kit with 3 262 oz/in motors for just under $300, or a hair 
over $400 for a 4 axis kit with 430 oz/in motors, but while those 430's are 
stronger, they are also slower unless the voltages are stepped up but the 
limit for that driver is about 30 volts before the smoke breaks out. It uses 
the Allegro A3977 driver chip.  Gecko drives can do 80 volts but are about 
$100 per axis just for the driver.

I have the 4 axis xylotex kit on hand now, and will use one of those motors on 
my z axis to get additional push for drilling holes, and put the takeout 262 
on my 4" rotary table, which I will probably use for sharpening end mills in 
addition of hobbing gears or sprockets for bigger stuff.   That motor mount 
for the rotary table is 'in progress' now, but I had to stop and make a 
larger boring bar, which is done except for the slot cut all the way through 
the 2" sq steel block from the back to the 5/8" hole so when the hold down 
bolt is tightened, the block is closed onto the 5/8" bar.  I may get that 
done today if I quit messing around and write the code to let the machine do 
it all, a 10 minute job if I just modify the routine I used to cut the bottom 
flat on the holder block, switching the character x to z in those commands 
that drove the x, and switching the y to x to switch the cut direction, of 
the 30 line routine I used for the flat top and adjusting the finished depth.  
I'll have to rig the air nozzle for swarf clearance, but that's a minor item.

This may be more bar than my carriage can handle, I see I need to adjust the 
way hold-downs on both front and rear, it was moving slightly sideways on the 
front v-way last night as I was using the head stock motor to drive the drill 
bit to align bore the block.  Hopefully the bore will be accurate enough. :(  
Milling that slot, about 1.3" deep, is going to be fun once I'm past the 3/4" 
mark.  Cutting going both ways, I'm hoping the mills flexibility will make 
the cut enough wider to clear the shank of the end mill when its deeper in 
the groove than it has flutes.

Its snowing here now, about an inch since I sat down here...  Got to go fix 
some breakfast & git to the day.

> I need to create a new boot disk to take advantage of the no-halt
>controller I have in this coco. I've just been lazy on that.

Shoemakers kids, version 391?

> I have a 6309 and have been thinking of installing it and converting to
>Nitros9 on this machine. It's a darned site easier than doing it on the
>other coco.

Git-r-done!

-- 
Cheers George, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Every time you manage to close the door on Reality, it comes in through the
window.