Re: [CAD_CAM_EDM_DRO] 3D Contouring - Loss of Accuracy

Boy Jon, Great Stuff !!! I'm keeping it handy.

Peter, I found Jon's writeup very interesting. This is the first time I
have heard that the Gecko's indeed require something unique. You are not
alone with Flashcut and Geckos losing a small amount ONLY on 3d files. I
can assure you that Flashcut would not be where they are today IF it was
some serious bug in the Controller software.....

One thing I noticed is that you had the same problems with
IndexerLPT.... I experienced this too years ago. It turned out that at
that time, and I am not sure what they do now, there was NO adjustments
for step pulse width or polarity. According to Indexers old Site, they
had made the statement that their controller sends out a signal that
will work with "Industry Standard Negative edge trigger" drivers. For a
long time, I could not figure out why I lost position.... I did when I
realized the problem was my MicroK drivers were "positive edge" ! I came
to find out the REAL industry standard just might have been opposite of
that statement.
Other than that, I had admired Indexers "motion"......

To take this issue further, Indexer always sold Anahiem Drivers with
their stuff. Anahiem drivers worked excellent with Indexer as I have had
experience with a few retrofits to machies that had Anahiems. Some of
the retro's were an upgrade to newer IndexerLPT software, but get this,
Two of them were also upgrades to FlashCut. The Flashcut had NO problem
running those drivers on two different machines!

When I moved my machine over from Indexer to FlashCut, I wrestled with
Step loss again. MicroK DM 4050's required a positive edge trigger and a
whopping 50uS of pulse width. Swapping the Opto's for transistors as
many of us have done helped quite a bit. But, it was only After I
actually ventured into really Big pulse width numbers in Flashcut,
things came around and step loss was a thing of the past. I was
reluctant to live with such high width settings at first because I could
get it to run much faster with lower width settings. But, those lower
settings allowed it to lose position just a smidget.

So, because of the super rediculously wide pulse width requirement,
"Speed" was always limited, as the pulse width took time away from the
Signal Generator actually being able to keep up to how fast I wanted to
go. In the case of the MicroK's, I couldn't get them to run at 50uS as
advertised either, I had to move all the way up to 65uS!! This is not at
all like most Drivers available today, most firing easily with under
1uS. The DM4050's would be wonderful for a mill, but, I had it on a
Router. Faster was always a goal. The drivers were too slow.

One day, I was back "tweakin" FlashCuts motor signals. I found that by
incorrectly setting my pulse polarity from positive to negative, and
then dropping my pulse width from 65uS to 7uS, I regained all of the
speed losses I had to live with. I ALSO did not have any step loss !!! I
am still wondering just how that could be the case, but it indeed worked
excellent with all the wrong settings. You may want to try some motor
signal settings you have never attempted. You might get a suprise !

Now maybe someone is familiar with the Drivers that Ah-ha throws in
their 6 amp box. I had put an Ah-ha system on a Router after a deal went
bad on a brand new Major manufacturers machine. They had designed a new
control and could not make it work. This was back before Flashcut
existed..... In fact before any other PC controls were available under
$4000.... After I stuck Flashcut on another machine, The Ah-ha was short
lived because I found I could get much more speed out of Flashcut even
before they had contouring, I could finally throw floppy disks where
they belong (Trash), and I could train any operator in minutes how to
run routine jobs. Flashcut had NO problems running those drives either.
Rock solid, bash stepper systems all you want, reliable, trustworthy
service I could walk away from, even go home and let it stop itself when
done. Within the same year and a lot of Testing beta "crash and Burn"
releases from FlashCut as they explored Continuous motion, we had a very
stable version with continuous that equally ran those drives without flaws.

Think I'm done yackin' yet ?? Nope !

A few weeks ago, I fought with positional error on one axis under a beta
release of Flachcuts V2, but ONLY when running 3d files. 2.5d work
always came back to my switches and the dial indicators I attach while
beta testing. I tried numerous Chip replacements. I could get the
drivers to work without loss, but only with those same high pulse widths
I knew worked well, but, NOT the "wrong" settings...... So, I had lost
my faster speeds I was so familiar with. Frustrated, I finally pulled
the old MicroK's and stuck in a set of API's I picked up when they were
a bargain. NOW I wish I would have bought 10 of them !!!

Flashcut V2 works very well with those drives, but V1 does not. Wierd
stuff eh ? Under V2, Things have really come around. Excellent multi
axis motion. Smoooth !

So, if you happen to have a beta release of a recent FC V2, you will
notice that the whole Gecko, Misc. driver, loss of steps issue is being
chased down. Newer settings are in place to adjust polarity, width, AND
leading/trailing edge. Of course some of this adjustment is only useable
with the right internal chip, and, I do not have Gecko's to even get
involved in tracking the problem down. Nonetheless, the whole issue is
being researched. For some really wierd reason, Flashcut has a tendency
to work best with the really expensive drivers ! I do not know what that
means, I can only tell you that that is my experience so far !

So, Peter, hang tight ! Your not alone with this issue. I friend of mine
just south of Chicago has the very same problem with Geckos and FC.
Maybe he is close enough that one of the Flashcut reps can go to his
place and play till they learn just what it is that baffles even the
best in regards some of this step loss under 3d motion. Fortunately, I
am not having it, nor do I think anyone with really expensive drivers
are ! :-)

Chris L

>
>