Zach,
the belt drives are extremely accurate much more so than screw
driven machines(no backlash or wear) and they are very fast the feed
rates can be set amazingly high. We cut lots of parts on a JWX-10
with the ProtoWizard fixture and they come out beautifully.
Regards,
James McMurray
David,
Actually George is my last name, an easy mistake to make when you
have two first names 
Didn’t mean to ruffle your feathers, I was merely pointing out that
there are other solutions available, and was it not meant to be
specifically directed at desk top solutions, but CNC machines in
general.
Flashcut, in many cases, is the default software that runs most Modelmaster machines, Sherlines, and the Revos. It is, as I said, the default standard for most small table top milling machines. Another common package is Mach 3. The manufacturers of each of these machines has most likely done some customizing of the chip that runs their machine and program
Which is why I made this comment shown below in my previous post, :-
So nothing more, nothing less may very well apply to desk top
machining, but it does not apply to all machine tools.
I still stand behind the fact, that it does not apply to all machine
tools. Not every machine tool uses Flashcut or the Mach 3 as a
solution, mine certainly does not.
Just for the hell of it I opened up my cutter compensation screen and will post the contents heRe:
Lookahead Buffer Size: 9999 moves Join Tolerance:.0200mm Preserve continiuity of Closed shapes Eliminate Moves that Cause Gouges
Think beyond desktop for a minute, and think of the following as
general CNC
The buffer size is one thing, but how fast does the control, process
the data is the real question?.
There are two main factors that need to be considered with look
ahead, and the first being the Block Processing Time, and the second
is the Block Machining Time. Both go hand in hand. The relationship
between Block Processing Time and Block Machining Time is critical
for maintaining smooth, fluid, fast and accurate surface
interpretation in 3D machining. If the Block Processing Time is
shorter than the Block Machining Time, the machine will run smoothly,
however if it is the other way around, the control is not ready to
give the servos new instructions and it will stall whilst waiting for
new If a line of code has a 1" move, then the processor
has plenty of time to process the next line, however, when a move
is.0005" well that just happened 2,000 times faster, therefore, to
maintain streaming data, the processor needs to process 2,000 times
more data in the same amount of time, to keep up with the machine to
cover a 1" distance. High speed processors/controllers are essential.
I am quite sure, that you would agree, that in watching your machine
running a 3D job, that in some areas the machine will accelerate and
in others decelerate and the areas of acceleration have different
characteristics to the areas that require deceleration. This could be
different types of slopes with more or less points, or it could be
attributed to the toolpath having to make a sudden move in another
direction.
Would a fair question be, why are you seeing deceleration?
No need to answer that, but you will better understand where I am
coming from regarding high speed controllers, if you have a look at
this article.
http://www.mmsonline.com/articles/039603.html
I think that’s enough on that point for now.
I would say that the other machines using Flashcut would also have similar settings. So this would put all of these small mills using this software into basically the same class and ability as to produce similar work output using the same cutter. If there is a difference it really should be minimal.
Agreed.
The cutter can not produce a better quality than it is actually capable of.
Agreed, and in addition, an optimized toolpath can maximize the
capabilities of a tool, and therefore become a more efficient entity
designed to do what it does best. A consistent chip load and feed
rate will serve the tool better than a tool being accelerated and
decelerated.
It is true that a crappy controller or a crappy software could make a difference...but most likely you are going to have to look really really close to find it.
Ok, I’ll give you that one, because of the size of the components
you are dealing with. I will add, that as items get larger it will
become more apparent.
The output I got on my Sherline and the output I get from my Modelmaster are very very similar. They both were Flashcut based chips, the Sherline was the Lightning chip, and my modelmaster is a Platinum chip. The ouput is still basically the same. I am talking wax milling here and not other larger machines. On these I bet you could see and find a difference, especially with larger tools and different stepover rates. But in wax machining and the machines we usually use, I really don't think you will be able to see a difference in the physical output on the wax itself. Machining time...It could make a difference.
As I mentioned before, it was not my intention to put you on the
defensive, as both machines and both controllers you mentioned do a
hell of a job. Again, I was pointing out something that may have some
educational value for you. Is it useful to you?, I don’t know, all I
can say is that the more I have at my disposal the
better my machining practices become.
Best Regards.
Neil George
954-572-5829
Rick,
Maybe if you have an intelligent tool changer....
Which I do. 20 tools in fact.
Sumner's orchid carving work is quite detailed, and, I would guess, hard to machine.
I am familiar with his work, and I must say he does fantastic work.
I often go to his site to see what is new. Personally, some of his
other work has more of a wow impact than the orchids do for me, but
that’s just a matter of preference. Never was one that paid attention
to flowers, unless I had to get a bunch for the wife:-) Still, he is
an incredible artist, no doubt about it.
Not likely even in 5 axis. The undercuts, reverse curves, thin areas and textures are quite a challenge- as are Daniel's (Race Car) ability to cast and finish them.
If the items are a one piece carving, then I could see where that
could pose a problem and quite rightly impossible. But each entity
done as an individual item and later assembled either in the wax or
with good engineering in the metal once cast, I really don’t see how
it cannot be done with 5-axis, and in reality, quite possibly in 4th
axis also. Depending on the areas of under cut and how extreme it is,
the tool could get into position and accomplish the task. Think of 5
axis turbine blade machining. I would imagine with the years of
experience you have Rick utilizing CAD/CAM, I am confident that you
of all people, could get a high percentage of the job done in
individual pieces, even in 3 axis, with further hand work to attack
the undercuts. As we all know, it’s a question of knowing where
CAD/CAM begins and CAD/CAM ends. That was a compliment by the way
I would love to be able to model them in ArtCAM but I really don't think the tools are quite there.
The tools are out there, however, I personally wouldn’t have the
patience to deal with it anymore at least at the CAD level. The CAM
level, well that’s another matter.
Andrew Werby likes the Sensable solution. I personally haven’t tried
it, because my sculpting abilities are zero to none, therefore my
orchids would look like mash potatoes
Having acknowledged my lack of skill in using such a tool, a guy
came to my shop by the name of Sebastian Pengue, and showed me some
of his work done with the sensable solution. I have to say, the level
of sophistication in truly representing life like figures was
awesome. Granted he is a sculptor by trade, and as a sculptor/artist,
he has more of an ability than I could ever have in understanding
human form and other objects of life itself, and his work even though
in digital format, truly reflected his abilities. Very Impressive.
Sumner spends weeks carving one wax, and the time and effort shows. RP will get there, but it won't spoil the pleasure of carving or constructing by hand really fine pieces of jewelry. Some people will probably derive the same pleasure from modeling them with software- we all love a good challenge.
Starting out with a Roland PNC 3000 and two Sanders prototyping
machines back in 1995, it has been a long journey. It seems so long
ago, it’s not even funny how time has flown. I remember Beta testing
Rhino before it even hit the market, and not to mention running the
Sanders, when most jewellers still thought a fax machine was cool
My challenge these days has moved away from CAD per say, and more
into the CAM. For the last 2 years, my passion and areas of interest
is machine performance, and just how fast can a machine handle the
job, or more so, how far can I push it. Through constant tweaking, I
am constantly evaluating new tools to speed things up, especially in
3D applications. An example of tooling choice is a 2.5D application
that I used to run with Garr tools. At the time, I would take a 1’’
Garr cutter, and feed that at 50 IPM with a full diameter depth of
cut with a radial axial load of 30%, meaning, taking a.300" cut off
of the side at a 1" depth of cut on a profile. Switching to Benchmark
tools, I can take that same cut with a better finish at 225 IPM with
the same RPM in Aluminum. Very cool to see it action. To know
translate those gains and capabilities from 2.5D to 3D moves, a new
obsession or passion for high speed controllers emerged. PCT tooling
is another area of great interest. These are particle metallurgy
endmills, and boy, do they kick butt also. Not quite as fast, but
1/5 of the price that a 1" carbide would be, and less prone to
chipping too.
Always a pleasure Rick.
Best Regards.
Neil George
954-572-5829
Neil
I am really not trying to be defensive about the look ahead
discussion. I totally agree with you on that aspect. I was basically
trying to point out that most of these controllers have been worked
over to squeeze all the horsepower out of them that they can produce.
They really are very accurate when you stop and think about it. And
as you said all these small machine controllers really do an
excellent job in turning out quality work. And other than speed and
fixtures and a few strategys there is not much difference among the
various machines that are currently in use for wax work. Now if you
are talking metal that’s a whole different story.
What type of machine do you have? What controller software and what
are you using for toolpathing software?
Protowizard software is definately optimized as it maintains the
same speed for a great period of time. If you sit and watch the
screen you will see the speed much more constant than say Deskproto.
Deskproto has much larger files than Protowizard and usually longer
cutting times. A really good toolpathing software that optimizes the
toolpath is a BIG help. It sure can reduce the cutting times, All
those speed ramping situations sure can take up a lot of time.
Regarding the Roland machines. They basically are an inkjet printer
on steroids, technology wise. I guess that could also be said about
the RP machines too. Are they accurate? Can your inkjet printer place
a period (which is actually several dots) anywhere it is told to, on
an 8.5 x 11 inch sheet of paper? It sure can…and do it over and
over again anytime it wants to. So I would have to say that they are
very accurate. For the money, the Roland machine with the Protowizard
software and fixtures is an excellent choice. The Roland, though, is
NOT the only machine out there that will consistantly produce
accurate parts as someone claimed here.
All of these small machines are equally capable of high quality
output. I guess the basic difference among the several machines is,
speed, cost, tooling and software that optimizes the toolpath better
than another brand and the nut behind the wheel.
David
David,
I am really not trying to be defensive about the look ahead discussion. I totally agree with you on that aspect. I was basically trying to point out that most of these controllers have been worked over to squeeze all the horsepower out of them that they can produce. They really are very accurate when you stop and think about it. And as you said all these small machine controllers really do an excellent job in turning out quality work. And other than speed and fixtures and a few strategys there is not much difference among the various machines that are currently in use for wax work. Now if you are talking metal that's a whole different story.
Point taken.
The point that I was making in comparing other controllers to the
ones that you are familiar with is this. And it is meant as a source
of
The look ahead buffer that the flashcut controller uses, is not the
same as the look ahead I am referring to on a high speed controller.
The look ahead on the flashcut solution, via RS-232 and DNC, looks
for as much as it can possibly store in the buffer, to
keep the data in the buffer/controller ahead of what the machine
needs. In realty, look ahead in this situation, is more of a coined
phrase that someone came up with for the buffer, meaning that the
buffer had sufficient data to eliminate servo starvation, therefore
it is not a true independent or intelligent function.
In a long straight line move, the machine has the ability to
accelerate from the start point to the end point and actually hit the
programmed feed rate and the controller has the next line ready. Once
the machine has essentially come to the end of that block or
programmed line, it needs the next line to know what it needs to do
next. I know, pretty obvious, but bear with me. However, when a
surface is being machined, same philosophy applies, but it gets a
little more complicated, whereby, the surface normals are offset by
a distance governed by the tool diameter, therefore, the offset would
be half that amount or rather equal to the tool radius. Now Chordal
Deviation comes into play.What this means, is that the surface
curvature is split into a series of linear segments, and a
theoretical curve approximates the surface. Chordal Deviation is the
maximum distance between the Chord and the curve with the chord being
a line that intersects a curve at two points. Chordal deviation will
change even with the same tolerance as the curvature increases or
decreases in size, or as a ball end mill increases or decreases in
size. Now given that a curved surface represented by Chord deviation,
will have many points, and that the number of points are determined
by the set tolerance. Less tolerance such as.001" will have fewer
points and the resulting surface will be faceted. Now obviously,
a.0001" tolerance will along that same chord deviance have 10 times
as many points, therefore resulting in a smoother surface. Now
because at a high level of chordal deviation or tolerance we have
many points, and as an example lets say it has 30 points, it
therefore, has to make 30 point to point moves to interpret the
series of moves required to define that particular chordal deviation
path. A programmed feed rate of 30 IPM in a straight line can be
reached very easily given enough distance, however, a point to point
move on a surface with very short moves can only accelerate so much.
Therefore the point to point move, can never achieve the programmed
feed rate, but will maintain the same feed velocity from point to
point until there is a longer move for it to accelerate back up to
the programmed feed rate. A.004" tip diameter pyramid tool, will have
a move of.002" or less along the chordal deviation path, which
clearly shows, that there is no hope in hell of any real
acceleration. I have seen many situations where that can slow a
machine down to 1/5th of the programmed feed rate, meaning you are
machining that surface at 6 IMP. Less points becomes faster and
closer to what you programmed, more points gets slower. Taking it
further lets say that the curved surface is tangent to a flat area,
and that the flat surface is maybe an inch long. The minute the
machine comes off the curve and on to the flat surface, you will see
acceleration. At the end it slows down, makes a step over, and comes
back at full acceleration i.e… 30 IMP, but as soon as it hits the
curved surface again, it will have slowed down back to the point to
point feed rate. Now, the interesting part. The look ahead on a high
speed controller will look ahead and identify the first 30 points,
the other two points at the end of the flat surface where it makes a
step over, and come back and will also have identified the next 30
points whilst the machine is taking care of the previously optimized
80 blocks of What happens here is still a point to point
move, however, because it already knows where the next 62 points are,
it becomes more efficient. For example, it will see the first 30
points and know that it can now keep accelerating from the first
point all the way through the first 30 points and prepare itself to
slow down just before the point to point move on the step over at the
other end of the flat surface. Therefore, except where it slows down
for the step over, it can hit full feed rate along the flat surface
and the curved surface, therefore reducing cycle times.
I hope it’s clear. As I have said many times, I understand what I am
saying, I just hope I convey it for others to understand:-)
What type of machine do you have? What controller software and what are you using for toolpathing software?
Haas machine with a Haas Controller and a High Speed Milling Package
with Optimized Look Ahead. The CAM side of things I use AlphaCam, Work
NC and Type3.
Protowizard software is definately optimized as it maintains the same speed for a great period of time. If you sit and watch the screen you will see the speed much more constant than say Deskproto. Deskproto has much larger files than Protowizard and usually longer cutting times. A really good toolpathing software that optimizes the toolpath is a BIG help. It sure can reduce the cutting times, All those speed ramping situations sure can take up a lot of time.
Absolutely. This is, in all fairness, where optimization originated
in part, however, it still required some human intervention to adjust
feed overrides up or down to try and speed things up. I know, because
before I upgraded the controller, I have manually adjusted the feed
and spindle speeds on many jobs on the fly. If the cut sounded good,
I would crank up the feed rates. The look ahead on high speed
controller can take even an optimized toolpath from a CAM system and
drastically enhance it even further.
Regarding the Roland machines. They basically are an inkjet printer on steroids, technology wise. I guess that could also be said about the RP machines too. Are they accurate? Can your inkjet printer place a period (which is actually several dots) anywhere it is told to, on an 8.5 x 11 inch sheet of paper? It sure can...and do it over and over again anytime it wants to. So I would have to say that they are very accurate. For the money, the Roland machine with the Protowizard software and fixtures is an excellent choice. The Roland, though, is NOT the only machine out there that will consistantly produce accurate parts as someone claimed here.
Actually the first machine I had was a Roland Camm 3, PNC 3000.
http://www.cadigital.com/camm3.htm
Excellent machine in its day and could feed at 1320 mm per minute or
just under 52 IMP. Pretty damn fast even by today’s standard at the
bench/tabletop level.
I talk like it’s gone, but I still have it, just don’t use it
anymore. No belts on this model, all screw driven. By the way, this
was using steppers, not that it made any difference at the time,
because I didn’t have a clue what I was dealing with anyways
All of these small machines are equally capable of high quality output. I guess the basic difference among the several machines is, speed, cost, tooling and software that optimizes the toolpath better than another brand and the nut behind the wheel.
Well said.
Best Regards.
Neil George.
954-572-5829
Neil
I sort of knew a good bit of that about high speed controllers and
points and Chords. I have read about it but you explained it a lot
better than what I have read.
I love watching those larger Haas mills, with the numerous tools,
work. It’s really neat to watch it go to the rack and change tools
and take off doing a different operation all by itself. You can
really do a lot of things with that mill. Haas has a 4 axis mill for
about 40K that they show at MJSA and tool shows that will do jewelry
work. It is a little bit of overkill for wax work but if you are
going to cut steel molds and other things this would be a really
great tool. For a small to medium sized manufacturer who does dies
and metal work this would be a good solution. I hear that some of
these used go for about 15K.
The belt driven Roland machines are still speed deamons in the wax
world. Those things really cut fast and are accurate. The
Protowizard fixtures are a great help with these machines. I do not
like the rotary ring cut solution that they use though…I prefer
using a large tube of wax rather than a friction based way of holding
the wax in place while it’s milled. Not that their way doesn’t work,
it is just a personal preference. For the money you have to consider
the Roland if you are doing a small to moderate amount of milling. I
would want something more robust if I was doing a lot of work or a
service bureau type of situation.
David