[Zeppelin1007]

Alright guys. Ive been enjoying my fla 100 build since about july/august now. Really been abusing it the past few weeks with great success.

But now its time to ramp things up and do some double sided machining.

Initially i tried as i would on a vertical mill. Square off the part, touch off, route...then flip the part, load up a new file, touch off again and go.

This worked, but things were just alittle too off to be usable.

So i read about the dowel method: Create your stock in your CAM program, have the machine drill 4 holes in the table, then the same 4 holes in your stock. Awesome. Seems to me then, i shouldnt have to cut the stock to perfect sizes this way no?

My workflow: Solidworks -> solidcam -> mach3. (love being an engineering student!)

i create the part in stock, set my coordinate system to the top left corner. Flipping the part is similar to turning a page in a book, flip horizontally 180 degrees, to which i have my new coordinate system to that top left

Once i flip the part and line up the dowels i secure it with screws. Pretty rough, but i have plenty of spoilboard.

I tried to do something simple tonight. Took a big ole' chunk of pallet wood. Did the dowel routine. Used a 3/4" router bit to "plane" my top side. Began routing 2 simple rectangles using 1/4" endmill. Went nice and smooth.

Flipped the part, DID NOT RE ZERO X AND Y, just Z. Planed. End mill. It worked, but was off ever so slightly.

Ok. Try again. This time hit "regen toolpath" and found out the hard way that mach 3 moves it to the original top left corner, so NOTHING at all lined up. Was that the problem?

Thinking this was my error. I tried again without hitting regen toolpath. It was close, but off an amount in the x an y axis to make it unusable.

So thats my software lineup, what did i do wrong?

Now for the possible mechanical issues.


Im cleaning this machine like crazy. Dust. What have you guys done on an FLA 100 to keep that X axis clean? Other than shop vac. I have one, i havent made a cyclone/dust shoe yet. Was thinking of making "walls" to go on the sides. Or, would simply overhanging the table work?

Next. Part of this i suspect backlash or missed steps. I say suspect because ive been happily cutting one sided for ages now, only tonight when using different router bits would i get the x axis being off slightly. How do you guys lubricate this machine? i ask because the backlash nuts are plastic. I know certain oils will not help my mission and eat my backlash nut!

So i learned ALOT tonight about toolchanges, zeroing, general feedrates and such. Very productive evening i must say.

Anybody have any info? i know. Theres many variables going on here and im trying to eliminate what i can. Maybe someone can tell me my software mistake? I cleaned the heck out of this thing tonight before shutting it down. Nothing was gummed up, just dust here n there. My floor and walls on the other hand lol....

Thanks in advance guys!

[vtx1029]

When I do 2 sided operations I put my X0Y0 in the center of the part it takes the confusion out of the equation for me. I also mark both sides of the part on what side is what and how it should be located when flipped. I did one double side operation and flipped it the wrong way once. I also use the dowel method and have had decent success, but I just drill thru the stock into the spoil board when I'm doing the first side operation (that way there is no need to re zero X and Y again).

Also a dust shoe is almost mandatory! It will greatly reduce dust on your rails.

[ger21]

Regen Toolpath shouldn't be changing your coordinate system, so I'm not exactly sure what you're doing there??

I'd also recommend setting X or Y zero in the center, rather than the corner.

You said you were off ever so slightly? How tight were the dowel holes? Wood has some give, so even with dowels, I'd say that it's possible for there to be a slight misalignment when flipping.

How are you zeroing your axis? Did you re-zero after drilling the holes?

[fizzissist]

You should NOT have to use the center of the part for X0Y0, what you MUST do is allow for the Zero to be appropriate for the part when it's flipped in the 2nd op program.
For example, when fixturing with a vise with a fixed jaw (let's say that it is fixed in Y), you'll locate/clamp against that jaw and a stop. When flipped, the part's X locate/stop will necessarily move to the other side.

Round parts at X0Y0 at the center (almost always), so it doesn't matter.

To sort out if it's software or hardware (my bet here is hardware,.... backlash or extra/missed steps) is to do 3 dowel holes to locate, much as you did with 4, and bore a hole thru at X0Y0. Flip the part and indicate the hole. Is it off?

If backlash is a problem you can't fix in hardware (and there's no provision in the software/parameters), you may have to compensate in your programming by always approaching from the same direction for features that must have a common accuracy.

That's the way the old school machinists have done it for eons with old, sloppy, manual mills!

[louieatienza]

You should NOT have to use the center of the part for X0Y0, what you MUST do is allow for the Zero to be appropriate for the part when it's flipped in the 2nd op program.
For example, when fixturing with a vise with a fixed jaw (let's say that it is fixed in Y), you'll locate/clamp against that jaw and a stop. When flipped, the part's X locate/stop will necessarily move to the other side.

Round parts at X0Y0 at the center (almost always), so it doesn't matter.

To sort out if it's software or hardware (my bet here is hardware,.... backlash or extra/missed steps) is to do 3 dowel holes to locate, much as you did with 4, and bore a hole thru at X0Y0. Flip the part and indicate the hole. Is it off?

If backlash is a problem you can't fix in hardware (and there's no provision in the software/parameters), you may have to compensate in your programming by always approaching from the same direction for features that must have a common accuracy.

That's the way the old school machinists have done it for eons with old, sloppy, manual mills!

I agree... If you locate 0,0 at the top left in the first op, then when you flip over you should locate 0,0 at the top right. This is, however, dependant on your clamp's or fixture's top edge normal with the X axis.

If you use dowels, you only need two in line. Use the left dowel as 0,0, then flip the part and use the right dowel as 0,0. Conversely you can pit the dowels center top, and bottom.

Yet another way would be to cut your blank to size on your machine, then make a fixture out of scrap by making a pocket that's a snig fit for the blank. Then in the same setup, do the top, flip, do the bottom. I do this a lot on parts that need counterbores or pockets on both sides and it works pretty well. This works especially well if your part is an odd shape.

If you're using 0,0 as the center you need to make sure your part is perfectly square, and you can perfectly mark the center on both sides. If an edge your referencing off of is off square your error will be double that when you flip it over. If you cut a pocket for your part and it won't fit when flipped over, then you need to adjust the squareness of your gantry relative to the table, and also check that your Z axis is precisely trammed.

[ger21]

If you're using 0,0 as the center you need to make sure your part is perfectly square

Not if your using dowels. The shape of your part shouldn't matter at all.

If I were using dowels, I'd keep my 0,0 the same for both sides, to eliminate any chance of error when re-zeroing.

[louieatienza]

Not if your using dowels. The shape of your part shouldn't matter at all.

If I were using dowels, I'd keep my 0,0 the same for both sides, to eliminate any chance of error when re-zeroing.

I should have clarified, if you're using a clamping device to hold down the work.... I do use dowels to line up both sides of my guitar bodies.

My thought originally was that the gantry might not be 100% square but re-reading, I guess unless the fit was sloppy the four locating holes wouldn't line up if the machine wasn't square. It's possible the Z axis is not 100% square, which could cause misalignment....

[smack ramen]

Your gantry could be racking slightly. I have a FLA-100 also and I did the anti racking cable mod seen here http://www.cnczone.com/forums/diy-cn...ock_solid.html

[Zeppelin1007]

Figured it all out guys!

For one, have my part fully inside my 4 dowels.

the other problem is that junk HobbyCNC board. REFUSES to cut an arc perfectly. I thought i had solved it, but apparently i only made the problem less noticeable. Any arcs have to be cut around 15-20ipm.

[BanduraMaker]

If I were using dowels, I'd keep my 0,0 the same for both sides, to eliminate any chance of error when re-zeroing.

For confusion avoidance - I totally agree.

In the end, it doesn't matter but when I machined a pretty complex 4 sided part, my brain felt like I was back in school when I was doing all of my CAM work. In the end, it doesn't mater where you put your zero so long as it's in the right place which is....where you think it is. Putting dowels along a centerline of the part avoids confusion and mistakes.

Turns out that when I did my 4 sided part I put the right left dowels along the centerline but the front back were off centerline by 1/8" or something and I didn't notice. Ran a test part and it came out perfectly. Ran the real part in more expensive stock and ruined it because I didn't pay attention to which side was the "front" and which was the "back". Had the dowel pin holes been centered, it wouldn't have mattered.

[PaulRowntree]

Figured it all out guys!

For one, have my part fully inside my 4 dowels.

the other problem is that junk HobbyCNC board. REFUSES to cut an arc perfectly. I thought i had solved it, but apparently i only made the problem less noticeable. Any arcs have to be cut around 15-20ipm.

If the shape depends on the speed then there is a good chance the constant velocity settings are at fault

[ger21]

Any arcs have to be cut around 15-20ipm.

The HobbyCNC board doesn't know when you're cutting an arc. My guess is flex in the machine.

[Zeppelin1007]

Im actually pleased to say its not flex in the machine, I've narrowed it down to the board, or, possibly Mach3 itself. Ive tried different accel/velocity/pulse widths.. CV mode on and off..for some reason i can't figure out, when cutting arcs/circles, the motor gets about half way through the arc, then slows down, then just like completely loses its direction/speed, you'll see the shaft spin erratically back and fourth!

Never fails. 37.5IPM as a feed rate triggers it like clock work. Its the worst here. I found that cutting around 50+IPM solved the issue, or so i thought. Apparently its been there all along, just at higher federates its not as noticeable on the end result, and at lower federates (15-19) its gone completely. around 20-25 it starts, depending on the direction of the arc.

Guys, this will do it with NO machine attached to motors. Exists if i swap motors, chips. Ive reflowed the board, hell, i took it to a technician and had him reflow the board on a Weller station. Ive tried different versions of mach3, I've changed computers. Changed parallel port cables. I just spent the most of the evening cutting and my arcs i had to go ridiculously slow..and I'm sick of it!

Tried Sherline mode. That seemed to make it run a bit..smoother..the chatter is still there, just smoother. I mean, it up and reverses direction! how?? Microstepping = same result. Ive had it! and it frustrating since i bought it second hand, i can't get into the support group.

Im not sure what other things to play with in mach3, so I'm going to try LinuxCNC just for the heck of it, but, yeah. i can't have this problem! Does it with Gcode from any source or the built in wizards. its something to do with slowing down/speeding up the motors in a circle/arc.

I will say if I'm cutting a part with not too much for curves/arcs, and i do the flip...its perfect! I use 0,0 in my "top left" corner for both sides. Have preset jigs on my table, that are simply bowel lines, predrilled screw holes, etc etc. I can easily come out every day. Fire up the machine, load an offset preset, load the machine and go. Minus this arc problem. With the holidays coming up, the incredibly slow rates for cutting projects is NOT an option! But thank you guys for assistance on the flipping issue. That works fantastically.

BanduraMaker, I've wondered about 4 sided machining. Any useful links/advice?

[alan_3301]

can you set your post processor for line segments instead of arcs?
I think it will slow the feed down because of the many many short line segments, but you could see if your problem still exists without g2/g3 commands.

[ger21]

Guys, this will do it with NO machine attached to motors.

What you're most likely seeing is resonance causing the motors to stall. Resonance is even more likely when the motors aren't attached to anything.
If you have dual shaft steppers, add some dampers to them. Even a hockey puck pressed onto the rear shaft can make a big difference.

[louieatienza]

It is possible that with fine leadscrews (10tpi for example) that you enter mid-band resonance zone before you get to a useful cutting speed with these less expensive driver boards and larger steppers. THis had happened to a lot of Xylotex owners who got the 425in-oz steppers and attached them to their Sherline/Taig mills. Gerry makes reference to using dampers; and there is a thread on how to make "rattler" style dampers. The person did eventually upgrade to a G540 which eliminated the need for dampers.

As to arcs... a lot will depend on your CAM as well as CAD. If you're cutting a 2.5D part, where X and Y are continuous and Z is indexed, many CAM software will translate the "arc" moves as G2/G3 commands. If however the machine moves in simultaneous 3-axis, most CAM will translate arc moves as line segments (unless you have a very high-end CAM system.) Some CAM will only do 3D arc moves with Z waterline or Z level finish strategies.

If you already hae CV tured on in Mach3, another thing to try is to change the LookAhead number in the GeneralConfig screen. I found this helped smooth machine movement when movement consisted of many line segments. It seems to be default at a measly 20 lines, but you may try experimenting with maybe 150-200 to see if it helps smooth movement.

My best guess without seeing your machine is, you probably would benefit by just switching your driver to a GeckoDrive G540, which has electronic resonance damping.

[Zeppelin1007]

Hey guys

Im using Solidcam thanks to my school. So I'm sure i can. just using the Fanuc output. But, like i said, the built in mach3 wizards will do it too.

Tried hockey pucks. No luck. :-( Just made me realize the skating rink near me is open and i should go lol.

what does work, is pulling the motor off the machine, laying it down, and pressing as hard as i can on it with the palm of my hands. For some reason, it acts fine then through the ugly parts. as soon as i release pressure, the problem comes back.

What the heck?

these are 425 oz-in keling motors but since its the hobbyCNC board, its at 305oz-in.

[ger21]

Tried hockey pucks. No luck. :-( Just made me realize the skating rink near me is open and i should go lol.

what does work, is pulling the motor off the machine, laying it down, and pressing as hard as i can on it with the palm of my hands. For some reason, it acts fine then through the ugly parts. as soon as i release pressure, the problem comes back.

What the heck?

Sounds 100% like resonance. Hockey pucks don't work for everyone, but the rattler type should make a huge difference.
On my Xylotex powered router, it was almost unusable until I added the dampers, now it's perfectly smooth at all rpm. without dampers, it only ran well in a very limited rpm range.

[Zeppelin1007]

lol knew it was the hobby cnc.

So i did a google search on these dampers. Neat. Though many say "eliminate the problem at the board" my question is, is how? Whats causing this? wonder if i can get someone with a digital scope over...

[Zeppelin1007]

i wonder how hard drive platters would work as a damper lol...