DWX50 skipping the fine milling stage

[CoolHandLuke]

k, so we've got a problem here. at first we though it was a CAD issue but now it seems to be cropping up at random times, for no apparent reason; we haven't changed any procedures for loading the machine with items to mill.

the mill will do the 2mm cut and then just jump to the next item and never finish what it started. some of them it finishes, some of them it mills only one side start to finish...

somehow the mill is skipping steps and it us ruining a lot of work. not the least of which is MOST of the Izir we are trying to do. sometimes on singles too however, and copings which is what tells me it happens at random times.

anyone have help to offer? we got a second roland dwx50 but the problem hasn't cleared. we've re-installed sum3d, even done rollbacks to older cam versions but nothing has been a permanent solution.

 

[Sevan P]

Luke did you try to do a preview of how it will mill? after you calculate do not send it to the machine click on the strategy line you want to preview and right click you should see simulate with material removal. See what happens there, it should show you exactly how it is going to cut it. You can increase and decrease speed by the + - buttons if I recall right. Sound like data is missing somewhere in the strategy or the calculation. Did you do a start or send style calculation?

 

[CoolHandLuke]

we send the calculations and all appears normal, we see each stage being sent and processed, and it works for most of the files, but sometimes some of them will only do the rough mill stage.

calibrate every Monday 9am regardless of workload (nearly 30 per day).

tech support last week walked through the preview stage with us and all appeared as normal. they said they never saw anything like this.

 

[zero_zero]

Try reinstalling the v-panel as well. Maybe its cropping the g-code whats being feed to it by the CAM.

Sent from my SGH-I747M using Tapatalk 2

 

[Sevan P]

On big cases I like to calculate them send one large file to the milling unit, assure calculation completed. Maybe a bunk stl file as well, I have had that happen as well. I broke wax burs because part of the calculation was missing. Weird stuff will happen.

 

[Glenn Kennedy]

Luke,

If the mill is skipping steps then it might be a communication issue between the mill and the PC. Make sure you are using the original USB cable that came with the machine, that it is tethered to the machine with the white clip mounted near the USB port (I have seen machines "hula" the cord loose). You are connected directly to the PC. No hubs or USB extenders. Make sure the PC is not doing other things like virus scans, backups, calculating toolpaths, streaming videos, etc. As mentioned above you might want to write the file to a .prn and send that to the machine from v-panel. This will let the PC focus on tool path generation for the entire job rather than tool path generation and file spooling.

Make sure you have the newest version of v-panel installed and the newest driver. It is always a good idea to keep both of these up to date. Both can be downloaded from the support section of our website (from the home page the support link is on the top right side). Roland DGA

If you still have problems contact Roland DGA technical support. You will need your machine serial number to open a case. You can get the s/n off the sticker on the back of the mill or from the v-panel.

 

[technician]

k, so we've got a problem here. at first we though it was a CAD issue but now it seems to be cropping up at random times, for no apparent reason; we haven't changed any procedures for loading the machine with items to mill.

the mill will do the 2mm cut and then just jump to the next item and never finish what it started. some of them it finishes, some of them it mills only one side start to finish...

somehow the mill is skipping steps and it us ruining a lot of work. not the least of which is MOST of the Izir we are trying to do. sometimes on singles too however, and copings which is what tells me it happens at random times.

anyone have help to offer? we got a second roland dwx50 but the problem hasn't cleared. we've re-installed sum3d, even done rollbacks to older cam versions but nothing has been a permanent solution.

Today I had the exact same problem. Everything looked good but it must have skipped one tool path but not on all crowns?
It was only at the inside top of some crowns were you could clearly see that it was not milled all the way.
So as Glenn mentioned I did actually changed the USB cable because the original was really long. I'll changed it back.

 

[Sevan P]

Today I had the exact same problem. Everything looked good but it must have skipped one tool path but not on all crowns?
It was only at the inside top of some crowns were you could clearly see that it was not milled all the way.
So as Glenn mentioned I did actually changed the USB cable because the original was really long. I'll changed it back.

If the above problem is true, on a USB cable the MAX length is 15' and no longer.

You would think that mill mfg's would switch over to cat5 or cat6 Ethernet cable allowing the user to put the mill farther away from the pc if that's how they want their setup to be.

Data loss and be the devil if overlooked.

 

[mir0]

Guys - most likely the problems you are having are the data corruption errors caused by your PCs during CAM tool paths calculations.
Small little (uncorrected) error or skipped bit and the whole calculated "thing" is just GARBAGE.

CAM tool paths calculation is quite intensive and resource demanding process - it needs REAL WORKSTATION PC
with REAL workstation components to run reliably and give you consistent results.

If you are using "regular over the counter PC" or even a "powerful gaming PC" -
and it was working OK for you - it is just pure LUCK - and it will end soon or later.
And when it does - you will loose more money (and time) then if you've invested in REAL workstation PC in the first place.

Personally I do NOT like buying ready made WORKSTATION PCs from big manufacturers like DELL, HP or Lenovo
because they use proprietary components and you are locked to their service and parts if you need one.

I prefer to custom build my own PCs from the best universally available components - much easier to service them later on.

Here are a few "MUST HAVE" when building or shopping for one:

• - professional Intel chipset motherboard with Intel XEON processor and ECC memory support (like: Supermicro X9SRA)
  
• - minimum of 32GB of ECC memory (Error Correcting Code)
• - minimum of 240GB SSD - Solid State Drive - Hard Drive (professional - like: Intel 520 SSD)
• - 90+ Gold or Platinium SeaSonic PSU (power supplay unit)
  
• - if your CAM software uses video card for calculation - your video card also needs to have ECC memory (like: nVidia Quadro 6000)
• - good computer case with plenty of air circulation for cooling of ALL components inside
• - Windows 7 x64 (64-bit) PRO or ULTIMATE !!!

 

[zero_zero]

All communication protocols ( 100Base-T(Cat5)/ USB / RS485 etc. ) have some sort transmission error checking implemented on hardware level. There is usually no such thing as data loss, its just slowdown.

If would be, we would see lots of garbage (corrupted files) coming through our i-net connection all the time, not to mention backups to external hard drives. The data is being sent in chunks, depending on protocol this could be from a couple of bytes to 100's of kilobytes. If the receiver finds a corrupt data package will signal the host and that package will be resent, causing a slowdown. The receiver end uses memory buffers to mitigate such hiccups.

A milling job data wise is nowhere in comparison with a music download or video streaming by far. Speed is not important at all, the amount of data what can be sent via USB in a few secs would mean long hours of work for the mill.

What I suspect is happening is, that the microcontroller what sits in the mill on the receiving end could use a bigger buffer (if there's any) and/or a more robust error handling. I'm pretty sure the manufacturer is well aware of this and working on it, hence the frequent upgrades for the v-panel and the firmware. By recommending cable changes and advising against any other resource hungry PC activity while milling is to reduce the chances of any miscommunication between the mill and v-panel.

As PC's tend to slow down as time goes (OS problem: growing reg entries, fragmentation, mem dumps etc. ),a complete reinstall would probably help. Also finish all toolpath calculations before sending it to the mill to have all resources freed for the v-panel. Setting the priority for the v-panel process (in the taskmanager) to above normal or high would also help.

 

[mir0]

Does USB protocol inherently have error correction built into it ?

NO ! - it does have error detection.

There is a principal difference between "error correction" as in ECC ("Error Correcting Code")
and as in "error detection triggers a resend".

USB has detection of errors via a CRC on each packet.
If corruption is detected, the USB controller will drop the packet and the host will have to (automatically) re-send it.
Unless it is an Isochronous type endpoint. For Isochronous endpoints, corrupt packets are just dropped and not re-sent.

 

[zero_zero]

Never said error correction.... The Roland mill is connected as a Class 7 USB device (printer with bi-directional transfer protocol) definitely not isochronous in my book... Will check nonetheless with a USB sniffer to see what's going on between the v-panel and the mill.

 

[CoolHandLuke]

Guys - most likely the problems you are having are the data corruption errors caused by your PCs during CAM tool paths calculations.
Small little (uncorrected) error or skipped bit and the whole calculated "thing" is just GARBAGE.

CAM tool paths calculation is quite intensive and resource demanding process - it needs REAL WORKSTATION PC
with REAL workstation components to run reliably and give you consistent results.

If you are using "regular over the counter PC" or even a "powerful gaming PC" -
and it was working OK for you - it is just pure LUCK - and it will end soon or later.
And when it does - you will loose more money (and time) then if you've invested in REAL workstation PC in the first place.

Personally I do NOT like buying ready made WORKSTATION PCs from big manufacturers like DELL, HP or Lenovo
because they use proprietary components and you are locked to their service and parts if you need one.

I prefer to custom build my own PCs from the best universally available components - much easier to service them later on.

Here are a few "MUST HAVE" when building or shopping for one:

• - professional Intel chipset motherboard with Intel XEON processor and ECC memory support (like: Supermicro X9SRA)
  
• - minimum of 32GB of ECC memory (Error Correcting Code)
• - minimum of 240GB SSD - Solid State Drive - Hard Drive (professional - like: Intel 520 SSD)
• - 90+ Gold or Platinium SeaSonic PSU (power supplay unit)
  
• - if your CAM software uses video card for calculation - your video card also needs to have ECC memory (like: nVidia Quadro 6000)
• - good computer case with plenty of air circulation for cooling of ALL components inside
• - Windows 7 x64 (64-bit) PRO or ULTIMATE !!!

we can all concede that certain computers are certainly better than others, however "workstation computer" are not two words that really should go together, as they do not form a coherent phrase. its like saying Amphibious Frog.

anyway all that to say the best computer in the world is limited by its software; in most circumstances that software is capable of detecting computational issues such as a memory stack overflow. in this case however it is a lost cause to be going through the computer looking for computational errors; the fault of the software - it is not likely. we've discussed how even doing the simulation in the sum3d has resulted in no error, so no arithmetic is at fault here.

i suspect the actual arms of the machine itself are at fault. if memory serves, they are powered and fed data (and controlled feedback) through the old IDE drive cabling, which as we may well know are aluminum wires that are easily broken (especially after repeated bending) and are not as conductive as a braided wire (which would be much easier to short circuit and lose data).

i believe the error to be roland-side, because if this were a computer hardware or sum3d error, it might be more prevalent in each user's system. software being so needy with the hardware, and all.

i say this based on the experience of working in a past lab where the same *symptom* arose, in completely different machinery, computers, and data transmission. different G code as well. same error - a machining stage was skipped, or more commonly in that system the issue of halting Z axis.

no error messages generate in both lab's circumstances. nothing in sum3d now, and nothing in the Laserdenta proprietary nesting code either.

the machine is at fault; the question now is how to service it. nobody but glen has been able to provide anything from roland to help with this error, but i got money says he isnt the Service advisor or team lead to robotic machine design.

 

[zero_zero]

If you think is broken cabling, then could be easily diagnosed with a multimeter. Even is is conductive, a broken wire will have a measurable resistance.

Sent from my SGH-I747M using Tapatalk 2

 

[CoolHandLuke]

could be it, could be the microcontroller... lots of parts that *might* have an issue.

i can't be the one to rip the machine open, i have a job to do. there should be people for this.

 

[sidesh0wb0b]

could be it, could be the microcontroller... lots of parts that *might* have an issue.

i can't be the one to rip the machine open, i have a job to do. there should be people for this.

didnt you guys just get this thing not too long ago? yikes!
everything seemed pretty well for the Roland, but i am starting to hear more folks having trouble like this in the past 3 months.

 

[DMC]

Just wait a little more....

I hear of many similar issues, but people are shy to post about it for some reason?

Maybe they are going to try and sell as a good healthy mill, but in reality......

 

[technician]

Today I had the exact same problem. Everything looked good but it must have skipped one tool path but not on all crowns?
It was only at the inside top of some crowns were you could clearly see that it was not milled all the way.
So as Glenn mentioned I did actually changed the USB cable because the original was really long. I'll changed it back.

Ok same thing happened again. I Milled only two units to try because its a 23 unit case and I didn't want to do all I did a 26 and 12.
26 ok but 12 the same thing not milled correctly in the inside top. But I decided to do the rest because theres no time to investigate what's causing the problems. Of all 23 units five of them were not milled correctly.

The tool path thats missing is always inside at the top and mostly it seems on units with conical shaped inside.
I called my distributor on friday and He said, (drum roll, wait for it)....................-Yes it happens. Nice support....
Anyway He said he will try to contact Roland or Noritake and ask about the problem.

When I look at milling simulations and at the tool path pattern it looks ok.
I saw somewhere theres a test file for Roland machines, does anyone have it?


Here you can see that the burr just didn't go all the way (in this picture) to the right. Its just a drill hole

This is the tool path and it looks ok to me. (not all of the tool paths included in this picture)

Later I will mill the same crown and see if it'll come out differently.

 

[DMC]

That looks like only the internal roughing, or semi-finishing tool-path.

What about the final finishing tool-path?

Did you verify the raw-curve for the margin was there?

Show us more please.....

 

[Labwa]

Can you share your cam file? Maybe it will be visible to someone.