What's new
What's new

Fadal Intermittent DNC Error w/ Cimco Edit

StirlingMachine

Aluminum
Joined
Apr 19, 2015
Hi all, I know aspects of this topic has been covered before, but I've been searching the forums for a few days and haven't found anything that quite matches my problem, so I decided it was time for a post.

Machine: Fadal VMC 40, (like 2216) 1991 vintage, CNC88 control.

Running Cimco Edit V5 for program load and drip feed.
Standard setup:
9600 BAUD
Stop Bits 1
Even Parity
Data Bits 7
DTR/RTS Yes (tried both ways)
\17 Xon
\19 Xoff

So here's the issue. Programs load fine with TA,1 assuming they're small enough. DNC works great, most of the time, but intermittently it does not. When it doesn't it'll be running along fine and then the code just stops and says DNC error, table is still, spindle and coolant still running. When this happens usually Cimco says program transferred successfully. This happens most often when using adaptive clearing type paths, but it doesn't happen consistently in the same part of the path, or even the same operation within a program if there are multiple adaptive style paths in a program. Initially I thought it was just the mass of code and how fast it was transferring, but after messing with the smoothing and tolerances to reduce the code massively and change to arc moves the problem still persists.

I'm using it setup as I bought it, PC, Cable, etc. The existing cable is off the shelf stuff, which was my first thought, I've got parts to make a cable I know has the right pinouts on the way. But the fact that it works most of the time and will load programs just fine has me questioning if that's really the issue. I've spent some time looking over the code that's produced, and the only thing that seems very different from a surfacing path or other high volume code is the number of feed changes, don't know if that would cause it though?

Any insight or experience with something similar would be much appreciated. I've been getting by just avoiding these types of toolpaths, but it would be great to be able to run them without worrying about wasting a ton of time restarting after errors mid stream. I'm currently running a program that makes this happen just to figure out what the deal is, and it seems like I get maybe 50% success of a DNC transfer without some error. I'm running Mastercam X4, and it seems to take dynamic core mill paths just fine, but dynamic pockets not so much.

Also, I've noticed the almost everyone runs Fadals with cables set up for software handshake instead of hardware handshake. Is there a particular reason for this?

Cheers and Thanks

Devin
 
Follow up to the first post. Just re did the program without any feed changes for the back feedrate between trochoidal moves, and same problem still happens.
 
How long is your cable? Hardware handshake requires voltage changes to the signal lines that sometimes get lost in the capacitance of a long cable. I'd try the transfer with maybe a 5 foot long cable to eliminate other issues. I had this problem using a cable 35 feet long so I had to park a computer right beside the machine. If that isn't convenient in your shop, you can go the route I did and get a Moxa Nport setup. This works over your ethernet and has a little server box that you can place real close to your machine's serial port. I've transferred millions of lines over that thing and never had a transmission error (distance of 40 feet, using my wireless LAN). It is my opinion that ethernet transfer has far more smarts (and capacity) to send and resend data that needs to be resent, far above the capacity of your serial port.
 
Currently the cable is about 12’. It does have several adapters in line at the moment so I’m hoping taking them out will help. Parts for a new cable are coming today.
 
I also have a VMC40 with a CNC88. I use a calmotion LANCNC-FAD DNC unit, which sits inside the machine cabinet, so the only cable is the ~16" ribbon cable direct to the interface card. Even with a brand new interface card, I also get DNC errors at 9600. 4800 works fine, so that's what I have resigned myself to. From some reading on this issue, an inline optoisolation unit may help, but I haven't bothered to make one up yet- all off the shelf units I have found use a DB9 connector, rather than the DB25 on the fadal.

If you haven't already, you might spend time assessing the actual feeds/accel rates possible with your machine relative to your intended programs to determine whether the extra bandwidth provided by running at 9600 is even a benefit. In other words, is the limiting factor for feed rate your DNC rate or the feed rates possible with your control/servos?
 
Have you tried the OneCNC or the Fadal DNC programs?

OneCNC I could get to DNC, I couldn't with the Fadal DNC program.

I have issues with the 4020 keyboard which can be affected by a noisy connection to the PC, when the PC is disconnected it's fine. I would look at the quality of the cable and connections.
 
I'll look into the inline isolation unit, seems like if DB9 is all that's available it could go at the PC end of the cable just fine?

Bandwidth has definitely been the issue so far, when running HSM toolpaths it's often stuttering running out of code even using fairly conservative feeds. That might be internal to the Fadal though, because when I load a program using TA,1 the transfer rate is obviously much faster than it can process code through DNC, or at least that's how it appears to me.
 
I'll look into the inline isolation unit, seems like if DB9 is all that's available it could go at the PC end of the cable just fine?

Bandwidth has definitely been the issue so far, when running HSM toolpaths it's often stuttering running out of code even using fairly conservative feeds. That might be internal to the Fadal though, because when I load a program using TA,1 the transfer rate is obviously much faster than it can process code through DNC, or at least that's how it appears to me.

What version of boards do you have? I have a 1991 4020 that was upgraded to -4 boards and machines pockets etc at 100ipm without stuttering, that's a fairly cheap upgrade, the -5 is even better, but more $'s.
 
Use the fadal dnc program.
If you need it, PM me... but its easily found in the wild ;)
Im running a 20' rs232 null cable at 19200 through a startech usb adapter.... zero issues
-5 machine
 
What version of boards do you have? I have a 1991 4020 that was upgraded to -4 boards and machines pockets etc at 100ipm without stuttering, that's a fairly cheap upgrade, the -5 is even better, but more $'s.

I hadn't ever looked before, didn't realize that was a thing. That's the problem with working alone and being self taught. Looks like I've got a mix of -0 and -1, I'm guessing that's part of the problem?

Been running the rest of the day at 4800 Baud with zero issues. Just got the new cable put together, I'm going to see if I can go back up to 9600 without issue.

Edit: Did a little research. The 1400 board which I assume is what you were referring to, is a -2A

2nd Edit: New cable with known proper pinouts didn't fix the issue at 9600. Optical Isolator should show up thursday, I'll update if that helps at all.
 
I hadn't ever looked before, didn't realize that was a thing. That's the problem with working alone and being self taught. Looks like I've got a mix of -0 and -1, I'm guessing that's part of the problem?

Been running the rest of the day at 4800 Baud with zero issues. Just got the new cable put together, I'm going to see if I can go back up to 9600 without issue.

Edit: Did a little research. The 1400 board which I assume is what you were referring to, is a -2A

2nd Edit: New cable with known proper pinouts didn't fix the issue at 9600. Optical Isolator should show up thursday, I'll update if that helps at all.

At the end of the board number there's the configuration level, I'm not sure if that's the right way to describe it. I think you should be -2 for that year.

Fadalcnc and itscnc both have upgrade kits to go to the -4 and -5 configurations from what it seems you have. Not that expensive, and maybe of great benefit.
ITS has an upgrade to -4 kit for $2750, and upgrade to -5 for $5500. I'd stick with the -4, it will be more than adequate.

Ebay maybe a good source for boards. There's a fadal document that has board compatability requirements. Using that you could see what boards you have to replace.
 
At the end of the board number there's the configuration level, I'm not sure if that's the right way to describe it. I think you should be -2 for that year.

Fadalcnc and itscnc both have upgrade kits to go to the -4 and -5 configurations from what it seems you have. Not that expensive, and maybe of great benefit.
ITS has an upgrade to -4 kit for $2750, and upgrade to -5 for $5500. I'd stick with the -4, it will be more than adequate.

Ebay maybe a good source for boards. There's a fadal document that has board compatability requirements. Using that you could see what boards you have to replace.


Yeah, looks like I'd also need to upgrade my 1010, and 1420 boards
 
Yeah, looks like I'd also need to upgrade my 1010, and 1420 boards

You might want to talk to ITS, who, the list of board compatibility may not be entirely accurate (so I've read on the internet for what it's worth)

I believe ITS made the boards for Fadal, when Fadal was in Chatsworth, so are probably the best source of current correct information.
 
Can you expand on your Moxa setup?
I have an old Moxa 2150 I could configure and setup but I don't think it ever supported XON and XOFF very well which I figured would be a problem with DNC to a Fadal.

Richard

How long is your cable? Hardware handshake requires voltage changes to the signal lines that sometimes get lost in the capacitance of a long cable. I'd try the transfer with maybe a 5 foot long cable to eliminate other issues. I had this problem using a cable 35 feet long so I had to park a computer right beside the machine. If that isn't convenient in your shop, you can go the route I did and get a Moxa Nport setup. This works over your ethernet and has a little server box that you can place real close to your machine's serial port. I've transferred millions of lines over that thing and never had a transmission error (distance of 40 feet, using my wireless LAN). It is my opinion that ethernet transfer has far more smarts (and capacity) to send and resend data that needs to be resent, far above the capacity of your serial port.
 








 
Back
Top