Am having problems connecting to 3 of our CNC machines. Connection is
serial and we are using Pathtrace's Edgecomm to communicate. Parameters
have been checked as has the wiring but no errors found. The PC
providing this service has been replaced with a new one but the problem
persists. Edgecomm provides and error 18 unable to connect to device.
Other programs tried produce similar errors i.e. cannot connect to
Last week we had no problems this week, with no apparent changes having
taken place, we have no connectivity.
Two of the machines have Fanuc controllers, one has Heindenhain.
Anyone get any ideas? The only idea I have is that it might be and
earthing problem that has possibly developed due to the exceptional
warm and dry spell of weather we have had. However, I wouldn't know how
to check this - see if there is a voltage difference between the PC's
earth and the CNC's???
Any ideas greatfully received since currently I'm stumped!
Any lightning storms in the last week?
Get one of those LED dongles that will indicate any activity on the 232
buss and see what's happening. If it was working the odds are that the
pc is bad or not set right. Since you have three controls that aren't
working the odds are that not all three controls would have gone bad at
the same time. It most likely the pc or the common wiring from the pc.
You do not mention how you are switching between controls, a/b box or
electronic switching but that could also be the source of your problem.
Try isolating the error by trial and elimination:
1) Make a new cable, new ends, clean the connector at the control
with a little air or vacuum. If you use electrical cleaner, be sure
everything is off, and then dry before turning anything on.
2) try a computer/laptop on the floor near the control with the new
cable, verify that it works on another machine before troubleshooting
the problem machines.
3) get a breakout box at any electronics supply store, and compare
the lights with a control that is verified to the trouble machines.
4) switch the cable ports at the PC, to test your switch box. What
are you using for switching?
5) Can a program be sent and not recieved?? Maybe one pin is
corroded or tarnished, and can't make a solid connection, this
includes th ground pins.
6) Check the parameter settings on the controls, making sure you are
using the correct port.
7) With the new PC, make sure the serial port works, and is connected.
How did you test this?
8) try different software, something simple for CNC machines, download
something like a CNC editor.
3 machines only, sounds like switch box to me. Take it apart, and
clean the switch with electrical contact cleaner.
Good luck, let us know what works and what didn't.
No lightening - that's this week apparently!
Switching is using a manual switch box. This has been tested and is OK.
Read the Fanuc and Heidenhain links - useful stuff, will be saving them
for future reference.
Didn't get around to buying a breakout box but will do to assist with
any future problems.
The PC I'm using is one with a multitude of USB ports and no serial so
I am using a USB to serial convertor (which is working fine in two
other locations in the factory).
With regard to the faulty systemm, I have an identical setup and this
is OK. The only difference between the two is that the working setup
had surge protection. Added this to my faulty setup and it now works.
Our electrician is sceptical, but the only way I can prove this works
is to remove the surge protection and watch another (probably) PC get
messed up. Can't see the point in doing this and will therefore be
leaving well alone. Whilst I cannot say that this is the definitive
solution, it seems likely and at least the overall situation is
improved with the PC now having surge protection.
Thank you all for your interest and comments.
Ive had a lot of troubles with usb to serial port cables. They seem to
be very touchy. I would bet that you have a ground loop in the system
and the surge protector is correcting the problem. It doesnt take much
of a ground loop current to blow out the inputs or outputs of a pc or
They could be AC or DC current so if you test for them use both scales.
A large ground loop current can be an indication of other problems in
the building electrical system. A bad ground and a leaky appliance,
motor or such will give you ground loop currents.
I've had troubles with these USB -> Serial, while seemingly easy to
install initially, they are a pain to troubleshoot.
Check the driver, make sure the settings are correct for the USB,
something about the clock cycle if I recall correctly. Be sure to
check the web for updated drivers and manuals.
Check the actual serial port number Windows uses Control Panel ->
System -> 'Hardware' tab -> Device Manager, under Ports. While there,
double check all the little settings compared to the other computers.
Since you changed the computer, and reinstalled the drivers, the
virtual COM port may have changed ( because of a registry issue with
XP, and number of times installing the driver).
Get a PCMCIA serial card from Qualtech or StarCom, they are very
reliable and durable.
This sounds like a driver problem, not an electrical problem.
Newer Computers, Hardware, Software using USB is more stable now than
it use to be communicating with CNC's but it's still not bullet proof.
With my laptop I use a PCMCIA/RS232 card and have not had a problem
(if you type this in by hand don't forget the "s" on automations or you
will be redirected to another site)
This could be an option for your desktop although I cannot recommend it
from personal experience.
You don't say what kind of 'serial' is being used. If using common
mode serial, then your problems are expected. Use differential mode
types of serial. RS-232 is common mode which is but another reason why
the original RS-232 spec was only for connections of 50 feet and less.
Voltage differences can exist AND not be measured by a meter. To see
them, an oscilloscope or something equivalent would be necessary.
If surge protector causes a change, then problem is probably a
peripheral characteristic of that protector - its parasitic
capacitance. That capacitance would shunt out common mode noise - a
frequencies that involve things the electrician would not be familiar
Again, all this would be due to using a wrong type of serial port
technology. A problem made worse if any cable interconnects buildings
or equipment is powered from different breaker boxes.
Steve J wrote:
How long is the run....I'd imagine a noisy environment.?!
Did you build the cable yourself? Is it shielded? What wires are used
(Tx,Rx,grnd?...which gnd? chassis or signal?)
We got hit with lightning a couple weeks ago and had a number of things
go down (not puffs but certain not working afterwards). If we had've
spent the few $'s on isolation we would've been ok!
I'd suggest an opto IC something like the 4n35's. Basically, it's a
little IC that will take the hit for you and will protect the computer.
Wiring them up is easy (would suggest putting them in a socket as it
would be easier to change them out in the future).
Steve J wrote:
Relatively new to group.
I have recently converted some of my RS-232 communications over to
This has solved a few problems for me. Now there is no "one dedicated"
PC to tie up so multiple users can shoot programs from any computer.
And this has done away with any distance problems. And, now I don't
have to deal with the port issues and switch boxes of years past. This
has saved me from having to take the time to set-up one computer
perfectly....only to be replaced or mucked up by someone else.
If you are interested in this, let me know.
In answer to some of the questions raised:
Current wiring is 5 wire -
and at the device end, DSR, DTR & CD wired together.
This is all Fanuc standard.
The cable being used is shielded, but the shielding is not connected at
The maximum cable run is of the order of 20m.
The current situation:
One installation is working fine after surge protecting the PC's power
The second installation (which is surge protected) works fine on boot
up but after a minute or two, any attempt to download stops after only
few lines of code, say 100 bytes. Re-booting the PC is the current
Having read all the above, I'm considering using a Nport serial to
ethernet converter to provide opto isolation to see if that clears the
I appreciate that this is a kludge solution and, as has been pointed
out, RS232 was not intended to be used for such long runs. What would
be involved in using RS422 or RS485? Is it possible to simply buy
conveters at both ends or would I need to upgrade my Fanuc controllers?
First consider whether your problem is noise or involves something
larger such as voltage transient variations between two locations. For
example, do electronics on both ends of the RS-232 cable get power from
different power distribution sources? This latter problem also demands
galvanic isolation. If communication crosses 'power source' boundaries
(ie connects between two buildings), then galvanic isolation is
preferred AND cable must route so as to be 'surge protected less than
10 foot' to the earthing electrode for both power source power panels.
More on this later.
RS-485 is a differential mode answer to common mode RS-232. Anything
communicating beyond its limited area should be using RS-485; not
RS-232. Blackbox is a source of these serial port converters. Also BB
Peripheral cards are sold that work and act like RS-232 ports on the
PC but are using RS-485. hardware. Just another and cheaper option if
making the upgrade to RS-485 solutions.
If you put a surge protector adjacent to a power supply, well, that
is not effective protection. The surge protector is effective when it
makes a short connection to earth. The surge protector does not stop
or block surges. It is effective when it connects those transients to
the protection - earth ground. A protector that completly avoid all
discussion of earthing is just not effective. The effective protector
has a dedicated wire to make that all so essential 'less than 10 foot'
connection to earthing.
There is no short connection to that 'all so essential' common point
earthing if located at power supply. Anything that an adjacent
protector would do is already inside that supply. Essential and short
connection to earthing is same electrode required by NEC; back at the
breaker box. That is where a 'whole house' type protector provides
effective protection for that computer supply AND all other electronics
powered by that same box. Effective protection is about earthing any
transient on any incoming utility where that wire enters the building
AND to a same 'single point earth ground'.
Also a surge protector is for hundreds of volts. Even tens of volts
causes problems to RS-232. Tens of volts is completely ignored by the
surge protector. So what would the protector do? Earth transients
that would overwhelm protection already inside the electronics.
Shielding should be 'equipment grounded' only at one end of that
cable. Trial and error might teach more, but only if you have
electronic equipment (ie oscilloscope) to see the noise. Shielding
does not work very well for common mode type serial communication;
works so much better for differential mode serial such as RS-485.
Apparently you are using RTS-CTS to throttle data. Therefore serial
converter solution must support that configuration. Better would be to
use X-ON/X-OFF on RS-232 making the RTS-CTS connection irrelevant and
simplifying any serial port conversion. Does the equipment support
XON/XOFF making RTS-CTS connection redundant?
Steve J wrote:
Fanuc don't need 5 wire, just 2-3 crossed, and 6,8,20 connected
together, while 4-5 are connected together.
20m is not long at all, even at 9600 baud.
I would prove that that your communications is working before trying
RS422 or Nport converters.
Have you tried a different cable, or sending CNC to CNC?? Or setup a
computer right beside the machine?
Have you double checked all the parameter settings, identical to the
By rebooting the computer, it still sounds like a problem with the
Eliminate the problem, don't work around the problem.
Samurai makes an excellent point. Nothing posted previously says the
existing hardware is functioning properly. Posted previously were
solutions that assumed hardware on both ends was working properly -
which was only assumed. Changing cables would not provide useful
information. Smarter techs use an oscillscope or equivalent test
equipment. A badly distorted communication signal from defective
hardware would work most times - fail intermittently - and only be
clearly defective on test equipment. Is the Fanuc or data terminal
Samuai also implies is that the Fanuc already uses an Xon/Xoff
protocol or something that makes hardware data throttling (pins 4-5)
unnecessary. If you connect 4 to 5 and it works does not mean
throttling is unnecessary. Only means that one location does not
require data throttling. Connecting 4-5 together must be stated by the
manual - else it might be required in some locations and not others.
20 meters is excessive even if is works in some locations. Why?
Because 20 meters must work in every location. It does not. But
again, little things such as where the AC plugs are connected can
affect RS-232 - which is why RS-232 is not for distance communication.
Again, samurai makes a valid point. Were machines at each end of the
cable partially damaged? How would you know? What do comprehensive
hardware diagnostics (or other confirming methods) report? Without
confirming facts, we may only be working 'around the problem' as
samurai has warned.