RE: [CAD_CAM_EDM_DRO] Print Servers for CNC
Posted by
Jeff Goldberg
on 2003-02-25 07:12:16 UTC
On further consideration, I agree with Tony. A single NIC to the print
server would certainly provide the speed with no collisions (or a switch as
Tony points out would work). I still think you're going to have a bunch of
interface issues as print servers are designed to work with printers, not
carry the information in the complex way that you desire for this
application. Even with printers, they tend to be a little flakey sometimes.
The chance of getting internal engineering support from an off-shore
manufacturer is pretty slim. I don't know if they still make it, but Intel
used to market a 3 port print server (or it might have been 2 port parallel,
1 port serial - I forget) under their own brand name.
Regards,
Jeff
-----Original Message-----
From: Tony Jeffree [mailto:tony@...]
Sent: Tuesday, February 25, 2003 9:32 AM
To: CAD_CAM_EDM_DRO@yahoogroups.com
Cc: CAD_CAM_EDM_DRO@yahoogroups.com
Subject: RE: [CAD_CAM_EDM_DRO] Print Servers for CNC
At 08:36 25/02/2003 -0500, Jeff Goldberg wrote:
ones that used coaxial cable, or used twisted pair cables interconnected
via "hubs") use a collision detection/backoff/retransmission algorithm to
resolve contention for the medium; however, that is NOT true of the more
recent "full duplex" Ethernet variants (that use "switches" to interconnect
twisted pair segments). Switched or full-duplex Ethernet does not employ
collision detection/backoff/retransmission at all, and a decently designed
"non-blocking" switch will deliver the full available bandwidth to all of
its connected stations.
Having said that, converting a print server into something that you could
use for CNC purposes sounds like hard work to me for other reasons - mainly
associated with the fact that they are designed on the assumption that what
is connected to them is a printer, which will accept data a character's
worth at a time, rather than the way that most parallel port-based CNC
controllers operate, where the controller is continuously monitoring the
state of individual port pins. So you would have to design whatever you
attached to the parallel port to perform the same handshaking protocol that
the printer uses to communicate with the server, and to accept data a
character at a time; similarly, you would have to design the control
software in the PC to output a stream of characters that the controller
could then sensibly interpret. This would potentially be further
complicated by any spooling mechanisms that Windows inserts between the
virtual parallel port on the PC and the print server, which could insert
delays that could be far more significant than any that a shared medium
Ethernet LAN might impose.
Also, communicating data in the other direction (for example, limit switch
contact closures) would be tricky.
Regards,
Tony
server would certainly provide the speed with no collisions (or a switch as
Tony points out would work). I still think you're going to have a bunch of
interface issues as print servers are designed to work with printers, not
carry the information in the complex way that you desire for this
application. Even with printers, they tend to be a little flakey sometimes.
The chance of getting internal engineering support from an off-shore
manufacturer is pretty slim. I don't know if they still make it, but Intel
used to market a 3 port print server (or it might have been 2 port parallel,
1 port serial - I forget) under their own brand name.
Regards,
Jeff
-----Original Message-----
From: Tony Jeffree [mailto:tony@...]
Sent: Tuesday, February 25, 2003 9:32 AM
To: CAD_CAM_EDM_DRO@yahoogroups.com
Cc: CAD_CAM_EDM_DRO@yahoogroups.com
Subject: RE: [CAD_CAM_EDM_DRO] Print Servers for CNC
At 08:36 25/02/2003 -0500, Jeff Goldberg wrote:
>I don't think speed will be an issue with a modern print server as theyI agree (see below).
>are generally the same throughput as a PC parallel port. The problems
>may arise from the ability to directly output signals on pins which are
>unusual for printers to use in the same way. Don't forget, they are
>not directly coupled to your PC, so any fooling around that the
>programmer does with the innards of your PC to generate or read signals
>may not work on a print server.
>Also, Ethernet works on a collision based retransmission basis. On aYou are right that the various "shared medium" variants of Ethernet (the
>small network this probably wouldn't cause a problem, but on a network
>with lots of traffic flowing, your data may be slightly delayed in
>reaching the print server which might cause issues.
ones that used coaxial cable, or used twisted pair cables interconnected
via "hubs") use a collision detection/backoff/retransmission algorithm to
resolve contention for the medium; however, that is NOT true of the more
recent "full duplex" Ethernet variants (that use "switches" to interconnect
twisted pair segments). Switched or full-duplex Ethernet does not employ
collision detection/backoff/retransmission at all, and a decently designed
"non-blocking" switch will deliver the full available bandwidth to all of
its connected stations.
Having said that, converting a print server into something that you could
use for CNC purposes sounds like hard work to me for other reasons - mainly
associated with the fact that they are designed on the assumption that what
is connected to them is a printer, which will accept data a character's
worth at a time, rather than the way that most parallel port-based CNC
controllers operate, where the controller is continuously monitoring the
state of individual port pins. So you would have to design whatever you
attached to the parallel port to perform the same handshaking protocol that
the printer uses to communicate with the server, and to accept data a
character at a time; similarly, you would have to design the control
software in the PC to output a stream of characters that the controller
could then sensibly interpret. This would potentially be further
complicated by any spooling mechanisms that Windows inserts between the
virtual parallel port on the PC and the print server, which could insert
delays that could be far more significant than any that a shared medium
Ethernet LAN might impose.
Also, communicating data in the other direction (for example, limit switch
contact closures) would be tricky.
Regards,
Tony
>I'd tend to suggest sticking to popular name brands because driver
>availability might be an issue if you change operating systems in the
>future (unless the price difference is too great, and you get involved
>up front with the assumption that it may end up as a paperweight
>sometime in the
>future)
>Theoretically, they should work, from a practical standpoint, you'd have to
>try it.
>
>Jeff
>
>
>-----Original Message-----
>From: ccq@... [mailto:ccq@...]
>Sent: Tuesday, February 25, 2003 7:28 AM
>To: 2ccq@...
>Subject: [CAD_CAM_EDM_DRO] Print Servers for CNC
>
>
>The following URLs have a print server on them; That's a 10/100
>ethernet port feeding three parallel ports. Would these be usable for
>CNC work? Would the software be able to access them properly? I would
>think they'd be speedy enough for smaller, slower machines, but.....
>
> http://www.pcbay.net/gen3parpor10.html
> http://www.amastore.com/gen3porparpr.html
Discussion Thread
ccq@x...
2003-02-25 04:30:28 UTC
Print Servers for CNC
turbulatordude <davemucha@j...
2003-02-25 05:15:22 UTC
Re: Print Servers for CNC
Jeff Goldberg
2003-02-25 05:36:52 UTC
RE: [CAD_CAM_EDM_DRO] Print Servers for CNC
Tony Jeffree
2003-02-25 06:30:06 UTC
RE: [CAD_CAM_EDM_DRO] Print Servers for CNC
ccq@x...
2003-02-25 06:34:49 UTC
Re: [CAD_CAM_EDM_DRO] Re: Print Servers for CNC
ccq@x...
2003-02-25 06:48:27 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
turbulatordude <davemucha@j...
2003-02-25 07:01:01 UTC
Re: Print Servers for CNC
alex
2003-02-25 07:01:38 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
Tony Jeffree
2003-02-25 07:07:44 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
Jeff Goldberg
2003-02-25 07:12:16 UTC
RE: [CAD_CAM_EDM_DRO] Print Servers for CNC
Jon Elson
2003-02-25 10:02:21 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
Jon Elson
2003-02-25 10:07:12 UTC
Re: [CAD_CAM_EDM_DRO] Re: Print Servers for CNC
jmkasunich <jmkasunich@y...
2003-02-25 11:21:36 UTC
Re: Print Servers for CNC
ccq@x...
2003-02-25 11:35:14 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
alex
2003-02-25 11:45:41 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
ccq@x...
2003-02-25 19:33:07 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
alex
2003-02-26 07:18:20 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
glee@i...
2003-02-26 07:33:14 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers and SBCs for CNC control
ccq@x...
2003-02-26 08:20:30 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
alex
2003-02-26 08:41:21 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC
alex
2003-02-26 09:49:51 UTC
Re: [CAD_CAM_EDM_DRO] Print Servers for CNC