Parallel Port Troubles?

[Treischl]

This one has me really stumped.

Mach 3 was running pretty good on an older PC I had. (766 mhz). It could go nuts once in a great while, but I put that down to the speed not being what is recommended. OK, so my wife bought a new PC, I got her old one, 2.6 ghz. Nice and fast. Here is the sequence of events:

1) New Windows XP pro installation on it.
2) Installed new parallel board on PC (LPT3)
3) Reinstalled Mach 3
4) Uses CandCNC box with 2 parallel ports
5) No changes on the machine at all
6) Configured everything

Here is what goes wrong:

1) Start up PC
2) Start Mach 1 using mini-io-sc from CandCNC
3) Go to Diagnostics
4) Check all home switches, they are fine, when pressed, they are steady on
5) Jog machine on all axis
6) Go to Diagnostics
7) X home is intermittent at best, if it works at all
8) Z home can be flaky too, but not as bad as X
9) Y and A are perfectly fine.

Keep in mind, all is well BEFORE I jog the axis. Nothing has changed with the CandCNC setup, so I am not going to make Tom nutz, even though he does answer his phone and is a great guy.

I have checked the parallel ports out in Device Manager, it says they are ok. I have reinstalled stuff about a dozen times, no changes.

Anyone?

[ger21]

Does it work OK if you increase the debounce setting to a few thousand?

[Treischl]

Hi Gerry,

No, that does not help at all.

I have removed the second parallel card. That did not do it either. So, there is no conflict in Windows between the two parallel ports. It just seems that the one built into the motherboard does not operate correctly.

As I said before, Mach was working correctly as well as the machine, then I swapped out computers. I have searched the eMachine site and even talked to one of their reps, no luck.

I really do not want to run the machine without home switches, and I do not want to give up the pendant either.

[sunmix]

Hi, Sunmix here,

have you tried doing mach3 optimization? if not, you could try doing it, here's the link :http://www.artsoftcontrols.com/optimisation%20XP.txt

it's best that you do it with a clean new pc. if you do not wish to format, you could uninstall all the softwares, leaving the standard and default softwares that are pre-install with windows xp. after doing the optimization, if it still does not work, try using the special driver located on the main directory of mach3. no matter what you try, make sure you are able to complete the mach driver test.

[Treischl]

What makes you think that Mach is going to change the way the bios in the motherboard works? As much as I would like it to be a situation with Mach, I don't believe it is. After all, when the computer starts up and Mach is started, the parallel port is just fine. But after jogging the axis (remember, they use the the parallel port also) the two inputs are goofed up. I doubt that Mach is doing that. I talked to the eMachine dealer here, and he tells me that eMachines is only concerned about using parallel ports for printers. That, and I quote: " The parallel port was never meant to be used for controlling a machine". Nice, neat, narrow thinking. If it was a printer port only, they would have called it a PRINTER PORT.

Anyway, at this point, I have swapped some of the inputs around so I can zero x-a axis which is very important on a gantry type machine. Fortunately, Y also works most of the time too. That leaves Z out, but it is not important since I do not use preset tooling, or a tool changer.

After I have caught up with my workload, I am going to try reinstalling the 2nd parallel port and hopefully get the pendant back in working order.

My gut instinct is that this fix is going to have to be a software patch to the bios.

Thanks for the help.

[NC Cams]

The parallel port was the defacto standard that was used to drive printers way back when. Once guys figured out how the signals were sent and interrupted, they learned how to control stuff that the port was not really intended to control. Afterall, the port was originally conceived and configured by Apple and IBM to simply drive printers when the computers first came out.

When you dump print, you can dump it willy nilly and NOT do so in/with real time control. The "farther" the Wintel standard gets away from the old ISA architecture, the more non adaptive the previously "friendly" parallel port can be expected to become. It is my understanding that the newer Wintel platforms do NOT concern themselves with real time control of the parallel port - they dump data to it when the computer hard and software decide that it needs to happen/should happen.

Moreover, as the newer systems come on line, the "built in" ports are wired in concert with the chip set operating criteria, not necessarily with what was the old ISA methods. Add to this the emergence of low power LPT ports and you have some unsolveable issues with OEM LPT ports.


Perhaps this is why your add-on port works and your OEM port doesn't. It is not surprising that stuff isn't going to work today like it used to in the old "build it yourself" days of system integration. I've even seen situations where things like video cards will work in one PC and yet puke in another motherboard.

I'd suspect that E-machines has simply wired their system to work and communicate with the LPT port a certain way. Sadly, once the machine accessses the port the first time, it decides when or if it will operate it the same way or time the next go 'round.

The E-machines guy is nicely telling you that "his machine wasn't wired or designed for machine control via a parallel port" which is why it won't reliably work. Moreover, they're NOT going to accept liabilty for machine control due to issues they probably conciously decided NOT to get associated with. Some of the E-machine issues are so irritating that my ISP/computer shop has a big sign hanging there that says they WILL NOT SERVICE ANY E-MACHINE PRODUCT. When pro's won't touch it, why should I????

Consequently, you may be chasing wind trying to get your machine to communicate with the LPT. We ran into a similar situation on a data aquistion board we got from Heidenhain. It simply wouldn't work no matter what we did/tried on M/B's with a certain brand of chip sets - and the card was supposed to work 'universally' in an ISA socket.

Not on your life.

As sad as it may seem, it might be more advantageous to get another computer to use with your Mach based CNC system. Hopefully and perhaps, someone may have a fix. Then again, if you have our luck, there won't be an answer....

[Treischl]

NC Cams:

Yup, you very eloquently said what I was thinking. Of course, I am not going to toss a 2.6 ghz machine in the ol' dumpster anytime soon. After repinning things as I mentioned above, it works pretty good, not exactly the way it should, but good enough.

One thing I did notice that is of interest, when I switched from the 766 Mhz (not recommended by Mach at all) to the newer one, machine performance definitely improved. Consistent 90 ipm rapids were not possible. Now the machine rapids nicely at 90 ipm. It has also lost that irritating habit of dropping steps about once a day (naturally that happened just when a full sheet of parts was almost done).

So, I am better off than I was. I do miss the pendant though, so I will be giving that a try again in the near future.

I am very impressed with the flexibility that Art has built into Mach. It was nice to be able to just swap a few wires, change some pin inputs and get the machine to this point.