CarveOne's Work Table CNC Router Build

[jsantos]

As for the drivertest, it's unlikely that any PC issue would give you such consistent results.

Yeah, you're probably right. :-)

[CarveOne]

Carve one, If you want I can skip a week of having my g540 if you want to give it a try. To rule out your electronics. I do owe you one. Just don't blow it up...lol

That's a very kind offer FandZ. The fact is, it might be too hard to give it back. I want one of those for the Solsylva when I add the 5 start screws to it later this spring. Let's wait a little while on that offer. It is very much appreciated though.

[CarveOne]

Yup, I still have your address. I can send the port via US mail if you want.

US mail is fine, but let's see what happens by Sunday evening though. It needs to be a PCI card, as that is the only slot available in the computer.

If the Linux/EMC2 computer runs the machine properly with the mainboard pport, then the parallel cable, control box stuff, and the machine is ok and it is in the computer and may be the parallel port card.

My two local sources didn't have a PCI pport card. What is the brand and model of the card that you have?

[CarveOne]

Sherline mode sets the pulse width to 40. Changing the numbers in motor tuning won't give you any more than that.

As for the drivertest, it's unlikely that any PC issue would give you such consistent results.

Correct me if I'm wrong, but you have two different axis' losing position by the same amount, at the same time. right?

Yes, X and Y.

X and A are slaved in Mach3. All four drivers are G203V. X and A are on a common heat sink. Y and Z are on a common heat sink. 72vdc Keling 20amp linear supply, C11G BOB. No home/limit switches (just jumpers on those pins).

[CarveOne]

This just triggered something. The 1909.86 steps per inch is for the Gecko G540 which has 10 microsteps. What controllers are you using. Do they have microstepping ?

They're supposed to be the same circuit as the G203V, so it should work the same.

[Zygoat]

hay carv. parallel cards are not that expensive. by the time you guys pay postage to and from you could have dam near bought one, with shipping they are only like $15. then you have one if you need it later.

not saying the offers anrt nice. (really generous actually). just that for almost the same money it would cost you guys, you can have a brand new one to keep. then you can use it for extra inputs and stuff down the line.

http://www.newegg.com/Product/Produc...e&Order=RATING

i would recomend the rosewill ones near the top. some have 2 pports. some have one. some have pport and serial ports. whatever suits your needs.

[CarveOne]

Do you also want a straight pport cable? I've got an extra 5-ft one. When I bought my cables, some guy was selling them for 3 for $5 in ebay so I have a couple of extras.

Changing the cable this evening didn't make a difference, so the one I have is probably ok and the problem is elsewhere. Thanks though.

[jsantos]

US mail is fine, but let's see what happens by Sunday evening though. It needs to be a PCI card, as that is the only slot available in the computer.

If the Linux/EMC2 computer runs the machine properly with the mainboard pport, then the parallel cable, control box stuff, and the machine is ok and it is in the computer and may be the parallel port card.

My two local sources didn't have a PCI pport card. What is the brand and model of the card that you have?

It's a StarTech, model PCI1PECP - Single Port PCI IEEE-1284 Parallel Card EPP/ECP Compatible is what it says on the box. This was the first parallel card that I used. It worked except that I couldn't get the charge pump to work. So I bought another one that supposedly works with the charge pump. Little did I know that none of the PCI cards work with the charge pump. So now I have two working parallel cards but still no charge pump. :-(

[jsantos]

Just for grins, did you ever try cutting just simple straight lines to see if they would drift? Like maybe,

g0 z1
g0 x1y1
g1 z-.1 f30
g1 x20 f50
g0 z1

[CarveOne]

hay carv. parallel cards are not that expensive. by the time you guys pay postage to and from you could have dam near bought one, with shipping they are only like $15. then you have one if you need it later.

not saying the offers anrt nice. (really generous actually). just that for almost the same money it would cost you guys, you can have a brand new one to keep. then you can use it for extra inputs and stuff down the line.

http://www.newegg.com/Product/Produc...e&Order=RATING

i would recomend the rosewill ones near the top. some have 2 pports. some have one. some have pport and serial ports. whatever suits your needs.

I've seen other recommendations for the Rosewill cards. The one I have has the same specs and MosChip MCS9865 chip as the SYBA card right above the Rosewill single pport card and looks just like it also. Mine is a Creative I/O brand.

[CarveOne]

Just for grins, did you ever try cutting just simple straight lines to see if they would drift? Like maybe,

g0 z1
g0 x1y1
g1 z-.1 f30
g1 x20 f50
g0 z1

No, but I'll try it.

[FandZ]

Yes, X and Y.

X and A are slaved in Mach3. All four drivers are G203V. X and A are on a common heat sink. Y and Z are on a common heat sink. 72vdc Keling 20amp linear supply, C11G BOB. No home/limit switches (just jumpers on those pins).

Makes me wonder if when all 3 axis's are engaged at the same time if you are having some sort of brown out and loosing a step. No idea if that is possible. Me and electricity don't exactly speak the same language. Come to think about it, I think Old hack had a problem like that. Something about he was drawing too much power on an old line or something.

[greenb_60097]

You know, we had to splice the parallel cord as well. Doesn't someone sell a same sex cord, can remember the seller. Anyway I'm wondering if your losing some signal due to the connection??

[tpworks]

You know, we had to splice the parallel cord as well. Doesn't someone sell a same sex cord, can remember the seller. Anyway I'm wondering if your losing some signal due to the connection??

http://www.google.com/products/catal...d=0CCAQ8gIwAA#

[CarveOne]

Makes me wonder if when all 3 axis's are engaged at the same time if you are having some sort of brown out and loosing a step. No idea if that is possible. Me and electricity don't exactly speak the same language. Come to think about it, I think Old hack had a problem like that. Something about he was drawing too much power on an old line or something.

I remember it, but I don't remember his exact solution.

I can plug in a voltmeter while running the machine and watch it.

The building has 200 amp service, is just a couple of years old, and has 20 amp outlets everywhere. There is a heavy duty appliance extension cord from the wall to the surge protected power strip that the machine runs on and another one like it for the computer and control box to run on. These extensions both go to the same wall outlet. The only other thing that is drawing power is the overhead fluorescent lights. I can run the milling machine (on a different power branch) with no effect on the cnc operation. All of the building is wired to handle much more load than I would ever be using at one time.

[CarveOne]

You know, we had to splice the parallel cord as well. Doesn't someone sell a same sex cord, can remember the seller. Anyway I'm wondering if your losing some signal due to the connection??

The pport cable I was using on the new machine is a new 6' male to male with straight through wiring. They are available, just not in local stores around here. If I order a parallel port card I'll get a couple of new cables as well.

The one I use on the Solsylva is a male to female. It is an old cable that was used on other commercial electronic equipment that I salvaged from the engineering lab trash bin after a clean up. I just used a ribbon cable with insulation displacement connectors to get it to connect to the Xylotex controller's ribbon header connection point. It was reliably running the Solsylva machine.

[CarveOne]

I spent the evening moving the Mach3 computer to the Solsylva machine and setting up a new configuration file for it. I cloned the new machine's configuration file and then made changes to the new Solsylva configuration file to get the motors running in the right direction and then calibrated the steps per inch.

I then started running five consecutive full cycles of the same file using the Mach3 computer to drive the Solsylva/Xylotex machine. The attached photo file shows that it does the same kind of slide, but to a much smaller degree. The Z is still lifting a little and the X/Y is moving about 0.025" instead of a little over 0.1" per cycle. The direction of the slide is reversed from the new machine. The ports and pins settings for X and Y active high/low are reversed from the new machine configuration.

I'll hook up the Linux computer to the new machine tomorrow and set up a new configuration file for the new machine. Then see what that combo does.

I'm beginning to think I'll need to call in a good exorcist before this is over.

CarveOne

[CarveOne]

Note that in the last photo all of the walls have no steps. The cutter is sliding toward the upper right corner. It is not leaving the steps. It cuts them each time the bit steps down. After the total depth is reached, the next cycle does the same thing. Each little step over goes away when the full depth is reached. It only leaves the evidence that the Z is drifting upward.

CarveOne

[Zygoat]

looks like it was the computer then. if the big one runs ok with linux, we found our culprit.

[jsantos]

This definitely looks like your computer or whatever is on it. Maybe it's time to call Mach3 support and have them look at your xml file to see if there's a configuration that is not correctly set.