Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Eradicatore on January 03, 2013, 12:00:12 AM
-
Hi,
So I have a sherline 3 axis metric mill and their CNC driver box with 4 outputs (I'm not using the "A" output). I grabbed the xml for the setup from the xml ftp site Mach3 has. The X and Y motors run quite smooth both in 50% and 100% (shift key) jogging. I can change directions without issue. But on the Zed axis, I page down and it goes down every time, but if I try to go up after going down previously, the only way to get it to really go up is to do a quick tap of the page up key a few times. Then it will go up on subsequent presses of the page up key. Very strange.
So to debug it, I've tried a few things but they all seem intermittent. I unplugged Z motor from the Z output and instead connected the Z motor to the the "A" output of the cnc driver box. I updated the pins in the pin config of course, and then Z was working. Seemed perfect so I figured my Z output of the box or PC port was bad. But then it started doing it again. So I tried swapping the X and Z and similar story.
What is very odd is the behavior when it fails. I just can't switch from jogging Zed down to then jogging it up without the difficulty. It goes down even when I try to go up until I tap quickly the up jog key on the keyboard a few times. It's like its remembering that I'm going down. Note that switching from jogging up (after I finally get it working) to jogging down always works the very first try.
I will try to reverse the zed direction now to see if then I have the opposite issue.
I looked and I don't think I have any "backlash" turned on. But I'm quite new to mach3 so I could be wrong there?
So do I have a bad motor, a bad cnc box, a bad parallel port, or just a bad mach3 setup? ANY HELP APPRECIATED!
Justin
-
Hi Justin,
Could be your keyboard.
Assign a different key for the Z+ and see if it works.
Or, try another keyboaed if you have one handy.
Russ
-
Also, are you running a licensed install ? or the Demo version ?
Pirated SW does strange things similar to this too.
Russ
-
Thanks. Good idea. Let me try a new usb keyboard.
One thing I did notice after playing around more. It seems worse when I have Zed reversed. I.e. all the way up zero, and down towards the table being negative. That's how I want to obviously run it, but it seemed to work fine if I was running NOT reversed? Again though, it may be just intermittent. Its very hard to tell due to the problem seeming to come and go.
The other thing it may be related to, is a question I have. Do I need to turn off the CNC driver box any time I'm changing the pin settings or the polarity (reversed or not reversed)? I assume it's a good idea if not absolutely required. And again maybe that was part of my problem? I'll keep looking.
Even when its "working" it seems to me that the direction voltage may just not be fast enough out the printer port? That may be consistent with the issue only being in one direction? Could the rise voltage maybe take longer or shorter than the fall votage? I did put the meter on the pins on the side of the cable that plugs into the cnc box, and I only saw 3.3 V high and 0.6 or lower for low. I used pin 25 for a ground, so is my computer not putting out the required 5V possibly? Any idea how to raise that?
-
having never used a Sherline, just guessing here.
Assuming you are in Sherline Mode.
Many breakout boards accept 3.3v and are externally powered which provides 5v for the I/O, if all else is OK, this may not be a problem.
I never turn anything off to make config. changes.
The pulse polarity can cause problems if reversed, but the dir polarity sould not make much difference.
Lots of Sherlin folks here that will most likely help you out.
Greetings,
Russ
-
Also, are you running a licensed install ? or the Demo version ?
Pirated SW does strange things similar to this too.
Russ
Hi,
I'm running the demo. I've only been up and running for a few days now. I plan on buying the software but was just checking out the system. I had linux CNC loaded on ubuntu on this machine but the learning curve was a bit steep on their instructions. My friends at the real machine shop I know use mach3 and told me about the videos and I'm very very happy now learning on this program. Can't wait to watch more vids. :-)
-
having never used a Sherline, just guessing here.
Assuming you are in Sherline Mode.
Many breakout boards accept 3.3v and are externally powered which provides 5v for the I/O, if all else is OK, this may not be a problem.
I never turn anything off to make config. changes.
The pulse polarity can cause problems if reversed, but the dir polarity sould not make much difference.
Lots of Sherlin folks here that will most likely help you out.
Greetings,
Russ
Hi,
So here is the spec: http://www.sherline.com/8760pg.htm
It doesn't appear as if my 3.3V is a problem. I'll have to go back and make sure my low voltage was lower than 0.5V which I think it probably was. Also, the thing is that ONLY the Zed direction is failing, and even then, when it fails it's not a massive failure. Its just this tiny period of time where it goes in the wrong direction. I will hook up my logic analyzer tonight and get a graph output of the PC controls for a good (X and/or Y) and bad (Zed) case. That should completely rule out mach3/configuration/PC port issue. Then it would be just the CNC box which could be bad. I need to talk with Sherline if that's the case since I don't want to really touch anything in that box without their input first.
-
Ok, so I was about to connect up the logic analyzer tonight and debug, but then when I powered up, it worked perfectly. Then after maybe 20 min it started failing again. I think that's the smoking gun. Seems like a bad solder or something else heating up that is causing ohmic connection. I let it cool, and again it worked perfectly. No changes to config of mach3, no changes to wiring or connection changes to motors.
Time to talk with sherline about the CNC driver box.
-
Just a courtesy update. I've sent my box back to Sherline and they have been checking it out on EMC2. They found one "bad" cable, but it was Y and not Z or A which are my issue. Just to recap my issue and what testing I did:
The issue: When jogging Zed down, it will work 100% fine. When CHANGING from down to jog up (not rapid jog btw) it fails to change direction until you try it a few times. Eventually it will change to going up, and then every jog up from that point (until I ever jog down again) is 100% fine. So my whole issue is just changing from down to up in the Zed direction. This failure happens not just with jogging by the way. It fails during a full run of a cut too, causing my Zed to be actually lower than mach3 thinks based on DRO's. DANGEROUS/VERY BAD!!!
The testing I've tried:
- Swap out the Zed cable of the CNC driver box to the A cable. I of course changed the pin out settings also in mach3 so A parallel port pins now controlled Zed. Both A and Zed fail the same way. That rules out the parallel port signal for Zed alone being bad. A and Zed could both be bad but seems unlikely.
- Tried swaping Zed and X. The failure described above follows the Zed cable output of the CNC driver Box. So that rules out the motor being bad.
- Verified my settings in motor tuning (that I got from the mach3 FTP site with xml setups for different machines) was the same for X,Y, and Zed
- Oddly if you let the CNC box sit off for a while, the issue goes away until it's allowed to warm up for 15-20 minutes.
Sherline has tested my box, but I've not yet been able to ensure they were really testing for the right failure. I'm still in talks now live rather than email to try to make a break through. They don't have a computer tech yet, or at least one that I've been able to talk to. Which is very interesting because of one more piece of information I haven't mentioned yet.
See the photo below. When looking at the circuit board inside the box for cold solder joints or anything bad, I noticed a strange thing. X and Y have PIC16F627 chips with a white dot on them. Zed and A have the same chip (in a socket btw) that has NO WHITE DOT!. The person I talked to so far at sherline didn't know why, but did confirm that the chips are programmed with different firmware! So the mystery is why would two of the linear axis' be using one firmware, and one other linear axis (along with the rotational A axis) have different firmware? And does this firmware mean I need different motor tuning settings for Zed compared with X and Y? Recall that Zed works mostly. Just has issues with changing direction. So is there a strange timing issue with just the Zed and A firmware?
Does anyone know of a good test with a logic analyzer to do on the direction change signal versus the square wave pulse train for stepping? What is the critical separation in time between those two different signals? I have a logic analyzer so I could monitor the parallel port. But I'm more thinking I need to monitor whats coming out of the CNC box.
Justin
(https://dl.dropbox.com/u/226919/sherlineCncDriverBox.jpg)
-
Eradicatore
I had the same problem except in my case it was mostly the Z axis.....it would mainly go down even when receiving a signal to go raise the spindle.
X and Y would behave erratically too but not as often.
Sherline replaced the controller twice but no improvement at my end. We tried everything you could think of and in the end, what did help was this http://www.cnc4pc.com/Store/osc/product_info.php?products_id=203 .
An alternative would be to find an old PC XP computer. I found out about that after I got the buffer.
-
MaNo,
Thanks so much for the advice! I was really hoping to find someone with the same issue possibly. It sounds like we have the exact same issue. You mention yours is mostly Z, and so is mine. Did you try using A axis to control Z instead of Z? Did that show the same issues? Did your issue show up right away or only after the box warmed up? Also, that's very concerning that you saw X and Y issues but just rarely. Makes me think I could see them too possibly if I use it a lot more.
So this buffer you mention sounds like an excellent way to ensure that the signal out of the parallel port is the proper levels. Are you suggesting that an older PC may also have a better on board parallel port that isn't "low voltage"? Do you know if getting a PCI parallel port may also do the trick? I guess for 20 bucks, this buffer you're pointing out seems to be about the same price as a pci parallel port card anyway, and be a good safey precaution overall. I think I'll give it a go.
One question. Could you look in your driver box and do you see the same "two white, two black" PIC chip situation? Not sure how many versions of that box they have out there too though I guess. Mine is very recently purchased (Oct 2012).
-
Eradicatore,
OK, I re-read your post and it sounds to be exactly the same problem I had.
Here is some background info.
1) I purchased my units (8621 package) through http://www.discountcampus.com/cgi-bin/webc.exe/store/cncSystemsWin.htm and I can highly recommend them. The system came (June 2012) complete with a Dell Optiplex 380 and PCI parallel card. I opted for Windows OS and March 3. Windows is version 7
2) starting Mach3, the very first moves in any direction were mostly correct. Subsequent moves were incorrect. When the controller was warmed-up, the errors were immediate...always Z- and X+ as Y+ I thought it was me because my lack of knowledge of Mach3. (My other machine has a Syntec controller)
3) Between David (of dPP engineering/discount campus) we spent easily 12 (maybe more) hours over several days on the phone. I have a VOM and we went over all the connections on the board. No problem there. We also checked and tried all possible configurations settings but the problem persisted.
4) I contacted Luiz Ally (TechAlly on youtube) and he graciously send me his XML file. Still, the same problem.
5) David then got a RMA and I send the controller back to Sherline for exchange. Replacement arrived and.....same problem.
6) David got a new RMA for the second unit and I got a new board again. Still the same story.
7) Having never heard of a problem like mine, David suspected too weak a parallel signal and suggested the parallel buffer since nothing else seems to work. Obviously Sherline hasn't heard of it either as they didn't suggest trying a buffer before sending a replacement.
The buffer corrected the problem immediately. In retrospect I wish I had ordered his (dPP) "CNC Extended Box" (without probe) as that has a built-in buffer but at the time I didn't see the need for the other extras and the switches.
As to the white dots on the IC, well, current unit and previous have them.
I had an old PC (7-8 years??) in junk pile storage and the THAT one works without the buffer.
I am neither a technician nor an expert in the CNC field but it looks to me like you have no other option than to get the buffer or see if a friend has an old PC laying around.
Cheers
-
Thanks again MaNo! I ordered the $20 buffer and my sherline box is on it's way back too. I hope by the end of this week I can confirm your analysis and put this to bed. I will be so happy when I can finally get to trusting my cut and making some real stuff. I'm totally new to CNC, so I'm having fun learning along with all this debugging.
So just out of curiousity, did all 4 of your chips have the white dot on them? Or just the X and Y PIC chips?
Cheers
Justin
-
Justin, please post results when you get the controller up and running with the buffer. I do believe it will work for you....I know how frustrating it can be to have new new machine sitting there and not being able to use it.
The the chips with dots are the same 2 as on your image.
Good luck
Norbert
-
Doh! I got the buffer and driver box back today, went to hook it up, and realized I need two gender benders! :( I should have checked when I ordered it. Now I have to wait another day to test. the suspense is killing me! :)
-
.....so,what's happening?
-
Ok, finally got 2 db25 gender benders from amazon prime. You wouldn't believe how hard it is (around me anyway) to find a store that sells db25 gender benders!! Its now all hooked up and working perfectly. I do however need to test it for a few hours I think to make sure it doesn't fail again once it gets warm. fingers crossed. So far so good.
-
Excellent!!!! I'm doing a cut now that never worked and now it works perfect. Thanks MaNo for the 20 dollar patch to my problem!!!
Sherline was nice about trying to look into it, but they now I think need to make sure to warn all future customers of this potentially very dangerous issue. Especially since it sounds like the newer your computer is the more likely it will fail with their CNC driver box alone.
I owe you a pint MaNo! Cheers!
-
Hi Justin, glad to hear it's working for you. Still, keep observing machine doing some additional projects before you can feel confident enough that the toolpath works out as planned.
Cheers
Norbert
-
Yup, I totally agree. I ran about a 30 minute cut and observed the whole time to make sure nothing funny happened. I will update maybe in a week or two if there are any issues ever again. Thanks again!
Justin
-
Yup, I totally agree. I ran about a 30 minute cut and observed the whole time to make sure nothing funny happened. I will update maybe in a week or two if there are any issues ever again. Thanks again!
Justin
Well, just a quick update. Been busy so haven't used the mill as much as I would have liked. But I can now say that I've used it enough to be 100% sure that this buffer was my whole problem. Thanks again MaNo! You really were a life saver on this one!!