Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: 13brv3 on April 04, 2009, 10:52:46 PM
-
Greetings,
I've been running a Fireball V90 using Probotix motors and controllers for about 6 months now. The computer was a fossil (Athlon XP2200+) that I literally couldn't GIVE away about a year ago. Nevertheless, it ran fine, and would jog at 220 ipm for X and Y, though I generally kept it set at 100 ipm max. I recently needed to replace the machine, because it wouldn't run some other software that I needed to run.
The new machine was a pretty cheap build, consisting of a Gigabyte GA-G31M-ES2L mother board (on-board parallel port, and video), 4GB RAM, 2.8 GHz Core2-duo, and a single SATA HDD. Much to my dismay, the new computer won't run the X and Y motors faster than about 50 ipm. It runs fine at that speed, and I could certainly use it, but I just can't accept that my old POS computer was running at 220 ipm, and the new one can only manage 50 ipm. I certainly can't deny it though, because I still have the old computer running, and can move the cable back and forth between the two.
Both systems are XP-Pro, and neither has much else loaded on it. The configuration files were copied straight over from old to new, so I know the settings are all the same. It would be easier to list the things I haven't done, than everything I have done. I'll take a shot at it though:
-Everything on the Mach3 optimization recommendations has been tried, except for loading XP without ACPI, since it's too late for that. I did change it to "standard PC" (loses one CPU core), and "multiprocessor PC".
-I've tried every combination of parallel port type, and two different addresses.
-I tried every pulse frequency for the port.
-I upgraded to the latest version of Mach3
-I changed everything in Windows that I could find to help, though of course none of this was don't on the old computer.
-I changed every setting in BIOS that could have any effect.
-I scoped the step and direction pulses using both computers, and while they're hard to see, they seem very close to the same voltage.
-probably some other stuff I can't think of at the moment, but a total of 14 hours worth of work with nothing to show.
I ran the driver test, and it says it's "excellent", though there is a small, fairly regular pulse. About every other screen, there's a very large pulse. I compared this to the old computer, and it also has the large pulses, though not as many of the smaller pulses. I had nice screen captures of these, but they got written over when I restored the backup image for about the 5th time today.
What I have not done is try a PCI-e video card, rather than using the on-board video. I do have one, but the on-board video was one of the reasons I got this mobo. I could also order a PCI parallel port adapter, but see previous comment about the reason I got his mobo. Of course if they work, I'd do either or both in a heartbeat.
If anyone can think of something I've missed, I'd love to hear it.
Thanks,
Rusty (tired and depressed)
-
What was the kernel speed on the OLD PC ?
Just in case....did you shut down and restart when changing the Kernel on the new one ?
You can CHECK CONFIG and it will tell you if the K speed is wrong too.
I know.....it's not much but it's all I can think of right off.
Good luck,
RC
-
-I changed every setting in BIOS that could have any effect.
Make sure you disable C1E and EIST (dependent on CPU chosen) in the
board's BIOS.
-
What was the kernel speed on the OLD PC ?
Just in case....did you shut down and restart when changing the Kernel on the new one ?
You can CHECK CONFIG and it will tell you if the K speed is wrong too.
I know.....it's not much but it's all I can think of right off.
Good luck,
RC
Thanks for the reply. The old computer was set to 25k, as is the new computer. I've tried other settings, but always put it back to 25k when it didn't help. I did see the note about restarting the app when changing speeds, so I've been doing that.
The check config is something that I haven't looked at, but it found no errors. When I set the Kernal speed above 25k, it complains that I have it set above what is required, and it tells me to put it back to 25k.
BTW, I should have described the problem a little better. When I try to jog at 100 ipm, it moves some, at least until the acceleration is nearly complete. Sometimes, it moves almost the entire distance across the router before it stalls. Most often, it's lucky to move a half inch at 100 ipm.
Thanks,
Rusty
-
-I changed every setting in BIOS that could have any effect.
Make sure you disable C1E and EIST (dependent on CPU chosen) in the
board's BIOS.
Good thought, and I saw it on a search earlier. Unfortunately, those were among the settings I disabled earlier. There was a 3rd in the same group, but I don't recall what it was at the moment.
-
It might be the parallel port on the new PC. Maybe try a PCI parallel port card?
-
The first thing that jumped out at me is that you are running 4gB of RAM on XP. As XP can't address more than 3 gB, more memory than that can actually cause performance problems. I'm not sure that it is responsible for your lack of speed issue though as that sounds more like a HW limitation of the on-board LPT or perhaps the high precision HW timer that mach uses. If you have an O-Scope it might be worthwhile to look at the stability of the pulse stream generated.
As other have said you can try a PCI parallel port card or my personal preference upgrade to a SmoothStepper! The SmoothStepper moves the pulse generation to a dedicated motion controller giving you rock solid pulse streams and a faster pulse rate than the LPT could ever do. However if you use backlash compensation with the LPT you should be aware that it is not incorporated on the SS yet. Take a look at the SmoothStepper section of this forum or the Warp9TD forum or give me a call for more SS info.
-
Thanks for the additional comments. I mentioned the option of trying a PCI parallel port in the original post. If I don't get anywhere with today's testing, I'll certainly order one to try.
I've scoped the parallel signals, but they're really hard to capture. Today, I'll capture the X step and dir signals with a Saleae Logic device. Maybe there will be a clue there.
I'll also try the PCI video card, to eliminate the on-board video as a possibility. While I'm in there, I'll pull one of the 2GB memory cards as well. I've run 4GB on my main computer for years with XP Pro. Each mobo has it's own way of handling the memory over 3GB, and the Abit board gives XP 3.25 GB, where the Gigabyte only reports 2.99 GB. It's certainly possible that this could be causing a problem, so it's well worth trying.
As for the smoothstepper, I haven't followed it that closely, but I did look into it a bit yesterday. I'll be considering this a last resort, since the parallel port can, and should work fine.
Stay tuned as I waste another weekend day on this.
Rusty
-
Rusty,
I think what you are seeing is the difference in the quality of the step and direction pulse out of the new computer. To see that difference you need a good scope.The pulse is generated and will carry right thru the PP. So don't think a new PP card will help.
Now using a video card instead of the on-board video may help. The above comment is from trying a number of PC's. I use a SS on my DELL GX520, has on board video, and no matter what ( video card for it is not cheap so never tried ) the stepper speed suffered dramaticaly but they run great with the SS and the increase was 2 to 4x.
Never know what you can end up with until it's tried. So my recommendation is that you buy a PC with the undestanding from where it is beinig purchased that if unstisfactory for intended use 100% refund
and if they can't agree .......walk away! The savings may not be worth the agrEvation and time spent on the bargin!
BEEN THERE ........DONE THAT,
RICH
RICH
BTW, still use the old "dead dog laptop" via PP and its use is rock solid ! ;)
-
As for the smoothstepper, I haven't followed it that closely, but I did look into it a bit yesterday. I'll be considering this a last resort, since the parallel port can, and should work fine.
The way I look at it is the the PC does many things very well. It can crunch number, show pretty graphics, access lot of memory, lots of different types of I/O etc. What it does not do really well is real-time I/O. Art has done amazing things with the LPT driver but in the end the bottleneck is still the underlying PC architecture; it was not designed for real time I/O. Motion control cards also do some things very well. Mainly they do all the time critical stuff VERY well. So, by marrying the two you get the best of both worlds. You free up a great deal of the PC processor by not trying to make it generate the pulse streams for the steppers and you get the faster/steadier pulse stream from the SmoothStepper.
I agree that you 'probably' can get the LPT to work. Each motherboard is different and some folks have found that some simply will not work. One caveat on the video card is that some built in video chip-sets can't be disabled, so make sure your's can before buying another video card. Good luck.
If all else fails give me a holler for a SmoothStepper :)
-
I tried removing memory, installing a PCI-e video card, a few more settings in Mach, and BIOS, all with no success.
The old computer was still running an old version of Mach3 that had the old Mach2 driver test. That shows full height spikes on every screen refresh, but I wonder now if those were false spikes. I copied the new driver test over to the old computer, and it was REALLY nice compared to the new computer. No doubt that's the issue. I also put a logic analyzer on pins 2 and 3 for the X axis, and verified the same occasional pauses in step pulses, so there's no doubt that the timing is not nearly as stable on the new computer.
That leaves me with the question of why this is a problem. The driver test program declares that the timing is "excellent", and the motors certainly don't expect the next pulse at any particular time. They would just see it as a speed change, as you have when you ramp up or down in speed. Best I can figure, the motor controller must be the device that's having an issue with this. Not much to do about that really, but I'd bet that another brand of controller might work fine.
In all my research, I've noted that some controllers generate a hardware timer, that's fed back to the software for timing control. Software that uses this clock looks for it on pin 10 of the parallel port, and if your controller doesn't generate it's own clock, you can use a stand-alone board to supply it. Pity that Mach3 can't accept an external clock, because that would be a pretty easy fix to the timing problems.
Bottom line, I can't argue with the need for stable timing, at least in the case of the Probotix motor controllers. It also makes sense to let a dedicated external device handle the chore, as is done with the external timers I mentioned above. In this case, the best option appears to be the SS, so I officially give up. SmoothStepper order placed. I sure hope it works, because otherwise I'll be stuck at 50 ipm.
Thanks for all the comments and suggestions. Now get to work Jeff ;D
Rusty
-
Hi, Rusty
Sent you a PM message, Top left of this screen, Click on your messages.
Chip
-
Have you actually gone and looked at your drive settings? I have seen MACH change these on a reloading of mach.
Sounds like the ACCEL is way to high. THe 25k mode should run fine on just about any modern comp.
Double check your settings ALL of them. (;-)
(;-) TP
-
Have you actually gone and looked at your drive settings?
Probably not more than 50 times. I'll probably have nightmares about them :-)
Rusty
-
Have you copied your licence file onto the new computer?
Graham
-
Have you copied your licence file onto the new computer?
Graham
Yes. No problem with the license.
Rusty
-
Is the LPT 3 volt or 5 volt on the output pins?
Graham
-
Both computers have just under 5V on the parallel port pins.
I spent quite a lot of time on this over the weekend, including using a logic analyzer to watch the step pulses. It's clear that the new computer's timing isn't as stable as the old computer was, and there's nothing that can be done about that. The only part that bothers me is that the driver test says my timing is "excellent", so I presume the Mach3 folks consider this OK. I have to wonder if the timing issue isn't more of a problem with my Probotix controllers. Perhaps they're just not forgiving of slight timing changes.
Since I ordered a SmoothStepper from Jeff, it's not such a big problem now, but I don't like to be beaten by something like this, so it still bugs me a bit.
Thanks,
Rusty
-
I had to go to a meeting today, and found myself about a block from a decent computer store. For the sake of ruling out everything, and to satisfy my curiosity, I picked up a PCI to parallel adapter card. As expected, it didn't make any difference, and it makes sense that it wouldn't. The driver test shows the problem, and doesn't actually use the parallel port for the test. No harm done trying it, since I can always return it.
Cheers,
Rusty
-
SmoothStepper saves the day!
http://www.machsupport.com/forum/index.php/topic,11152.0.html
Thanks,
Rusty
-
Glad you like it, they are nice :)
Hood
-
Happy things are working for you. I know that the SS was another expense but compared
to the agrivation of that computer the SS cost is only a memory now. ;)
RICH
-
Hi All. I am an old user remotivated I have purchase a S/hand Xp 32bit computer approx three yeras old to use with Mach3. My old computer worked fine with Mach 3 but was getting old and intermittant boot up. I have pinned my problem with slow running motors down to Pulse frequency . Mach3 Pulse frequency set at 25k but when in use on Diagnosis page only getting 4500-5000. If if run LPT Debugger frequency is 25K and moturs run ok. Don't want to run with LPT debugger on all the time any suggestions where I may look for a fix have spent many hours so far to get to this point.
Thanks in advance Graeme
-
Have you installed any software since it worked correctly or done any windows updates?
If not then it could simply be the motherboard dying slowly, might be worth checking that it is not clogged up with dust and also have a look in the BIOS and see if the voltages are close to what they should be.
Hood
-
Sorry the computer I have the problem with was a fresh install on a clean xp computer.
No dust clean board
Checked Bios found frequency set at auto I changed to 2500mhz.
Tried mach3 again in diagnosis screen pulse frequency. approx 5500 motors run slow.
When I run a lpt port monitor the frequency changes to 25000 Or more.
Motors sound right but sound like missing steps
-
Ok first thing is BIOS has no control over the frequency of Mach, I suspect what you have set is the processors frequency. You will have to be careful with that as you may be over or under clocking it, it should be set to whatever your processors speed is.
Ok so if this is a new computer that doesnt work then have a look at the optimisation document. It should be on the Mach3 main site but you may have to search for it as it is a new site and I am unsure exactly where it is now.
Hood
-
I found windows XP professional Optimization guide.
XP home edition came pre installed on computer can I run optimization steps from after install or do I need a fresh install? might not have disk for home only Dell xp Professional?
I also tried running USB controlled HyCNC-4P-60 from Hytechworks. This is a controller unit generating its own Frequency etc the motors sounds much different than when running on LPT (sound grindy)
I am not sure if it is running at the right speed what is the best way to test speed of motors?
I thought I could Program a square and check Time and distance to complete.
Thanks Graeme
-
You can do some of the steps I think, never really looked as I have never had a problem myself and thus never needed to do them.
People have mentioned that one on or two of the steps have helped so often not all are needed, just do what you can and see.
Also some people have tried all of the steps and still it has not helped so they had to either try another computer or go with an external controller.
Time will be correct if distance is and distance is determined by you setting the correct steps per unit.
Hood
-
Ok went through xp optimize couldnt do acpt pc as this killed my computer an had re install xp twice
Still same problem with slow motors frequency low around 6000 not 25000 any other ideas
Graemecns
-
Only things I could really suggest is to try another computer or possibly try a PCI parallel port or get an external controller such as CSMIO OR SmoothStepper or similar.
Hood
-
thanks for the reply.
I am able to use my older PC which I now have working again. But the plan was to upgrade the computer hardware. I did purchase a Hytechworks USB controller but have returned that unit as I could not get it work properly.
the motors sounded like they were grinding and with the speeds set correctly I was getting 85mm travel DRO indicating 65mm.
I will look and see if there is something else I can purchase to go between the computer and new stepper driver I have just received (has jog controls and a screen on top of package) LPT input.
I will do some more research
Thanks Again Graemecns
-
I am glad to have stumbled upon this thread. I too have experienced vast differences in my steppers performance with a simple computer change. Computers with very similar specs. I have a video where I was testing my mill during a warm up routine and 300 ipm split on the 3 axis simultaneously. I am lucky to get 60 out of any single axis now.
Didnt seem to really see a solution here but I do know some lower spec'd PCs have given me some higher machining speeds. I was thinking I was having a machine problem only to disconnect the stepper and find the screws spin freely. The stepper was stalling. I do have a few SS's but that isnt the point.
-
Hi all,
my take on the timing variation is
When a stepper is running at max speed, lets say 1200 rpm at 500 step per rev (I use Vexta 5-phase steppers) corresponds to 10,000 pulses per second.
If the pulse generator (ie the PC) is stable the period between rising edges is 0.1ms. If the PC isn't doing as well then the period between pulses will vary.
If we assume just 10% variation the periods will vary from 0.09ms to 0.11ms which corresponds to a variation in speed of 1333rpm and 1090rpm.
Further these speed variations can occur in rapid succession and constitute VERY high accelerations/ deaccelerations and is no wonder the steppers can't
keep up.
My 'big' scope, while old, is quite good and has a histogram function which allows me to measure % variation on a nominally constant pulse stream. I have found
that more than 2-3% severely limits high speed running. At much lower cutting speeds variations of 10 and more % can be tolerated.
I don't believe that the PP, breakout board or stepper controller have that much influence provided they work of course!
As comments in this thread and others make clear an external motion controller is really the only way to eliminate pulse speed variations. I use a single board
Atom based PC for my controller and it works fine so I haven't bothered with an external controller.
I also use an embedded OS (Windows 7) which allows me the choice of what modules run in the machine. I don't run any Ethernet or Ineternet capable modules
as they have security checking processes which 'clogs' things up when trying to run Mach3.
Craig
-
Craig, Thanks for the response. I dont understand a lot of what you said but it does seem to confirm my suspicion. There are differences in performance.
Though I do have a few flavors of SS, USB and ESS. My thought is that if the computer is causing the issue, while will it not show up via the SS? Trying to quantify just what is responsible for stepper performance is difficult. It isnt as simple as meeting the minimum requirements for Mach or adding RAM or graphics card.
I have had great results from a few PC's over the years via a PP and would like to pursue that avenue before any other approach even if ancient by todays thought.
I see you and a few others are running Atoms or mini itx systems (computers are not my forte either so excuse my terminology if I am off). From what little I can tell, They seem to have integral graphics cards and parallel ports. You dont have any issues running Mach with them? What kind of specs would be required? I wouldnt mind building one. There has to be a youtube video on how to do that! ;-)
-
Hi Fastest1,
the board I use is a UNIGENS UMB1 which is a MiniITX dual core Atom running at 1.8GHz. The graphics unit (GPU) is built in and shares main RAM.
The NM10 chipset includes a PP. So yes the graphics and PP are effectively included. My board is fitted with 4 Gig of RAM but in fact I wasted money
there, 2 Gig would have been ample and a 64 Gig SSHD. It has one PCI slot which I use for a second PP.
The wee Atom is not a powerful computer by any stretch but has run perfectly from day one. I have seen people spend big bucks on powerful platforms
only to find they don't work as well as an old XP from the ark.
I believe the real reason that this platform works as well as it does is the OS. As I mentioned I installed Windows 7 Embedded. Embedded OS's have been around
a long while, there are a few flavours of XP Embedded for instance. They are typically used by manufacturers for specific or otherwise dedicated devices, Point-Of-Sale
machines as an example. Windows 7 Embedded (Standard) has a very user friendly setup and installation suite which is why I chose it. XP Embedded has 15000 software
modules to chose from when installing, I am no expert and that looked daunting to me.
When you install an embedded OS you can chose certain modules that you want and deselect those that you don't. For my purposes I elected to do away with the modules
associated with Internet and Ethernet security. Those processes have to run in the background continuously and interfere with what you want to run. I don't wish to
expose my machine to the Internet nor put it on any network, I want it to cut chips and nothing else.
If you search for 'single board computer' or 'miniITX' you will be flooded with hits....
MiniITX.com has some very useful combinations that would suit a Mach3 controller for under $100 US.
If you use your existing power supply and monitor and if you have a SATA HD you are there....
Craig
-
Craig, Thanks again. Do you see any issues running a touchscreen with your set up? Or a similar system?
-
Hi Fastest,
I haven't tried a touchscreen so can't say definitively. Certainly the OS has the necessary software to recognise a touchscreen and others use
them so I believe that it would work.
I have found that my budget does not run to such things. If I had several hundred to spare I would uprate my 23 size steppers with planetaries to
grunty 34 size steppers without gear reducers or finish my high torque spindle for spinning face mills or finish my 4th axis drive or buy a direct
drive servo for fifth axis or... and so the list goes.
The main advantage of Mach3/Mach4 is that it is useable software for readily available and cheap computing platforms at a very good price. I accept the
limitations of Mach3 including the PP as the best I can afford. A Seimens 328D controller is on my list but way down there.
Craig
-
I have been very happy with old computers and Mach. The price is fantastic as well as the forum. It is just when odd unexplainable things get in the way, it is frustrating.
Sent from my iPhone using Tapatalk