Jog is sluggish, overshoots, and EMC halts

More
26 May 2010 18:19 #3017 by dcassyc1
Hi all,
I got my router going and proceeded to run 3 part files through it. First one turned out OK( 300kb) , then toolchange and loaded and ran second file (1.3MB). I noticed after file 2 completed that the machine wouldn' t respond to jog commands unless the key was held down for a second, and when it did respond it overshot the destination point considerably, although the jog was set to 'continuous'. Decided to restart EMC and the problem disappeard, then loaded and ran file 3 (1.3MB) but the machine stopped about 2/3 of the way through. Looked in the gcode window at the active line and there was no stop signal in the file.
Would anyone have an idea why the jog is sluggish, and why the part file only got 2/3 way through before machine halted? Computer is AMD 1.15 Ghz, Asus computer, with 1GB ram. I conducted memory tests on the ram for several hours which reported good.
cheers,
Denis

Please Log in or Create an account to join the conversation.

More
28 May 2010 02:49 #3028 by dcassyc1
Hi All,
I tried runnig the same part file twice today in EMC 2.4 which I'd loaded on to a compact flash. The part program stopped running twice and messaged that limit swith on axis 0 had tripped, but it was no where close to the switch to trip it. Each time it stopped running was in a different part of the file, and I started to run the part file over again and it did not error on the second attempt. After the part file had run, the jog performance was still sliggish as compared to before running the part file. I closed down EMC and started it again and the jog response is fixed. I got a real time error two times early into the first part file and adjusted the latency timing incrementally in units of 1000 up to 17000, even though the test came back at slightly less that 15000ns. Since increasing the setting to 17000ns I'm not getting any more rtapi error messages.
I'm using proximity switches for home and limit switches and wonder if it's possible that the switch sensitivity may be drifting, so that it sets even though there's no reason to. I'm seriously considering going to optical switches.
Would anyone have any ideas why there a message reporting that the home limit switch has tripped, even though it has not? And why the jog performance is sluggish after running a part file.
regards
Denis

Please Log in or Create an account to join the conversation.

More
28 May 2010 10:50 #3034 by BigJohnT
The only time I've seen the limit and home switch messages is when a signal has been received on that line. Depends on how you have them hooked up you might be getting a spike on one of the lines. Are they DC 2 wire proxes? You can't chain 2 wire proxes as they have a leakage voltage that will add up to enough to turn on an input.

John

Please Log in or Create an account to join the conversation.

More
28 May 2010 16:28 #3040 by dcassyc1
Hi John
thanks for the heads up on EMC performance. Sounds like a machine problem. Both the X limit/home switches are 3 wire, capacitive, PNP output trnasistor, NC output, pull up output when not at target. Both of these switches feed into a 2 input positve nand gate for isolation from one another and level switching. I check for noise and possibly substitute the switch for another.
I got 2 problems at once, the other is the jog performance is bad after running a part file. The jog performance is OK after I restart EMC.
Have you got any idea why the jog is sluggish and overshoots after running a part file?
cheers,
Denis

Please Log in or Create an account to join the conversation.

More
29 May 2010 11:17 - 29 May 2010 11:25 #3041 by BigJohnT
I recall something about a memory leak at one time and a problem with large files at one time. What version are you using? The memory leak thing would act just like having your base period too fast and the processor would not have any time left over to process user commands like nml communications between keyboard and real time EMC. You might pop up the system monitor and watch your memory when you run the file.

John
Last edit: 29 May 2010 11:25 by BigJohnT.

Please Log in or Create an account to join the conversation.

More
30 May 2010 20:29 #3047 by dcassyc1
Hi John
I'm running EMC v 2.4 on a compact flash, just to see what it's like to run with out hard disks.
I found that the PC power supply was intermittently putting out noise and replaced it. Changed the proxy switch too - no difference. Scoped the optical isolators and found a lot of noise on the power supply rails so replaced the 5 and 24 vdc supplies and it cleared up.
I'm running a file now and it's good so far, no home/limit switch errors. Now, maybe the latency test will return lower figures, I'll see later on.
An added bonus maybe; I noticed the jog problem was gone after the first tool change, so perhaps there won't be any more sluggish / overshoot jog response as a result of these changes to eliminate the noise conditions.
I'll let you know how it turned out.
Thanks for your help so for
cheers,
Denis

Please Log in or Create an account to join the conversation.

More
01 Jun 2010 12:47 #3056 by andypugh
dcassyc1 wrote:

Hi John
I'm running EMC v 2.4 on a compact flash, just to see what it's like to run with out hard disks.
I found that the PC power supply was intermittently putting out noise and replaced it. Changed the proxy switch too - no difference. Scoped the optical isolators and found a lot of noise on the power supply rails so replaced the 5 and 24 vdc supplies and it cleared up.


Fixing the issue in hardware is better than masking it in software, but...
Consider adding a "debounce" function to your HAL file for the limit switches. You can probably handle a 5mS delay between hitting the limit and the axis stopping a lot more than you can handle a job stopping halfway through. (How far does your axis travel in 5mS, 25 thou?)

www.linuxcnc.org/docview/html//man/man9/debounce.9.html

Please Log in or Create an account to join the conversation.

More
05 Jun 2010 03:01 #3069 by dcassyc1
Hi All,
I checked the control and found a lot of noise coming the servo motor drives and the spindle inverter, so I rewired the control with more attention to separating the motor conductors and control wiring. It just completed the same file with no limit switch errors. As you mentioned John, it was probably noise causing the false limit switch signals. In my case it was the control wiring scheme and an intermittent noise fault in the PC power supply. The PC power supply replacement alone enabled a reduction in the latency period from 17000 to 15000ns.
Thanks to everyone for the help
cheers,
Denis

Please Log in or Create an account to join the conversation.

Time to create page: 0.184 seconds
Powered by Kunena Forum