Spindle Speed control issues with Super-PIDin 2.4

Sorry. I didn't know that limit. That file is 2500 so that explains it.

i can't see anything that would cause the problems you describe , excepting the RT errors ,
can you try another Motherboard or remove devices not needed and see if that improves ,

your files could be easily modified to suit 2.5.1 , i'd upgrade

I upgraded to the latest Linuxcnc version this morning and it had no affect on the spindle or the latency issues. Before I upgraded, I talked to the guy I bough the drivers and pc from and they gave me direction in the stepconfig parameters. I adjusted them and they made no difference. I made sure that there is no additional hardware attached and that the bells are turned off and all that.

I did the SMI procedure shown on these pages because my log files were showing a message that indicated something about latency happening every 64 seconds. After I made the SMI change, that message quit appearing in the log files, but I still got the RTAPI errors.

I also tried using the isolcpus=1 process from David Kent (I think that was who). I have an Atom (quad core) processor so this was set to isolcpus=3 instead. The procedure seemed to work according to the instructions but It seems to have had no affect.

I have gotten in the habit of running the latency test quite often and yesterday when I ran it, I opened a large pdf file (O'reilly's Learning Python manual) and resized the window while it was opening. The Max jitter jumped to 840,000! It normally goes to anywhere form 14-40,000 so this really got my attention.

I think a real clue to understanding the issue with the spindle not starting is the difference between manual mode and MDI mode behavior. These are so different that there must be something in that to help me find the spindle root cause. I have literally gone back and forth hundreds of times testing this and there is absolutely a difference. Manual mode response is instant on or off. MDI mode sometimes "never" actually turns the spindle on.

I played a little bit with the HAL file lines entered for controlling the spindle. At first, it looked like they made a difference but it was just momentary. What it led me to is that the first time I enter the M3 in MDI, the spindle starts relatively quickly (acouple of second delay). As I do it over and over, the delay seems to get longer but I have not done any serious data logging about it or anything.

I have not tried to have the spindle signal bypass the Breakout board with a jumper. That will require some solder work in tight spaces and I have large clumsy shakey hands.

Andy

I have not tried the Linuxcnc status window. I am totally new to this world and so I am not even sure what that is. I will look into it and post what I see.

BTW - my motherboard is the Gigabyte GA-D525TUD - 2 Atom processors, not a single quad core as I stated earlier.

BigJohnT wrote:

Do you have the LinuxCNC Status window open when your testing your spindle? It might show something.

John

Thanks a lot, BigJohnT! I turned on that status window and then ran my tests with manual and MDI mode. When I did so, I saw that manual set the speed to 1.0. Otherwise, it all looked the same as MDI. I went back and tried it in MDI using M3 s1 and it worked great. Repeated it several times successfully so I went ahead and added the line "m3 s1" into some programs and it worked great. I tried 3 programs that I had trouble with and they all worked. It means manually editing my Gcode or setting up a template or something, which is not preferred, but I can work with it.

Thanks for pointing out a useful tool!

Now if only I can figure out what to do with the insanely large latency.

Andy

Thanks anypugh, but I actually had figured that part out a little earlier. I just read the G code manual a little more. The confounding part was that M3 s5000 should turn it on but it does not do so consistently. I never thought to set S=1 since that's not actually a real speed my spindle will turn. Since my minimum speed is 5000, I would think s1 and s5000 should do the same thing or that s5000 would work and s1 would not, but I am obviously still missing some piece. I am running fine with S1 though.

Andy

I have also now eliminated my RTAPI errors. I just went into the BIOS at login and disabled hyperthreading. It took like 2 minutes. I am still running 2 CPU's but doing this killed the other 2 and the latency got much better. The speed is noticeably slower for opening pdf files and things, but this machine is a dedicated machine controller anyway so no big problem there. I ran 2 full programs and never saw an RTAPI error, so I think this issue is resolved for now. Surfing the email community history has made me aware of the MESA I/O cards, so now I may have to get one of those and a breakout board in their kit.

I hope everybody else is having as much fun with this stuff as I am.
Thanks, again for your help, folks.
Andy