Tuning issues (RESOLVED)

Not nescicerilaly will you be able to see the movement. Remeber there is probably a long distance between the point of the tool and the way bearing, so a small movement like you are seeing at the tool tip, could be practically microscopic at the bearing and not easily noticeable with the naked eye.

Big industrial router manufacturers like Biesse (I have no experiece with that company though) usually have very good tech support. You, might be able to give them a call, tell them your symptoms and they could probably tell you exactly what part or parts are most likely the problem, their prices and instructions how to change them. (be warned the prices may be shocking.)

Todd wrote:

Not nescicerilaly will you be able to see the movement. Remeber there is probably a long distance between the point of the tool and the way bearing, so a small movement like you are seeing at the tool tip, could be practically microscopic at the bearing and not easily noticeable with the naked eye.

Big industrial router manufacturers like Biesse (I have no experiece with that company though) usually have very good tech support. You, might be able to give them a call, tell them your symptoms and they could probably tell you exactly what part or parts are most likely the problem, their prices and instructions how to change them. (be warned the prices may be shocking.)

They wont even talk to you. They will not help in any way if it is older than 10 years. I have already spoken with them many times. They want you to buy a new machine.

Anyway,

I checked with dial indicator and made some realy violent x movements and it always returns back to 0 as in the video. If the X mechanics were bad that would not happen I dont think.

Like I said I have no experience with Biesse. SCM was quite helpful, when I recently resurrected a 17 year old machine of theirs. (Of course I did spend over $40,000 in parts in the process.)

The machine repeatedly returning to the same point is not necessarily an indication of no problems. It just means it's consistent. On the two large rack and pinion machines I regularly struggle with, they would perform similarly with that same test, yet still cut awful (similar wavy marks). Usually it is one or more bad (loose) bearings somewhere in the pinion drive train. I tear them all down and replace what ever is rotten, put them back together, and they'll perform almost adequately again. (I have no love for those machines.)

What is the backlash like? Can you affix your dial indicator so it touches near the end of that beam, then have someone push/pull at the other end to check for looseness.
Another check, with your dial indicator on the spindle push/pull on the spindle to check for play in all directions

Todd wrote:

Like I said I have no experience with Biesse. SCM was quite helpful, when I recently resurrected a 17 year old machine of theirs. (Of course I did spend over $40,000 in parts in the process.)

The machine repeatedly returning to the same point is not necessarily an indication of no problems. It just means it's consistent. On the two large rack and pinion machines I regularly struggle with, they would perform similarly with that same test, yet still cut awful (similar wavy marks). Usually it is one or more bad (loose) bearings somewhere in the pinion drive train. I tear them all down and replace what ever is rotten, put them back together, and they'll perform almost adequately again. (I have no love for those machines.)

What is the backlash like? Can you affix your dial indicator so it touches near the end of that beam, then have someone push/pull at the other end to check for looseness.
Another check, with your dial indicator on the spindle push/pull on the spindle to check for play in all directions

I thought if the rack or pinion or bearings were bad, it would not be repeatable at all.

I will try your idea for checking the backlash.

I did manage to fix these issues. It was "P" too high in the PID.

I don't understand that at all...... My scope traps dont show that at all. I max the P until I got Oscillation backed off a bit divided by 2 and used that P value. I got them real close in terms of the halscope traps.


Any way, it is working good with the lower P. That changed all the other PID so I still have to get the f-error back down using the hal scope, but I am quite happy.

Thanks Todd and all for help.

Poor FF2 settings can excite vibrations in a too aggressively tuned PID as well.

Todd wrote:

Poor FF2 settings can excite vibrations in a too aggressively tuned PID as well.

Yes but you would think you would be able to see the issues in the halscope. But you cannot at all.

I always have spikes on accel and decell, I can limit them but never get rid of them.

Unfortunately you will never completely eliminate spikes of following error at changes in acceleration with Linuxcnc as it is now. At least not until they figure out how to add jerk limiting to the motion planner. The best you can do is minimize them and limit max accelerations to levels that don't excite instability.

Todd wrote:

Unfortunately you will never completely eliminate spikes of following error at changes in acceleration with Linuxcnc as it is now. At least not until they figure out how to add jerk limiting to the motion planner. The best you can do is minimize them and limit max accelerations to levels that don't excite instability.

Yes but I don't see any instability in halscope, but on the machine you do. So the interpretation halscope is doing is not that accurate in my opinion. You can clearly see, hear and feel it on the machine, but halscope does not show it.That is why I was scratching my head not understanding. I was taking halscope as gospel.

If you have halscope monitoring the feedback position, it _is_ accurate, at least as far as the encoder position goes
(mechanical issues past the encoder are not traced but may have a small effect at the encoder)

If the resonance is not mechanical, I suspect something like this is going on:

The resonance is only happening during running of real gcode programs and not static tuning.
This would be detected (causing a fault) with reasonable following error settings.
To see this in halscope, you would probably need to run halscope in the normal or single
trigger mode and trigger on the following error level.