First Build 5 x 12 Steel Frame CNC router

[JAZZCNC]

Now your ino the realms of Black Magic.!!

First thing first are you sure the Drive is tuned correctly to the motor.? If not then wasting you time tuning the controller.

However 500in/s/s is very high(12700mm/s/s) and could be the cause of your tiny overshoot. This is very abrupt stop and from high speed. You can see in the video how abrupty it stops. Try lowering the Accelration and see if the error disappears or reduces.

Also I notice you using kD value which often if required is much higher valule than 10. 100's even 1000's isn't unusual.
The kD value will fine tune the oscillation or any humming noice from motors.

You also don't mention the Kvff value. With such high acceleration then this value will come into play slightly as it trying to predict the next action and can help reduce errors if matched well to PID.

To be honest no one can truely help you with this unless in front of machine. Even then it's mostly trial and error. Also I wouldn't waste much time fine tuning untill the axis is fully finished and working as it will be when cutting. Just adding little more Mass or lowering resonance can change the Tuning when comes to fine detail.

Good luck and keep cool it will be worth the effort.

[Scott Damman]

Yeah I figured the cart was ahead of the horse so to speak but it was fun to get something moving. I was planning on just doing a quick tune so that I could jogg the axis... 3 hrs later I was so frustrated I had to say stop.

I dont know that the motor is tuned correctly, these are the units from BST and I just made sure to set the drive to the correct number for the motor. I will need to do a little more learning on servo drive tuning before jumping into that can of worms... Should I need to dig into the servo driver setup? I was hopeful that just setting the number for the motor would get me where I needed to be.

The acceleration was set just trying to push the limits of the servos a bit, I ran all the way down to 50 inch/s/s and it is always 0.003 inch off the mark with the PID set the way I have it. Then if I changed the PID, the overshoot and hunting was always essentially the same, so with that I came to the conclusion that the acceleration may not be in play? But obviously more tuning needs to be done.

I do need to play more with the KD and Kvff, but I think I will take your advice and wait until the whole thing is together before performing the final tuning.

Have a Mach4 question as well, DRO readout with the CSMIO-IP/a, in Mach3 if I would turn a servo motor by hand the DRO would update. In Mach4 I am not getting that, hoping that its just a setting somewhere that I overlooked? Anyone have any thoughts? I was using the demo of Mach3 and decided to buy Mach4 so I dont need/want to upgrade as quickly.

So onto finishing the mechanicals. Thanks for the help

Here is the Z axis jogging, I didnt do much of any tuning here just got it running making it easier to work on.

[Boyan Silyavski]

Are you sure about that acceleration? My machine is tuned at 3000mms2 and is much snappier at 10000mm min feed rate. And from servos is still on 50-60% of their ability. Frankly when working at around 16000mm/min it looks really dangerous and leaves you no time for any reaction.

[Scott Damman]

Are you sure about that acceleration? My machine is tuned at 3000mms2 and is much snappier at 10000mm min feed rate. And from servos is still on 50-60% of their ability. Frankly when working at around 16000mm/min it looks really dangerous and leaves you no time for any reaction.

No I am not sure, its very possible your machine is snappier I will have to take a look at your video, pretty hard to compare on video I suppose. Anyway, I was just going based on the settings in Mach that I had in at the time, quiet difficult to confirm at what speed its running. Also I was just jogging at a % of max (I am not sure if that % reduces both the feed rate and the acceleration) If both then what the video shows would be slower...

[Boyan Silyavski]

No I am not sure, its very possible your machine is snappier I will have to take a look at your video, pretty hard to compare on video I suppose. Anyway, I was just going based on the settings in Mach that I had in at the time, quiet difficult to confirm at what speed its running. Also I was just jogging at a % of max (I am not sure if that % reduces both the feed rate and the acceleration) If both then what the video shows would be slower...

I agree, its difficult to guess from video. All I was saying that you have to check that acceleration in Mach3, the plugin you are using and the drives. If either one of the 3 is set differently then its the lowest value active in reality. Though i dont know CSMIO and if you have to set the acceleration in the plugin separately, so yo better check manual.

My motors are only 400w but i assure you when i raise the acceleration snappiness from the servos themselves, its like shooting a bullet. Thats the sound how machine starts and stops. In fact my problems when tuning were related to Undershooting, cause i was not brave enough at first. So problem was it goes to position fast and then you could see how the motors for a millisecond were further positioning it in place. So i thought i am Overshooting. Yes but No. When i raised the servo response this disappeared. So auto tuning did not work i my case. Also it would be better if there is a total gain adjustment which auto adjusts all details, instead of separately adjusting each parameter. Which by the way is well described on many places around the web, but i still fail to comprehend it entirely .

[Davek0974]

I got my servos from BST too so likely the same, I think the thing with the drive number setting the drive up is that it has no idea what load the motor has - its just a ballpark setting. The downside is the manual is written Chinglish and unless you know exactly what parameter does what, then its all guesswork.

I spent hours and hours tuning, its made harder as you also need to test each tune at multiple speeds, I test at 3, 13,23,33,43 and so on as % of full speed - its easy to tune so that it runs great at say 60-100% but oscillates badly below 50%

In the end i tweaked in the order P - D - I until the following error was as low as possible, then added some kVff if needed.

I'm still not happy with it though but at least the mill is running now. No doubt an expert could tune it in minutes but I don't know any mobile servo experts ;)

[Scott Damman]

Made a tiny bit of progress last night. Got one of the Y axis limits and also the homing prox installed.

Chinglish and unless you know exactly what parameter does what, then its all guesswork.

Dave, This is exactly what I was worried about when digging into the servo itself. I think if I get a bit more understanding of servo driver tuning I will have a better chance of understanding the Chinglish... I hope.

Boyan, that is a very good point, I never checked the max acc and decc in the servo drive. Likely that is the limiting factor. I didnt think about it until now but as I increased the acc I didnt notice an increase physically so probably the servo drive is limiting. I will have to take a look tonight.

Which by the way is well described on many places around the web, but i still fail to comprehend it entirely .

I totally agree, seems like each spot that describes how to tune a servo does it slightly different, making it more confusing for those of us just getting into it... Like Jazz said Black Magic

Thanks for the help!

[Chaz]

Let us know how you manage with the PID tuning.

I have done 3 machines with CS Labs and it's been hit and miss with all of them.

The first was my Denford. If the settings are 'weak' the machine is slow and cumbersome. If too aggressive I get PID errors although I have found a happy medium.

In all cases I have tried auto PID tuning (the controller, after tuning the drives themselves) but limited success.

Even the manual indicates its better to do manually in most cases.

For another machine I retrofitted, we managed to get a nice speed / rate on the machine but its small and doesnt carry a lot of weight.

With my Concrete machine (Thor), it will be interesting. The design is fixed gantry, Y moves a table of around 100kg and I am looking at speeds of >10m/min using a Panasonic 1.5 KW Servo and very beefy ballscrews.

The drives have an auto tune feature which tunes all the time. How this works in reality, I dont know. I will have the Y running again this weekend, just need to do some alignment and will see what I can get from the machine.

I have a video of the machine doing slow movements of I think 4000mm/min. I have had one servo run away (my own fault, made a mistake in config) and I can confirm that the table can move a metric fuckton faster.

https://www.youtube.com/watch?v=5sMb4XmzuXc

[Davek0974]

I have a video of the machine doing slow movements of I think 4000mm/min. I have had one servo run away (my own fault, made a mistake in config) and I can confirm that the table can move a metric fuckton faster.

https://www.youtube.com/watch?v=5sMb4XmzuXc

LOL ;) nice unit of measure :)

[Chaz]

LOL ;) nice unit of measure :)

Ye, fairly standard UOM.