Motor tuning - acceleration and velocity

[Simko]

Now that my table is moving under its own power, I'm trying to figure out how to determine maximum velocity and acceleration. Right now the X, Y, and A are set for 1000 ipm velocity and 50 in/sec/sec. I read a few articles, but the best I can find is one that says to start with the acceleration very low (2 ipm) and keep raising the velocity until you start losing steps. Then with the new max velocity in place, start raising the acceleration until you start to lose steps and then reduce by 10%.

What pattern do you run to test velocity and acceleration. Would a square work? Or would you want something more complex with a bunch of tight twists and turns to really put the motors through their paces. Or is the velocity and acceleration something that gets fine tuned over a period of time with real world use?

Velocity. This is currently at 1000. I can't think of a time that I would ever need the table to move at 1000 ipm. Should I lower this to say 500, which is still faster than anything I would need to cut.

Acceleration. Currently it is at 50 in/sec/sec. Again this seems to be faster than you need for plasma cutting, should it be closer to 35 in/sec/sec (assuming my table can handle that, which at this point, I am assuming it can.)

I did some searching on the site here, but didn't find much. Maybe I wasn't searching for he correct term. Any help would be appreciated.

[Simko]

I did a little tinkering... Tell me if this makes sense.

Max velocity. With a one inch pinion and 3.2:1 belt reduction, my axes will move .9817 inch per motor revolution.

Circumference of 1" pinion = pi X diameter = 3.14159 X 1" = 3.14159 linear move per rev if direct coupled

Final linear inches per rev with belt reduction = 3.14159/3.2 = .9817"

According to CandCNC the 980 oz-in motors at 65V and a moderate load will spin at approximately 800 RPM.

800RPM X .9817 inches per rev = 785.36 IPM max velocity under moderate load.

[Brand X]

35 is a good all around setting to start with.. I still think your gears will want to pull up from the gear racks at anything over 400 ipm. Unless you have a huge amount of tension on your adjustments. Although the 34 gears might be a little better setup over my Nema 23 from the same supplier as your table. Having your rapids that fast is not needed for starting out.. Set your machine to 300 ipm, just to get the feel for it.. No reason to beat your machine up at this point. Things will happen fast enough.. you can change the motor tuning to suit the job anyway..

Start conservative, best bet. You are setting up your machine for world record top speed runs, while you are still in pit road..

[Simko]

Yeah.. I know, i need to keep it slow at first. I'm not trying to run it super fast, I just want to understand what goes into calculating the max speeds and accels.

I set the spring so that it just barely touches and then not even a quarter turn more and it seems to hold fine. I even did some jogging at 800 ipm / 35 in/sec/sec and nothing jumped out of the track.

I got each of the axes calibrated fine. The X,Y, and A needed a little calculating which is pretty normal from what I read. The Z was right on the money by using exactly 10,000 steps (2000 micro steps per rev X 5 TPI)

[motoguy]

I too am interested in learning to tune my table for best performance. It's a manufactured table, but I realize each one can have a bit of "personality"

[Capstone]

You can work out the theoretical, but once you get the steps matched to the proper travel, it's best to work from max "G's" based on the weight of your gantry. If you can work your way up to 350ipm and don't have any shudder at the "Z", then stop there, you're good.

[exapprentice]

Hi Steve
If it was me I would set the velocity to the max cutting speed based on what material and which plasma machine you intend to go with.
Then keep increasing your acceleration until any quick changes in direction stalls the motors, By quick I mean really aggressive changes (forward & backwards) and keep trying to stall the motors don't try not stalling by being gentle, then back off say 15%.
If you can achieve the fastest cutting speed you anticipate using coupled with the highest acceleration you can dial in without stalling or losing steps then that's a good starting point

What your after is your max cutting speed with the ability to change direction as fast as possible.
At this point you will see the effect of inertia on your machine (but hopefully not).

Its awesome when it all stays together when cutting a file with lots of fine detail at high cut speeds, that's when you start questioning if the machine is up to it

All depends on what materials you are interested in cutting (thick = slow / thin = fast) just try and be as realistic as you can.
Have fun this is the easy bit
Most important bit is get your steps per spot on, check your numbers over the longest distance possible for calibration

0.5mm over 3 mtr aint bad (0.020 over 10')

[tcaudle]

Since the tuning is largely a function of gantry weight, friction, final drive type and several other factors, its difficult to give rules of thumb. The larger motors and higher voltage of your unit makes the RPM closer to 800 and with the EtherCut even higher. Now, that number is at the MAX RPM and at that point the motor has virtually no torque left and "stalls" (loses steps). That "point" is dependent on the load and how fast you try to get there. As a design rule I try to make it so the max CUTTING speeds are about 1/2 the top speed . With a ratio of close to 1:1 that makes it about 400 IPM. You can he higher rapids so you set the velocity up to something like 600. As for the acceleration: It's better to have acceleration than high speeds. To get sharp turns and good corners tha higher the number the better BUT there is a practical limit. Its like driving your car hard and aggressive with every start and stop resulting in smoking tires! We have found that a good number to try and hit is about 1/10 the IPM feedrate but in IPS/sec. So if you want to cut at 250 IPM you need an acceleration of about 25 IPSec We find 50 IPS/sec to be pretty aggressive unless you have an ultra light gantry and need to cut 26ga!. 35 IPS/sec seems to be a good number for most cutting down to 16ga. We ran our test table up to 100 IPS/sec and did a sawtooth cut and I though the machine was going to shake itself to pieces! You need enough to let the toolpath track but not so much that parts start to fly off your machine!. If you run the inertial math on a 100 lb gantry going from 0 to 300 IPM and back to 0 in less than 1 sec you come up with some pretty big numbers!

Like it says in the mirror: "Caution, objects may be closer than they appear" so It's one thing to find the limits of your table . Its another to use numbers that give you good cuts cut don't put undue stress on the mechanics.

[Simko]

Since the tuning is largely a function of gantry weight, friction, final drive type and several other factors, its difficult to give rules of thumb. The larger motors and higher voltage of your unit makes the RPM closer to 800 and with the EtherCut even higher. Now, that number is at the MAX RPM and at that point the motor has virtually no torque left and "stalls" (loses steps). That "point" is dependent on the load and how fast you try to get there. As a design rule I try to make it so the max CUTTING speeds are about 1/2 the top speed . With a ratio of close to 1:1 that makes it about 400 IPM. You can he higher rapids so you set the velocity up to something like 600. As for the acceleration: It's better to have acceleration than high speeds. To get sharp turns and good corners tha higher the number the better BUT there is a practical limit. Its like driving your car hard and aggressive with every start and stop resulting in smoking tires! We have found that a good number to try and hit is about 1/10 the IPM feedrate but in IPS/sec. So if you want to cut at 250 IPM you need an acceleration of about 25 IPSec We find 50 IPS/sec to be pretty aggressive unless you have an ultra light gantry and need to cut 26ga!. 35 IPS/sec seems to be a good number for most cutting down to 16ga. We ran our test table up to 100 IPS/sec and did a sawtooth cut and I though the machine was going to shake itself to pieces! You need enough to let the toolpath track but not so much that parts start to fly off your machine!. If you run the inertial math on a 100 lb gantry going from 0 to 300 IPM and back to 0 in less than 1 sec you come up with some pretty big numbers!

Like it says in the mirror: "Caution, objects may be closer than they appear" so It's one thing to find the limits of your table . Its another to use numbers that give you good cuts cut don't put undue stress on the mechanics.

Thanks for the reply Tom! This is great information.

Yesterday, I went through and set the max velocity to 400 IPM and the acceleration to 35 in/sec^2. I had no intention of ever running it at 785 IPM and 50 in/sec^2, but I just wanted to understand how to calculate what the max capabilities of the table should be.

I did some dry runs on a few fairly detailed files at 350 IPM and 35 in/sec^2... the table doesn't sound strained at all and the gantry is solid. Once I get a better pen holder, I will make sure that I am not getting any oscillation. Right now, everything looks pretty promising. I am just a water table and a plasma cutter away from the finish line.

[davek0974]

One little note that may / may not apply here - with massive acceleration/deceleration you also risk the plate being cut moving and/or the bed slats wobbling.

I have de-tuned my table somewhat as i was getting bad slat wobble on aggressive cuts in 3mm sheet. I built the table to run at 9000mm/min as that is all my PM45 can do according to the book, max velocity is set at 9500mm/min and acceleration has been lowered to 900 from 1800 so pretty close to the 1/10 rule.

Things were jiggling about all over the place at the higher setting, it only showed as a little ripple in one direction so although it was dancing around a bit, the whole machine was moving so the cut didn't suffer too badly! It was the plate wobbling the slats that showed the issue.

[Simko]

Most important bit is get your steps per spot on, check your numbers over the longest distance possible for calibration

0.5mm over 3 mtr aint bad (0.020 over 10')

Thanks!

I got it close for right now. I used my flaccid PVC holder to mount a sharp scribe and made pin pricks in cardboard. The biggest piece of cardboard I had was 48" so right now I am right on the money (within .25mm at 1219mm or 48"). Once I get a better pen holder and some bigger paper / cardboard, I will set it over the distance of the table (10').

[exapprentice]

Hi Steve

I didn't use your method (couldn't find a piece of paper big enough)

What I did was placed a piece of tap at the home end of the axis, on the tape was a fine line that I used as a datum.
Then measured using a tape measure (nice new one)between the datum line and a known fixed point on the carriage.
Input a distance using the MDI and check how far the axis actually moved, crunch the numbers in the CandCNC manual to make any adjustments to the step / per.
0.5mm (0.020") is pretty easy to see whether it's over 3mm or 3mtr.
How many times you do the check its up to you but I checked, double checked and checked some more until I was happy it was spot on
(you could say I got very pedantic during this exercise )
On my table the rails allowed me to rest the tape flat along them during the checks which made my life a little easier but I did consider getting a laser measure (not sure of the accuracy of these) my nice new tape did the trick for me

[SeanP]

Gantry Cam
https://www.youtube.com/watch?v=8n6CiGs8jAk

[Simko]

Gantry Cam
https://www.youtube.com/watch?v=8n6CiGs8jAk

Lol.. You have way too much time on your hands.

Very cool though!

[Simko]

Hi Steve

I didn't use your method (couldn't find a piece of paper big enough)

What I did was placed a piece of tap at the home end of the axis, on the tape was a fine line that I used as a datum.
Then measured using a tape measure (nice new one)between the datum line and a known fixed point on the carriage.
Input a distance using the MDI and check how far the axis actually moved.

I was going to come up with something like that, but I didn't want to waste any more time with my crappy pen holder. Thanks for the ideas!

[exapprentice]

Nice one seanp, love it
A picture is worth a thousand words and if it's a video even better