THC accel/decel

[Vmax549]

Hi Balazs, The S curve solution could be a year or more away. "IF" teh S curve solution will actually work correctly for THC. I do know that Art and Tom tried everything they could think of to make THC z control work with accel work in Mach3 and never found a solution other than bang bang.

Creating that function now in UCCNC is NOT a half baked solution. It is a viable solution to a known problem and proven to work and work well by thousands of users world wide that use it every day. Then later " IF " teh Scurve solution works out to be better, Great. But there is NO garrantee that it will work properly or any better for THC at this point.

(;-) TP

[beefy]

I do understand that CncDrive do not want to make modifications now that only last a short time and are then obsolete when the new S-planner comes out.

So I guess the big question is what's a rough estimate for when the S-curve version could be up and running.

Could we be talking over a year as Terry mentioned.

Keith

[ger21]

Time estimates for software development are somewhat meaningless. I doubt they know how long it will take.
But it's promising that they are going to be starting soon.

[Vmax549]

Another thing to think about is S curve is not teh holy grail of CNC. It is really only better in a few instances of motion. Will there be teh option to use teh trap planner if teh Scurve is not best for your application ?

(;-) TP

[cncdrive]

Yes, you are right, S-curve profiling does not solve this at all. What solves this issue is that the new profiler will have a setable to non zero start step frequency.
The S curve profiling makes this issue even worse the larger the S is, because the acceleration is actually slower in both the bottom S and top S compared to trapezoidal profiling.
But you will have the ability to shift the S location from 0. And you will also have the ability to zero the S parameter to have trapezoidal profile when the S is zero size.
So Terry, what I described is basically the same what you calling "bang bang".
And this is not something what Art figured out, this is a fact solution based on how stepper motors work, that you can jump start and jump stop them under their start/stop stepping frequency.
Doing so causes a huge jerk, but it is possible to do it.
And you can do it also with servos, a servo will not jump like a stepper, but it will accelerate up the fastest it can, the only limitation is the PID gains and the motor and load inertia.

I don't know how much time it will take, but we will be more smart about it when we can actually get there to actually start making it which seems to be not far away now.
I mean we might encounter into non pre-calculated in difficulties. We will see these better when we will start making it's base code.

[cncdrive]

Furthermore this solution which I have described about the new S-curve profiler will work better with plasma control than just accelerating and deccelerating without a ramp (called bang bang by Terry), because with that you have to limit the velocity below the motor's start/stop frequency while with the solution which we will implement you do not have to, you just have to shift the start frequency point to the start/stop frequency of the motor.
So, the acceleration happens without ramp up to that velocity point and then the set acceleration is used, so the motor will not loose steps. Same with the decceleration ramp, it will deccelerate down to that velocity point and then it will just stop without using decceleration ramp.
With the "bang bang" solution you must limit the velocity to below the motor's start/stop frequency, otherwise it is risky that it will loose steps...

[Vmax549]

HIYA Balazs you are missing teh point that teh Z under THC control will NEVER reach max velocity because teh moves are so small. But those moves must be at max accell . S curve is unneeded for that function. Bang bang has always worked and always will under THC control. Now if you run teh Z as a servo PID that is another story.

(;-) TP

[cncdrive]

No, I'm not missing that point at all.
With infinate acceleration the target velocity is reached in 0 seconds, so it will reach the programmed velocity no matter how small the moves are.
But ofcourse you can (and should) limit the velocity if they are all G1 moves because then the target velocity is the F programmed rate.
because if it happens with G0 then the stepper will loose steps no matter how short the movement is.

[Vmax549]

Reaching target vel in 0 sec is impossible.Even Bang has an accel ramp that is defined by the physical charetoristics of teh motor design and the power being supplied to it. It is as steep as teh motor can physically accel but it is not zero for sure.

But now that you mentioned it EACH machine could have a different accel rate how will you define the cutoff point with an Scurve. Is the cutoff point going to be programable from inside UCCNC ?

Many Commercial controllers never developed Scurve for a reason. Even Fanuc if I remember correctly never developed an Scurve planner they have what it called a time limited planner which seems to be teh best of both worlds. Scurve can cause problems in some areas of motion like threading.

(;-) TP

[Robertspark]

Reaching target vel in 0 sec is impossible.Even Bang has an accel ramp that is defined by the physical charetoristics of teh motor design and the power being supplied to it. It is as steep as teh motor can physically accel but it is not zero for sure.
(;-) TP

I am just wondering if you have tried to increase the acceleration in uccnc and lowered the velocity (actually tried it physically?) In order to try to emulate Mach3.

Whilst Mach3 THC Z axis in theory has "infinate" acceleration, it is not actually infinate, it is the pulse width of the velocity setting

say you have 400Steps per unit (mm) [5mm leadscrew, 10micro steps, common 200 step per rev stepper motor, direct drive]

If you set the velocity at say 1200mm/min [20mm/sec], therefore the velocity is 8000 steps per sec, so the pulse width is 62.5uSec

So in Mach3 the machine will accelerate from standstill to 1200mm/min [20mm/sec] in 62.5uSec so the acceleration is 320000mm/s/s

Therefore have you physically / actually tried to enter in something like the above acceleration and limiting velocity to see if it works for you

EDIT: or the acceleration could be 1/2 this if its HZ / pulse width.... I've got other things on the brain. Point being has it been tried?