THC accel/decel

[beefy]

Balazs,

I came across a thread post on the OrientalMotor forum, and the technical guy was explaining about doing exactly that.

I thought about asking you to do that but I was scared you may think I was asking too much so I stayed quiet, especially as I thought you had forgotten about this thread. Obviously you had not and you were thinking about it behind the scenes.

So what do I think ???????????? Ha ha, I think it's absolutely fantastic. I mean I don't know how it could be any better for a stepper motor and basic THC up/down control. You will be giving us everything you can for this situation.

The only bad side to this will be the waiting

As long as there is a separate THC Feedrate setting to limit the max THC speed, then perfect.

I would only see problems coming from a users lack of understanding. They may think it's great to have high THC Feedrate, without realising accel/decel is applied above the start/stop step frequency / rpm, and that effectively reduces the resolution of the Z control (e.g. time/distance it takes for Z to stop from max. THC speed). They also may not realise that with high THC Feedrate, the Z travels further between each sample period, when it's traveling at THC Feedrate.

But that is their problem, not yours, and what you are giving us, is the option to tune the settings to our liking. Overall I think this is great.

Keith.

[stirling]

@Balazs: On the face of it that seems to be a way forward - Thankyou.

@Beefy: Re: why I asked for your steps/per. I've run a simulation of your setup here and I'm getting a swing of around 0.8 mm around the set point. Does that sound about right for what you're seeing? (I should have asked for your accel setting also but forgot, so I used 2500mm/s/s just for a reasonable shot in the dark).

FWIW, even with massive accel, I still get some swing albeit somewhat reduced.

[beefy]

Hi Sterling,

can't remember the exact figure, but what I do remember is my eyes going big and wide as I realised the distance traveled being more than the cut gap distance (0.25 mm !!!!) for doing stainless with a Powermax 1250 Finecut nozzle.

So with your figures being based on an accel setting of 2500 mm/s/s I'd say you're in the ballpark of what I calculated. I basically just punched figures into an online calculator and don't have exact records.

I've got my accel set at 1500 mm/s/s

And yes I tried upping the accel to silly figures that I knew my stepper could not handle (at high speeds) and still got travel distance I wasn't happy with. THAT is when I thought, "Aha Art you sneaky guy, so that is why you didn't apply accel. to THC moves"

Keith.

[Vmax549]

Was there ever any resolution to this problem.. ?? Many years ago Tom Caudle who created some of the first torch height controllers for teh DIY masses got together with Art. They worked out teh combination of the Bang Bang method with no accel values as teh only way it was going to work in Mach1 as ANY accel values caused teh Z to be too SLOW in responce to be workable.

So Art adopted teh method of normal Z accel for any NON THC moves in Z and bang bang for THC generated moves. And that method still works fine today for general DIY Plasma cutting.

SO until teh proposed S curve method is tried and proven workable could we please have it teh Mach3 way for now. At least it is workable.

(;-) TP

[cncdrive]

With our controllers it is not Mach3 and not the UCCNC who controls the THC, it works in reverse. The motion controller directly controls the Z axis using the THC inputs to the controller as the control input when there is a THC control and it only feeds coordinates back to the software.
And ofcourse you can set the acceleration parameter as high as you want. If you setting it to a very high value then it will act as if there is no acceleration profile but it will make a speed step jump (bang bang).
S-curve profiling will make this functionality no better. It would only make THC control to work worse because then the Z axis would response slower than it will with standard trapezoidal profile.

[Vmax549]

Hi Balazs I have tired very high accel and it is not teh answer as it effects normal Z moves as well. That is not a good compromise. Even with high accel values there is still the LAG that occurs and effects cutting. Especially thin sheet where extremely fast short moves must be made by the THC function in order to keep up and maintain cutting height. Proper cutting height is teh scecret to good plasma cutting. That and constant velocity motion.

I have been plasma cutting with mach3 for a very long time now using teh bang bang method and have seen ZERO problems with teh process. With UCCNC there seems to be NO sweet spot when using it for plasma cutting.

I have been trying to get UCCNC to cut at least as good as Mach3 so I could convert teh last mach3 machine I use to UCCNC but so far it has not happened to teh point where I can switch over.

Just a thought, TP

[cncdrive]

Hi Terry,

I have to say that your observation is wrong.
When using very high acceleration, let's say double.maxvalue what is the maximum possible value that is the same as if there was no acceleration, because that value is so high that it goes to infinity.
Then there is not even a single step happening with acceleration, but the programmed velocity is output immediately.
And there is no lag, the THC Z movement does not effect the XY motion in any ways, it works fully independently, because in a plasma application the XY motion is controlled by the software and the Z is controlled by the motion controller independently, they have no effect at all to eachother.
And it actually works the same in mach3. We had a large plasma machine machine mostly cutting very thin sheets down to 0.3mm thickness for years without problems.

[beefy]

Hi Balazs,

I understand what Terry is saying. He does not mean that Z moves affect XY in any way. He means that there is only one acceleration setting applied to Z, whether it is for THC moves, or normal moves which are commanded by gcode, jogging, etc.

The way UCCNC is now, if we apply infinite acceleration to the Z, we have to limit the maximum velocity very low to prevent lost steps. This of course means any rapid Z moves are not very rapid at all

What would be great would be to have TWO separate acceleration settings for the Z. One of these acceleration settings is applied to THC moves, and the other setting is applied to all other moves.

That way we can have our infinite acceleration moves for THC control, but with a much reduced maximum velocity.

And for all other Z moves, we have our normal acceleration setting so we can have our maximum motor velocity for Z rapids, jogging, gcode moves, etc.

I believe the S-curve stuff is quite some time away so if the above could be given to us in the meantime, that is the next best thing to the great idea you mentioned in this thread some time ago.

Keith.

[Vmax549]

Hi Balazs, No my observation is not wrong it is spot on. The proof is in the fact that Mach3 plasma and teh Z bang bang setting works just fine for what it does. Thousands of users prove that every day.

It in no way effects teh YX motions just teh Z motions.

Now is it perfect in all situations ? NO , but it is very much useable in most situations that the DIY cutter is going to run across.

I have 2 controllers running on teh Plasma machine Mach3 and UCCNC.(Well actually 3 if you count PathPilot , but Lcnc is in a different world) Everything else is teh same THC, etc I simply uplug one and plugin teh other. There is a big difference in teh cutting quality and teh time the operator has to spend baby sitting teh cutting on the machine. Most here do not want to run teh UCCNC version because of teh way teh Z works when cutting AND material setup using teh Z.

That is simply an observation from both running in the real world side by side. Take it for what it is worth.

(;-) TP

[cncdrive]

Hi Keith and Terry,

Terry, from what Keith wrote your description was not clear at all if you really ment what Keith described.
From your description I understood a totally different thing as what Keith wrote.

The thing is that stepper motors really can jump start, but only under their so called start/stop frequency.
They can jump start up to that frequency, because if you making it to jump up the step frequency above that frequency then the motor might loose steps.
So, it is possible to jump start a stepper motor only under that frequency and so the "bang bang" acceleration is only possible safely up to that frequency.
If you do the 1(t) velocity step (means infinate acceleration) then you have to limit the max. velocity to under the motor's start/stop frequency or you can 1(t) step the velocity to that frequency point and then you have to use an acceptable value acceleration above that. This is a workable solution in real world and could be useful for plasma THC control

The 2 different accelerations with cutting is not a bad idea, or an acceleration below a velocity point and another one above that velocity point is another idea which is implementable and would be usefull for THC control.
However we will soon start implementing the S-curve velocity profiler and with that the zero start velocity (frequency) will be also a parameter.
You can set that value to non zero and then the step frequency will jump the motor to that velocity point without acceleration and then it will use trapezoidal or S velocity profile above that point based on the S parameter value.
So, I see not point in making a temporary half-solution now for this, because the S curve motion profiler will solve this problem also.