THC accel/decel

[beefy]

High THC speed is no good if you lose tight control of THC moves, i.e. delays are introduced via acceleration and deceleration, and oscillation occurs.

That's why my suggestion of having an enable check box.

Check box un-ticked, no acceleration applied but user must limit the max Z speed to the stop start frequency. This will be identical to Mach3

Want higher Z velocity with THC control, tick the check box, and increase max THC feedrate, and now acceleration will be applied above the stop start frequency, but at the risk of the introduced accel/decel delays causing Z oscillation due to overshooting THC set point.

The user chooses what works best in real cutting conditions.

That is better than Mach3.

Keith

[Vmax549]

(;-) The only thing I have seen thus far that is better than Mach3 bang bang is a real time tightly controlled servo PID and it is ONLY better with things like currogated sheeting where teh Z has to lift higher per command to follow teh countour of teh corrugation. Other than that I have not seen teh advantage and teh Mach3 BB is brain dead simple to tune to conditions.

(;-) TP

[beefy]

High THC speed is no good if you lose tight control of THC moves, i.e. delays are introduced via acceleration and deceleration, and oscillation occurs.

To further comment on this issue, many users (myself included) have had to reduce their THC FEEDRATE in Mach3 because they got the dreaded "jack hammering" Z action. The Z motor was operating within the start stop frequency of the motor yet still, set point overshoot would happen.

There's more delays than just the 1 milli-second reaction time of UCCNC. The THC itself must filter the voltage and that itself introduces a delay from the real time measuring of the arc volts. Put several delays together and you have a recipe for oscillation:

1). Controller loop reaction time (the logic analyser scans in the previous post seem to show the real reaction time is bit longer than 1 millisecond)
2). Time to filter raw arc volts
3). Time for THC to give the UP/DOWN signal to UCCNC after filtering/reading raw arc volts. My own microcontroller THC is fast in this regard because it's all solid state and TTL logic I/O, but any THCs that use relays will introduce significant delays.
4). Acceleration/deceleration time if we allow Z feedrate to go above stop/stop frequency

It's all coming back to me now.
When I first set up Mach3 on my table, I had a Z rapid speed of 3800 mm/min. I set my THC FEEDRATE to 25% of this value but I got terrible jack hammering due to Z shooting past set point. Only when I reduced my THC FEEDRATE to 15% did the jack hammering stop and THC control was smooth.

I've seen MANY threads with users having this exact same issue.

So that is a simple bang bang Mach3 system with no acceleration applied to THC moves, and operating within the stop start frequency, yet the MAXIMUM feedrate for THC had to be reduced in order to stop oscillation. Imagine what will happen when you go ABOVE the stop start frequency (feedrate) and introduce acceleration delays.

Don't mean to be a pain here Balazs, I'm only using my experience and knowledge to try and help UCCNC be the best controller it can be, and to try and get things right first time, which is also good for you guys doing the development.

Keith

[cncdrive]

Keith,

Yes, I know about control loop stability, I've made my diploma work from them.
Basically this will never be stabile, because it is a P plant with a P controller which pair usually does not provide good stability.

Also sampling times and control loop times can be a problem as you described.

So, yes, the theory tells all what you experienced.

The loop could be further stabilised for example if a PD or PID loop would be used.

[beefy]

Hi Balazs,

Sorry if I came across as trying to teach you about closed loop control, I was definitely not trying to do that. You will be light years ahead of me in that area. I was just explaining my thoughts on why adding acceleration above the stop start frequency and the associated higher Z feedrate could actually cause oscillation under THC control.

Yes, I'd love to have a step/dir driven servo motor with PID control for my Z axis, but I have no idea at present how I'd keep track of the Z position, while a microcontroller was controlling the Z axis step/dir signals. I've never had any formal education with programming, microcontrollers, etc, and so I just head for Google and Youtube to learn stuff, so it can take me quite some time to find and learn the info I'm seeking. Haven't got that time now. It was a very long learning curve just getting my bang bang system working.

Keith.