THC Design Question for you Jim

[beefy]

High Jim,

I'm working on my own THC design as I get time. It's just a basic type that sends out THC UP & THC DOWN signals, no fast PID or anything (yet ).

Do you have any recommendations for what rate I need to be able to send out the up/down signals. I mean with electronics and microcontrollers I could send them out 500 times per second, or much greater, but a mechanical Z-axis simply cannot take advantage of that speed.

Do you think 30 per second is too slow. That would THEORETICALLY mean you could send out a THC UP signal and the Z starts moving up. 1/30th of a second later you send out a THC DOWN signal and the Z must reverse direction and starts going down. In one second that UP/DOWN movement could be repeated 15 times, and the Z must be CAPABLE of mechanically doing it.

So basically I'm wondering, physically, how fast a well made Z axis can react and thus how many THC UP/DOWNS per second should I design for. Or put another way, what rate would be needed for good THC control.

I'm sure there's a lot of completely un-necessary THC specifications out there claiming super fine volts resolution, and super fast correction outputs, yet for all practical purposes they don't make any difference above certain PRACTICAL values because the limitations are more mechanical, such as rigidity, inertia, etc.

Cheers,

Keith.

[jimcolt]

While industrial machines utilize z axis / height control systems with 1000 ipm up and down ability.....this is primarily to retract the torch out of the way during travers to the next cut (collision avoidance) without having the effect that slow movement would have on productivity. These same systems also have the ability to make "smart" decisions to ignore retract based on the part geometry and where the next pierce will be....all in the name of saving precious seconds of production time.

On most small shop machines productivity is of lower importance.....good cut quality and long consumable life is ranked higher. The z axis needs to go fast enough to maintain +- .005" of the correct height under normal plate warpage conditions. If you plan to cut corrugated roofing...it will need to go much faster! So...depending on the speed you need to operate at and the height differences you will see in materials and your slat bed (remember the slat bed gets worse the more you use it)...then you need to do the math to figure out minimum speed. Some CNC outputs also are sluggish in the respect (Mach 3 based THC).

Here's what I like to see in minimal height controls:

1. Smooth, fluidic motion. If the up/down is jerky, you will see deep lag lines in your cut.
2. At least 50 IPM in the up direction. It is nice to be slower on down direction as then things like kerf crossing and other anomalies in the cutting process won't cause material collisions.
3. If the feedback loop and electronics are too tight....expect a sewing machine motion. Have the ability to adjust sensitivity or gains to filter this out. Height controls use arc voltage feedback....which is very nasty wildly fluctuating voltage. The height control generally needs to average it's readings over time.....I often see averaging every 40 miliseconds on height controls that work well.

Jim Colt

[beefy]

Thanks so much Jim,

that magical number in the last sentence was what I was after. At every 40 milliseconds that's 25 times/sec, not too far from the 30 I mentioned.

Yes I'll be using software averaging on multiple readings. That's after electronic filtering before the AD converter. I remember what you told me also about volts resolution too so I'll just go for 1 volt increments, not 0.1 or anything like that. Then I'll chuck in void lock and dead band adjustments.

If it all works, that is

Cheers,

Keith.

[tcaudle]

If you are going to use MACHthen the sample rate is of less consequence. If you intend on designing a full stand alone torch height control then you WILL have to deal with servo loop feedback and PID (at the least PI) The more samples you take per second the less effect a few bad samples will have if you average them. Average them too much and your response will suffer. As soon as you get loop instability you get Z oscillation.

[beefy]

If you are going to use MACHthen the sample rate is of less consequence. If you intend on designing a full stand alone torch height control then you WILL have to deal with servo loop feedback and PID (at the least PI) The more samples you take per second the less effect a few bad samples will have if you average them. Average them too much and your response will suffer. As soon as you get loop instability you get Z oscillation.

Thanks Tom,

yep, just using Mach3 and the basic THC UP/DOWN inputs for now. Gotta be easy on myself with this one. I don't tend to do any high speed or corrugated cutting at the moment in any case so should serve me fine.

I WAS going to use an analog low pass active filtering circuit but was a bit shocked at the phase delay, so after a bit more research it seems a basic RLC filter + high sampling rate and digital filtering is a more viable method. Pretty much learning all this as I go.

Keith.