Rules for THC On and THC Off

Problems and questions
1

Since the rules that are pending do not work well in all cases I have a simple suggestion to Les. Do two simple things like functionality without choosing the rules:
The first is to analyze the contours and holes in the software and to select where the speed is not 100%, or at least 90%, and the monitoring is turned off.
Second - the possibility of choosing the size of small holes where the speed is smaller in order to improve their quality - there is no THC automatically.

Excuse me for my Google - English!

2

You can all this through the post processor which is what I have done

3

Hello, show how this is happening please, because I obviously do not know how it works with the postprocessor.

4

I do not think it can be done in the postprocessor. It can be set at a change in speed or right in the arc or corner itself but that is not enough.
Our company developed a THC that works completely automatically and independently at a speed of 3500mm / min, and it needs code to turn off THC in all places where speed is reduced or increased!

5

This really needs to be done in the control. The control needs to send a signal when the machine’s speed drops below the commanded speed. Trying to predict this in the post or in SheetCam probably won’t be perfectly reliable because it would have to be trying to predict what the control will do.

6

Hello mr Newell, you know that such a CNC does not really exist, so if you bet it as a software functionality you just would improve it is my modest opinion.
Depending on the weight of the portal, except for shapes of size below xx.xxx mm, it will be enough to give an angle smaller than xx.xx degrees with the possibility of correcting in the postprocessor rather than rules, do you understand my thought. Thus, shapes will perceive all lead in / out, arcs, and so on. , as well as all angles.

7

  1. VAD in CommandCNC does exactly this. When the feedrate falls below a fixed threshold, it disables THC to reduce torch dive. When the programmed feedrate is again reached, THC turns back on. The % is selectable on-the-fly by the user.

  2. While not automatic, I put all small holes on their own layer, reduce feedrate, and disable THC on that layer.

8

Hello motoguy, I did not have the pleasure to use CandCNC so I’ll trust you that’s right. But can all people use CandCNC? This is precisely the role of Sheetcam because it combines the needs of customers on different CNC systems. What I’m suggesting is very simple - the software analyzes the DXF shapes and defines the action point itself where it is likely that the speed will be reduced, and these are the small arcs and corners, and the user will only be able to give them the size. The size it will give will set the element itself and an equal distance before and after it.
Thus, by setting once only the action point dimensions, if the calculation is related to % of speed F, it will work on each machine and at each speed.

9

Actually several can. For example LinuxCNC can and I believe Mach3 can also be made to do it with a small amount of external hardware.

What I’m suggesting is very simple - the software analyzes the DXF shapes and defines the action point itself where it is likely that the speed will be reduced, and these are the small arcs and corners, and the user will only be able to give them the size

It isn’t that simple. You need to be able to predict what the machine is going to do. This will vary depending on how the machine is set up.
You can get close with SheetCam’s rules but it will take a bit of experimenting to get them set up to suit your machine.

10

I do not say it’s easy Les. If it was easy, we’d all do a program like Sheetcam and we would not write in your forum.
I am able to predict the behavior of the machine - no machine can make a right angle of 90 degrees without slowing the speed before it and increasing it after it! Thus, the inertia of the process can be easily measured by setting a motion giving a square with sides of 100mm. The run time will give us the average speed or the average time in which the machine is not 100% of the set speed. The same 200% speed of the same figure will show an exponentially increasing time through which the speed is not 100%.
All the time my desire is to achieve the highest possible automation in code creation. In our THC controller we have worked on any sheet metal, tension and speed without any correction and communication. Only with the permission to work at speed 90-100%.
With Sheetcam’s rules, I have not been able to work correctly in various forms so far and always have to set a manual action point to fix things.

11

Both Mach3 and UCCNC have this functionality built in. No doubt there’ll be other CNC controller software that also has this functionality.

Just because it MAY be possible for Les to do it in Sheetcam does not mean it is worth his while. This is the first time I have seen someone requesting this functionality be put in Sheetcam when several CNC controller softwares already do it.

Remember this is a business for Les and there needs to be a market that will give him a financial return on whatever time and effort he puts into something.

My guess is there is a lot of programming mathmatics involved in such a modification to Sheetcam. Have a look at a trigonometry book to see what’s involved in calculating arc lengths, travel direction, etc in 4 different quadrants, then add in calculating acceleration/deceleration and working out distances along the lines/arcs of the cut path to put in a THC ON/OFF code. After such code is developed there will no doubt be bugs found which need to be traced and corrected.

If Les thinks this will not make any difference to his sales, then in a business sense it’s a complete waste of his time, and he could be focusing his efforts on something else. I’ve been in business and you learn not to put time and effort into something that gives you very little or nothing back, or worse it costs you because you could have been doing something much more productive.

I’m also developing my own THC and so far have not had any problems using Path Rules and Action Points to insert code wherever I want it.

12

Strangely, what you wrote seems to be Les Newell’s secretary. All that I have written here is just something I suggest I do not want to force anyone to do something they do not want. Besides, I can make this offer to the Chinese competitor of Les, but I do not like anything Chinese except their food.
The way you write about Mach3 and the UCCNC means that you do not know them deeply. Only a person who has worked enough and has survived Match 3 knows that this is an amateur level program. And about the UCCNC please if you have the opportunity to make a file you want to cut a plate with a few holes where lead in want to be perpendicular. Then you will find out how many times they have to go before their program becomes semi-professional.
I’m interested in seeing your THC and how and at what speed it works. What do I mean? I can offer you the THC that works so slowly that you will not have to exclude it in the corners and demonstrate on a thick flat sheet that it seems to work. But it will not be right?
And one more thing, as you obviously do well with the rules in Schechtam, would you have lived here a set of rules that work correctly without corrections in all the shapes you cut.

13

Sounds like you did not like something I said. And no, I’m certainly not Les secretary, I’m not cute enough.

You say I do not know Mach3 or UCCNC deeply. That’s very interesting coming from you who said:

__

you know that such a CNC does not really exist

when indeed there is. Seems like you do not know them deeply either.

You are correct, I do not know Mach3 deeply (depending on what you mean by deeply).

However, when I first abandoned Mach3 and bought UCCNC I did a great deal of testing. I programmed microcontrollers (no, not Arduino) to inject my custom test signals into UCCNC. I used a logic analyser to test the timing of many inputs and outputs, including the anti-dive which was activated by a slowdown in feedrate. Knowing the stepper steps per linear distance travelled I could see at what speed the anti-dive was activating and de-activating. It worked great and accurately.

I further tested the timing of various synchronous commands (like the THC on/off output for example). The timing of everything was great. Where a bug was found, Cncdrive quickly corrected it. Synchronous torch on/off outputs work fantastic without any interruption to the step pulse stream as proven by the logic analyser testing. I would even program the code an exact known distance and verify the accuracy of UCCNC by counting the step pulses shown in the logic analyser results. UCCNC was pretty much accurate to the step pulse.

So basically I have no idea what else I need from UCCNC. Perhaps you can explain in detail what UCCNC cannot do. And as far as perpendicular leadins on holes (starting at the centre of the hole), that’s exactly what I do when plasma cutting holes. Les added that as a feature some time ago.

As far as I am concerned I know that UCCNC has good control for plasma cutting. I have no idea what “professional” top end requirements you have but it seems like you need to be speaking to the likes of Hypertherm for top end plasma related cam, or perhaps even Siemens or Mazak or one of the other big CNC boys. Then you will get the top dollar professional CNC software that may keep you satisfied.

If there is some big untapped market for your ideas for plasma related cam software, then specify what they are and why they will sell so well. I personally do not understand the need for your wants. If you can explain to me why your “predictive” ideas are better than a built in anti-dive that has proven to work good with testing then I will bow down and consider you my teacher. Until then I remain your skeptic.

At present my THC is not “fast”. I have no need for a PID controlled servo motor driven THC. One day if I’m cutting very thin, quickly warping metal at high speed, then I may need one of those. But all I’ve ever done is a minimum thickness of 3mm and super fast THC has never been needed.
In any case you often have to dampen down the THC response based on the material you are cutting. It’s not one size fits all.

I’m more interested in overall system functionality, like the ability to fire the torch during motion (without motion pause), turn if off during motion, semi-automated mid-cut restarts with torch refire during motion, closed loop hole centre marking at any current level (regardless of set amps) with closed loop control for precise timing to within a hundredth of a second, auto torch refire, puddle jump at pierce point, etc, etc.
And everything controlled from the PC screen. My “slow” THC is just a “black box” that communicates via a noise resistance RS422 connection with the CNC controller, or direct I/O connections for time critical communication.
And to top it off I’ll have my touch screen pendant with as many functions as I desire on an effectively unlimited number of “pages”. Already done a basic prototype screen touch pendant and I like it.
So I’ll pass on the “fast” THC for now and have awesome functionality and control instead, which for me personally is of much greater value. I’m designing my system for what suits me, not to shout out how fast a THC I have.

For plasma cutting Sheetcam does everything I need and so does UCCNC. If it’s less than semi-professional as you suggest, then less than semi-professional is good enough for me. And I also consider them great value for what you get.

14

stivemaster please keep this polite. I won’t tolerate attacks on members of this forum.

I am able to predict the behavior of the machine - no machine can make a right angle of 90 degrees without slowing the speed before it and increasing it after it! Thus, the inertia of the process can be easily measured by setting a motion giving a square with sides of 100mm.

Yup, that is very easy and an ‘on all corners’ rule will do the trick.

Let’s take a more complicated example. Say you are cutting a complex curve. As g-code can only handle arcs and lines these complex curves have to be broken into a mix of short lines and arcs. How do I predict how the controller going to handle this? A controller with no lookahead will nearly stop on each join between the lines/arcs. A controller with 1 segment lookahead may blast around one corner but slow right down for a series of line segments. Even controllers with full lookahead use different algorithms and will cut at different speeds. If you take four different controllers and configure them for exactly the same speeds and accelerations they will all cut at different speeds on complex shapes. This functionality belongs in the controller, where the amount of slowdown is known under all conditions.

Rules are an approximation but in most cases the approximation is good enough. What issues are you having with rules? What situations do you find where rules don’t help?

15

I want to apologize first, I did not intend to offend anybody, especially if it offers something constructive. I hope the mancavedweller will forgive me.
Please, if my tone is considered rude to give it a bad translation.
On the occasion of the UCCNC I will give a link and only say that they have codes that work in sync really BUT they MUST BE ACTIVATED AT THE PRESENT TIME OF THE G CODE !!!
And if you read this, what you will think about this software?
I was not lucky to buy 2 controllers from them and put them on two machines. And then I made a trail to them! In case of a problem like mentioned in the second link I did not know what to do. I was reading a CNCDrive and I was told to just make the software up to the last version. Well, I did not do it. I just forgot to write that my key will not work on the new version . I wrote again a request to generate a new key, and it is right that the key is issued to the owner of the machine who can update himself in the future. I have no answer for 3 months.
Regarding the rules I can not describe exactly what the problem creates, just the rules do not work as a system but they interfere with each other. In the end, it is not just for me, even in the example mentioned, people mention this. For me, I only have to use the rule for small shapes and manually set the action point at the beginning, in the corners and with small arcs. If I include these rules, always at the end just before M5 is running THC and this is a mistake.

16

And if you read this, what you will think about this software?

That is an annoying bug but it is easily fixed in a post processor. Use:
post.SetOptions(post.ARC_SEGMENTS)
This will break circles into at least 2 arcs.

Regarding the rules I can not describe exactly what the problem creates, just the rules do not work as a system but they interfere with each other.

I put quite a lot of work into making the rules not interfere with each other. They can (and often do) overlap but SheetCam does it’s best to sort out the overlaps in a logical way. For feed rates, the rule with the lowest feed rate will have priority. For THC on and THC off codes only the first and last THC on/off codes will be used.

Make sure you only have one code snippet to turn THC on and one snippet to turn THC off. If you have multiple code snippets for these functions you can get problems when rules overlap.

17

If I had a simple straight line 100mm long and I used Path Rules to insert M205 (THC on) 10mm after the start, and M206 (THC off) 10mm before the end, the generated gcode will put M205 at the 10mm point and M206 at the 90mm point on the line.

There is no “timing” involved, it is just generated gcode and the synchronous commands being placed in the gcode at the correct positions. In UCCNC the gcode commands are then sent to the motion control in a transmit buffer, and the “real” timing is done there in a fast ARM microcontroller.
In my experience so far, synchronous commands do exactly what you expect them to do, and where you expect them to do it.

As I mentioned earlier, my testing showed there was great accuracy in the positional placement of the synchronous commands and very accurate execution of those synchronous commands, without any interruption at all to the step pulse stream. If Sheetcam Path Rules generated gcode to put a synchronous command at say X10.41 mm, then that is exactly where that synchronous command would be placed and executed.

And as mentioned earlier I tested this accuracy by recording the step pulse stream and the synchrounous O/P with a logic analyser, and painstakingly counting the quantity of step pulses to get an idea of the accuracy of the operation of the synchronous command. Accuracy was very high, to about plus or minus one step pulse.

18

Unfortunately, I would not say that this will help even I think it will worsen the situation. Holes should be treated as holes or even shapes and holes like arcs because in the case of offshore, it should be off-off and other off-on. I get a picture to show what I want Sheetkam to do.
The purpose is if I have set only one variable, we give (minimum distance) = 15mm, the software will define all zones with R <15mm including holes, arcs and angle (in the picture are red) and add where necessary 15mm are zones with blue) before and after these zones, where THC does not work again.
Primer za Les-Model.pdf (4.86 KB)
Dear mancavedweller, we are going back to the beginning because we are talking about the same things, but you just do not want to find out what I suggested. That their codes work well does not mean that they turn alone when needed !!!
And I’m sure you have what to say constructively if you think about it because the ability to explore UCCnC actions means you are the right person , in my opinion

19

Looking at your pdf it looks pretty straight forward to do that using a few rules.

20

I have already described the problem - in the end, just before M5 there is always a code for the release of THC.