Advanced Search

Hello brianm,
Thanks for your answer . I was looking ( a lot ! ) in the web how to modify the thc component .
Basically I understand that may be changed and recompiled for suit the needs , but when I was trying to find more info i did not find nothing for a beginner .
Reading your reply and the reply from PCW ( many thanks too ) I come with a different appproach .
Would be relatively easy use the component input and "read analog values" like usb joisticks but I do not know the resolution and speed of this converters.
Also using Arduino to make the A/D conversion and why not something else like torch on, comparations etc...
There are also inxpensives USB SOUND CARDS that usually dont have ac coupling letting use them for dc A/D conversion ( but I dont know if HAL can use them.

At the end I think I will follow your idea !

Just will make a small external comparator and drive the voltage comparation from a parallel port pin (PWM) to set the voltage from Axis UI.
I will try also an experimental way to have the ARK_OK signal from the torch volt ( when working with a good ark I never saw voltages less than 80 v and more than 150 v )
Another experimental idea is to read the ARC_OK - UP - DOWN signals with just two pins ( when UP and Down signals are active will mean ARC_NO_OK ).

Now the big problem for me !
I NEED/WANT to learn how modify the thc component to do that.
Any help wil be greatly apreciated .
If everithing goes well I will be proud to post the hardware schematics and send some prototypes

Best regards
Roberto

Any V-F converter would work, though the parallel port would limit your control loop bandwidth vs voltage resolution tradeoff somewhat. For example say that you can read the parallel port at 20 KHz so you can reliably sample an input square wave up 10 KHz. A t a 1 KHz servo thread you would only get ~10 counts full scale ( about 3 bits resolution) so you would have to low pass filter the velocity signal get a usable feedback signal (more voltage resolution), and this low pass filter will limit your Z Torch height axis control loop bandwidth.

Hi Roberto,

You do not need the mesa board, I use a parallel port. However, I do not read the voltage. I think the easiest thing for you to do is probably develop some hardware that will read the voltage and then decide whether to move up or down and send a signal to the computer on what to do, (since that is your strong point). That is basically what I do. I bought a THC controller from Campbell Designs and then connected that to the parallel port. I then modified the THC component per the suggestion by andypugh. I am by no means a great programmer, but it was fairly simple and it works fine for me. I can post it later if you like - I do not have it at this computer. One thing I am learning about EMC is you can basically do anything you want.

Hello

I am a new user interested to use EMC with a plasma table THC .
I am tecnichian in electronics more than software developer , so is a liitle bit difficult for me change configuration .
I was reading about the HAL THC component and I am very intersted to use it .
Now my question is ...
It is necessary to use mesa board or can we use a small hardware ( voltage controlled oscillator )
to give the signal on the parallel port for THC control ?
Can we use ( as an example ) spindles inputs to "read" the torch volts ?
If some electronics is needed I will be happy to develop an open source schematic for that

Best regards to all
Roberto

Brian,

At one point in time I did play with an up/down THC component for "smart" THC boxes but never did finish it . The current THC uses a velocity from the Mesa THC card that converts the voltage to pulses and it is what I use on my plasma table. An up/down one would be fairly easy to mod from the current THC comp. Andy is correct and you have to hijack the position and sub your own and lie about the feedback and say it is where EMC wants it to be.

John

It might be simpler to edit the thc component and re-compile it.

You will find the current version at:
git.linuxcnc.org/gitweb?p=emc2.git;a=blo...02fa7bb54473;hb=HEAD

You can edit that, then recompile with
You might need to sudo that, and you might need to sudo apt-get install emc2-dev to get comp to work, but other than that it should be straightforward.

There are bit-type input pin declarations to copy, and the C-code function is quite short.

I was looking at the THC component and that looks exactly like what I need. However, it looks like I need to bring a scaled down voltage from the tip into this component. It does not look like it has a bit for just move up or down if you have an external box that sends those signals.

Would it be reasonable to just use some hal components and "fake" the voltage going into the THC component. Where if my external box says move up I could put a 6 into the THC component voltage and if it says move down I could put a 4 and have 5 when it is the correct height. Not sure if this makes sense, but I would rather use the component if possible.

Thanks for the help.

Have you considered using the THC HAL component?
www.linuxcnc.org/docview/html//man/man9/thc.9.html

Your following error is because the stepgen feedback signal no longer needs matches the command signal. Ideally you need to subtract the THC offset from the stepgen feedback before passing it to the axis-fb pin. The THC HAL component does this, but might not suit your inputs.

I get very confused by the old-school newsig / linksp / linkps format, but let me have a look.

These lines could be combined as:
You are setting sum2.0.in0 to both Zmin and Zmax there, and linking zpos-cmd twice. That is probably an error.

Also, You can set the sum2 pins directly if you want, iecan just be

From following the dallar example, I thought this was basically how it worked... However, I run a program everything is fine, then I press the switch to move up, it moves up a tiny amount and then gives me a joint following error.

The quick and dirty solution would be to loop the axis feedback to itself, ie
net zpos-cmd axis.2.motor-pos-cmd axis.2.motor-pos-fb

But ideally you want to arrange to feedback the stepgen feedback with the THC offset subtracted, which will give you proper f-error detection of the commanded step rate exceeds what the stepgen can follow. (this is set only by the base thread period, and the axis limits are probably much, much, lower)

I am slowing learning EMC2 and have a few questions on THC.

I have a plasma machine that works good. I have successfully connected the ARKOK signal and it works. I have made a few parts. I am now trying to move the Z-Axis up and down with a THC unit. (It is a THC 300 from Cambell designs).

I have tried to follow the wiki and the examples but seem to be missing something.

From what i understand i have to decouple the axis.2.motor-pos-cmd from the stepgen.2.position-cmd

Which i have done with the following lines:

net zpos-cmd <= axis.2.motor-pos-cmd
net zpos-cmd-final => stepgen.2.position-cmd

I then create a move up bit and move down bit. Currently I have them connected to 2 switches I hold in my hand, when it works I will connect it to the correct pins on the THC unit.

newsig TorchMoveUp bit
linksp TorchMoveUp classicladder.0.out-13
newsig TorchMoveDown bit

linksp TorchMoveDown classicladder.0.out-12

I also create the max and minmum distance to move

newsig ZMinCord float
sets ZMinCord -0.15

newsig ZMaxCord float

sets ZMaxCord 1.0

I then set the maximum and minimum distances to move
linksp ZMaxCord sum2.0.in0
linksp zpos-cmd sum2.0.in1
linksp Absolute_Max sum2.0.out

linksp ZMinCord sum2.0.in0
linksp zpos-cmd sum2.0.in1
linksp Absolute_Min sum2.0.out


I then use a mux4 to select. If TorchMoveUp is selected I go to the absolute_Max, if TorchMoveDown is high, I go toward the absolute min.
linksp zposcommand => mux4.0.in0

linksp Absolute_Min => mux4.0.in2

linksp Absolute_Max => mux4.0.in1

linksp TorchMoveUp mux4.0.sel0

linksp TorchMoveDown mux4.0.sel1
linksp zpos-cmd-final mux4.0.out

From following the dallar example, I thought this was basically how it worked... However, I run a program everything is fine, then I press the switch to move up, it moves up a tiny amount and then gives me a joint following error.

A couple of things I am confused on, what is the feedrate it will be moving? The feedrate given in the program?
If I find my min and max from a sum command of the zpos command, will this change as the z pos command changes.

I do not have a float switch, so I set the zero of my torch to the top of the material. Command it to go to a certain height, it pierces through and then i want to invoke the Torch height control.

Any help would be appreciated. Thanks
Brian

Thank you that should come in handy.

WoR wrote:

I just wish it didn't have that many hal files to make it work. The hal's have parts commented out that are placed in other hal's and it just gets confusing lol..

It is, and I found it confusing too. This seems to be a consequence of the way that the wide range of sample configs are assembled.

I would suggest combining them all into one file. In fact in my setup I pretty much ignore the INI file too, and edit direct in one HAL file, just so I can see all the parts.

If you look at the parallel port enable pins in Hal Meter, do they behave as expected?

The "enable" connections on my amps are active-low. There is a +5V supply to the opto-isolator, and the P-Port pin connects to the low side and then current flows through the LED when the pin is low, and not when it is high. This is because P-Ports are a lot better at sinking current than sourcing it.

However, if your Mach setup works then it seems unlikely to be an issue with the external wiring.

This might be helpful: wiki.linuxcnc.org/emcinfo.pl?Parallel_Port_Tester

wish i had good news but i still can't seem to get the motors to enable. I've got the inputs working (estop, etc..) but it seems like i can't get any output. The only action i'm seeing outside of EMC is that when i turn off the E-Stop the THC301d goes from 0volts to the preset volts as it should. So the charge pump must be working. But i still can't get movement in any axis. I even tried starting with a fresh config, leaving XYZA and setting the pins up accordingly. And adding the enables for each axis (this board has 3 enables, two of the axes share an enable so i just need pins 4,5,17 enabled). I thought i had found the problem but it didn't help. in stepper.hal there was:

and i had "xenable" for the pins, not "EnableX" in the hal that sets the pins for the enables

At one point i had got rid of the joint following errors but now i can't seem to do it again. When i manually turn on the torch (spindle) the relay doesn't click either. It just seems like there are quite a few buggy things or maybe i just plain don't know what i'm doing, but i've went threw the config files and i belive i'm starting to grasp the basics of how it all ties together. Problem is its like a giant jigsaw puzzle, you change one part and nothing works until you match it up in the other hal files.

I just wish it didn't have that many hal files to make it work. The hal's have parts commented out that are placed in other hal's and it just gets confusing lol..

But i thank you all for your help and please any more info would be great. I still would like to use emc2 on this machine and future tables we build.

Here is a link to a vid of the machine running with mach (still in baby stages but working ok). This was all home/shop built, no kits or plans.

In the longer term, the ja3 branch of EMC2 will break the implicit link between joints and Axes. However it is not recommended to use that branch at the moment.

For the time being you will need to keep the AXES=4 and COORDINATES=X Y Z A and just ignore the A axis. (it should never move)

You will start in Joint mode, then when you home you can switch to World mode with "$".

Any time the motors get out of step with each other you will get a following error, and EMC will stop, not letting you re-start. The trick when this happens is to switch to Joint mode, where there will be no f-error and then re-home the gantry axis, at which point you can switch back to World mode and there will be no f-error any more. (manually jogging the joints in joint mode so that their positions are very close will also work, if re-homing the axis is not a good plan at the time)

andypugh wrote:

Firstly, you don't have an A axis, so should change the AXES = XYZA to read AXES = XYZ

This appears to be incorrect advice. I am experimenting at the moment with a similar simulated config to try to understand it.

WoR wrote:

t
net xenable => parport.0.pin-04-out
net xenable => parport.0.pin-05-out
net xenable => parport0.pin-17-out

I wasn't clear how many enables you needed for your 4 motors, but if they each have a enable pin, then clearly they need connecting too.

You can check that emc thinks it is setting them using the HAL meter, Halscope or Hal Config viewer in the "Machine" menu.