Plasma Questions

More
18 Jan 2017 12:34 #86049 by rodw
Replied by rodw on topic Plasma Questions
I think you and I are both finding a number of weaknesses in the LCNC Mesa/THCAD/THC model. I have an external THC which might have been much simpler to implement.

Chainsaw and sandpaper approaches aside, some things I think about
  • Why we are stuffing about setting cut voltages when we are really interested in cut height. Surely we should be sampling the voltage once we are at cut height during the leadin.
  • Lack of PID control
  • Not waiting for commanded moves to complete before the next adjustment
  • Inability to sample voltage fast enough to apply any signal processing (40 usec only gives 5 bits of resolution)

I think the whole thing needs redesigning and eventually I'll have a go at it.

Using an Arduino is viable to read voltages at 40 usec. I have not pulled one out to play with but it seems they are able to sample voltage reasonably accurately at 20 usec so sampling using a ISR at 40 usec should give enough time to get other stuff done without lockup from ISR overruns. I think it would be viable to generate a frequency to send to THC to send voltages back to LCNC as the generated frequency would only be changed every 40 milliseconds.

Please Log in or Create an account to join the conversation.

More
18 Jan 2017 12:50 #86050 by andypugh
Replied by andypugh on topic Plasma Questions

  • Lack of PID control

  • PID shouldn't be hard to set up. But it might be fun to tune, and PID enable/disable conditions would need some thought.

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 16:57 #86064 by grijalvap
    Replied by grijalvap on topic Plasma Questions
    I agree with most of the points you make here
    however, I think Arduino by itself is not reliable enough for the application
    at any point I asked about the following board from Mesa.

    store.mesanet.com/index.php?route=produc...83_88&product_id=126

    I think a reimplementation of the THC component need to be done to add PID and also decoupling from the Z axis
    and let it work as a standalone entity that can receive all parameters need to work.

    At the beginning, I was thinking to use a raspberry or beagle bone to implement the THC using machinekit, but this lead me back to LINUXCNC, and I realized that it can be done on the same machine using a new component.

    One of the advantages of having it as a stand-alone entity is that the g-code gets less complex, ant the implementation easiest (once the proper component is developed)

    Now I using a good quality stand alone THC plasma, the communication is RS232 but I have a lot of delay in the communication

    Also used the proma but it is also not what I need

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 17:03 #86065 by islander261
    Replied by islander261 on topic Plasma Questions
    rodw

    The THC mode of operation you describe is one I call "arc voltage sampling". This mode is implemented in most high end commercial THCs. Most small shop/hobby THCs don't offer it. You will find that I described it and my desire to implement it in one of my previous posts. Also if look at the posted screen shots of my GUI modifications you will see a radio button that at this time does nothing to select this mode of operation. The voltage sampling wants to take place after torch reaches programmed cutting speed plus any additional selected servo delay. This is the usual problem for most small shop/hobby THCs is that they are unaware of actual machine velocity on a real time basis. I have note from a HT guru on another forum who states that for good THC you need a servo loop with at least 25Hz bandwidth. I expect it is over kill to go much faster at typical air plasma speeds. You can download the manuals for HT THCs from their website if you want to check out the features and control modes. One of your fellow country men has done a lot of THC work with the UCCNC controller, He goes by beefy over at plasmaspider, I think he is in the southern part of the country.

    I still intend to use the stock THC component at first. If it doesn't work well then I will investigate changes. I am sure it work better than my existing stand alone Torchmate AVHC box! The corner hold ability alone will be a big bonus.

    John

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 17:32 #86068 by islander261
    Replied by islander261 on topic Plasma Questions
    Hello

    My present THC uses SBC very similar to an Arduino. It uses an Atmel Mega 8 micro processor. So I am sure that the Arduino will be more than good enough as a low cost development platform. Of course one should only run fully compiled code outside of the Arduino operating system for a production machine.

    I originally bought my THCad card to use with an Arduino to make my self a new THC before switching to the Linuxcnc route. The main problem was I couldn't figure out how to get real time machine velocity information out of my present controller. When the manufacture finally admitted that it couldn't be done without access to the source code I gave up this approach and went with Linuxcnc.

    Yes, the stand alone THC for machine Z axis control makes for very simple G code and a very simple Post Processor.

    John

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 19:05 #86079 by PCW
    Replied by PCW on topic Plasma Questions
    The 7I87 has been used for THC applications, its not quite as rugged as the THCAD
    (more capacitance from input to output because of the isolated power supply transformer)
    but has better resolution.

    It also should be possible to get min/max and average readings for THC sampling use with a 7I87
    (it already samples all inputs at about 50 KHz each but currently just presents the average values
    to the host)

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 19:16 #86082 by grijalvap
    Replied by grijalvap on topic Plasma Questions
    I agree a good AFE (Analog Front End) need to be designed including Voltage divider.
    but the resolution and speed are good.
    the other possible option can be, design an ADC compatible with Mesa serial port

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 21:29 #86095 by rodw
    Replied by rodw on topic Plasma Questions
    Great to see so many people get involved. Here is the journal article I found.
    app.aws.org/wj/supplement/WJ_1992_08_s277.pdf
    It took me a few hours of methodical study before I understood how simple the final algorithm really is. It outlines the rationale behind their choice of 40 msec sample period (so that they can account for the 3 phase ripple.) on a single phase machine, the ripple period should be 100-120 depending if you have 50 or 60 Hz but I doubt you'd want to get down so quick.

    The 7I87 has been used for THC applications, its not quite as rugged as the THCAD
    (more capacitance from input to output because of the isolated power supply transformer)
    but has better resolution.


    I don't think the lack of robustness would be an issue in a modern plasma that does not use HF start and includes an internal voltage divider. Mine has a blow back start design and a configurable 16:1 or 50:1 voltage divider.

    It also should be possible to get min/max and average readings for THC sampling use with a 7I87
    (it already samples all inputs at about 50 KHz each but currently just presents the average values
    to the host)


    It would be awesome if Mesa could provide software drivers that presented LCNC pins for min/max and average volts for a multiple of the given frequency. Another alternative would be to leave the sample period where it is and allow the number of samples included in the min/max and averages to be defined. either way, we need to get the min value every 40 ms.

    A couple of other observations.
    There is already a complete config for Plasma with no Z axis stuff in it for Independent THC. Tomylight reposted a link to it recently in response to a question of mine.

    The independent THC I have also requires connection to the X & Y pulse generator signals. I assumed this is so it can monitor torch velocity to solve the problem islander had with his arduino.

    Keep the discussion going guys.

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 21:49 #86096 by rodw
    Replied by rodw on topic Plasma Questions
    I forgot to mention there is no real difference between an Arduino compiled binary and code output from the Atmel compiler. Remembering that the feature of the Atmel chip used by Arduino is that they support in circuit programming. eg. You download new firmware when the circuit is powered up. You can take an Arduino binary and download it using the Atmel programmer. I've done this several times. You can also program an Atmel chip in an Arduino and transplant it into a non-Arduino circuit and it will function accordingly. Yes there may be some overhead in the libraries but when you get to interruprts and Analog sampling frequency, there is not the overheads you might think. I would not use floating point anywhere though!

    Please Log in or Create an account to join the conversation.

    More
    18 Jan 2017 21:53 #86097 by islander261
    Replied by islander261 on topic Plasma Questions
    rodw

    I think everyone that has considered the THC problem has looked at that AWS article. From what I have been able to tell unless you are cutting quite thick plate the arc source point movement get swamped by the accumulated system errors and filtering. This is a good intellectual exercise, I was advised by HT guru to not worry about it, a working solution is much simpler.

    Just monitoring the the step output of the controller only provides the information to solve the speed problem on slow down at the end of the cut. It has no way of knowing what the target speed is during acceleration. Of course you can just use a fixed time at initial cut height or make a bunch of assumptions based on recent motions to get around this. Now if the THC can accept an analog or digital representation of the Feed Rate then it can work. With LCNC this is no problem, but with most closed commercial system and MACH your out of luck.

    John

    Please Log in or Create an account to join the conversation.

    Time to create page: 0.149 seconds
    Powered by Kunena Forum