LinuxCNC, Plasma, THC questions

[Greolt]

Long time Mach3 user, via LPT, on Mill, Router and Laser.

However I am in the final stages of building a plasma table and the particular requirements are a little different. So I am researching LinuxCNC and in particular how it incorporates THC.

With Mach3 when the controller is given up/down signals by the THC (eg Proma) the LPT driver takes care of these movements.

The disadvantage of that is that those Z axis moves are bang bang. That is, there is no acceleration ramp. Full go or stop.

This has the problem that the Z axis performance, and hence the THC performance, is limited by that lack of a proper acceleration ramp.

For example my Z axis can happily run at 1800 mm/min with acceleration ramp but only 600 mm/min without it.

So to my questions,

Does LinuxCNC, when responding to up/down signals from the THC (eg Proma), does it incorporate acceleration ramp into those moves?

Is there an advantage in respect of the above question, to using Mesa THCAD?

Another Mach3 disadvantage seems to be latency in the response to up/down signals. Is this better with LinuxCNC? Again is this better or not with THCAD? Or Proma?

I hope I am asking this in the right place.

Thanks

Greolt

[beefy]

Just some food for thought regarding the THCAD and Mesa products.

My THCAD has now been on order for SEVEN WEEKS and they haven't even put it in a box and posted it out. On top of that I've heard a rumour that Mesa is trying to get away from us DIY little fish and migrate to the OEM big fish.

The only time I can get Mesa to reply is when I get the shits and go on the Linuxcnc forum to complain that a single stock board is not being posted out for weeks on end and Mesa won't answer emails. I was there 2 weeks ago and the Mesa guy did not say that rumour wasn't true. Surely it would have been a priority to say such a rumour was totally untrue ??

Of course the die hard Mesa supporters say that rumour is not true, and they have a hard time beleiving Mesa won't reply to emails, despite others having exactly the same problem (single in stock boards not being posted out and emails being ignored).

The talk at the moment is that Mesa have a lot of OEM orders to see to. Seven weeks and counting to put a single PCB in a box and send it out, then ignoring customer communication, is definitely putting us little fish right at the bottom of the pile. Seems to support the rumour about Mesa wanting rid of the little fish.

I shudder to think that I could be waiting months, not weeks to get a replacement PCB from Mesa to get a broken system working. May as well just shove your cnc machine in a corner for a long while.

One more point regarding the THCAD. It is not a torch height control board like the Proma unit, it is simply a voltage conversion card which outputs a frequency proportional to the voltage input. You would need one of the other Mesa boards to read the output frequency then you've got to know how to use that in the Linuxcnc system (over my head). I'm working on programming a microcontroller to read the frequency and give the THC UP/DOWN signals. Also looks like I'll have to try and make my own voltage to frequency card seeing as I don't know how many more months Mesa will take to send the THCAD out.

Keith.

[Greolt]

Thanks for that Keith.

My idea would be for a friend to front up to the door in Richmond with cash in his hand.
Hopefully that would do it.

But before I get anywhere near that, I need to educate myself on what THC on LinuxCNC looks like.

I won't move away from Mach unless there is real advantage. Hence my questions above. As a starting point.

[acourtjester]

Interesting I went the their site and this is what it shows.

Brand: Mesa Electronics
Product Code: THCAD-300
Availability: 9

I went to the "better business bureau" and this is what I found

This Business is not BBB Accredited

Mesa Electronics
Phone: (510) 223-9272
Fax: (510) 223-9585

[islander261]

Hi

I am presently working on modifying the current LinuxCNC THC and probing support to be more like members here are a custom to. This is a long term project for me as I am not a software wizard. The present LinuxCNC plasma uses the Mesa Electronics 5i25/7i76 (if stepper driven) combination with the arc voltage measured by a THCad card. There are two versions of the THCad card one is 0-10V input and one is 0-300V input. You need to use the 0-10V with RF filtering and a voltage divider if you have a HF start power supply. Almost any modern blowback start plasma can use the 0-300V one without a divider. If your plasma unit has the right built in divider the 0-10V should work.

Mesa Electronics has been around for a long time. At the present time their shipping is painfully slow. I waited 6 weeks for my THCad card. The good news is that they didn't charge my card until it shipped. You have been warned!

Look at these links:
http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Plasma

http://wiki.linuxcnc.org/cgi-bin/wiki.p ... smaMachine

http://www.gnipsel.com/shop/plasma/plasma.xhtml

Look around JT's site as there is a wealth of LinuxCNC info there.

John

[beefy]

Interesting I went the their site and this is what it shows.

Brand: Mesa Electronics
Product Code: THCAD-300
Availability: 9

I went to the "better business bureau" and this is what I found.

This Business is not BBB Accredited

This is what a few people are finding. Their website shows plenty availability yet they can't even stick one in a box and send it out. And when you TRY to contact them to ask why they are having such a hard time doing that, they completely ignore you. That's the bit that does the damage. They don't even have a stock standard reply explaining why it takes so long to send out an in stock part.

I believe Tormach are now using Mesa products and that has resulted in Mesa ignoring us little fish.

[beefy]

Mesa Electronics has been around for a long time. At the present time their shipping is painfully slow. I waited 6 weeks for my THCad card.
John

The shipping is only supposed to take 6-10 days to most international locations. The painfully slow part is that they simply don't see to your order for many weeks, even when the part is in stock.

Good luck with your Linuxcnc adventures by the way. I started looking into the same path but got a bit frustrated with always searching for answers to the many questions I had. Lcnc is supposed to be great stable and powerful software once you get to know it.

Keith.

[Greolt]

Really makes you wonder if you want to get in a postion where you are dependent on a company that is reluctant to deal with you.

However I will repeat these questions,

Does LinuxCNC, when responding to up/down signals from the THC (eg Proma), does it incorporate acceleration ramp into those moves?

Is there an advantage in respect of the above question, to using Mesa THCAD?

Another Mach3 disadvantage seems to be latency in the response to up/down signals. Is this better with LinuxCNC? Again is this better or not with THCAD? Or Proma?

And now I have another,

Is the THCAD dependant on using a Mesa IO card?
I was under the impression that the LPT would be OK with it.

[Greolt]

Does LinuxCNC, when responding to up/down signals from the THC (eg Proma), does it incorporate acceleration ramp into those moves?

OK, I got this question answered on the LinuxCNC forum.

Linux step generator handles all stepper movements, no matter if THC or otherwise.

So ALL moves include acceleration ramps as configured.

That sounds to me like a big tick for LinuxCNC over Mach3 regarding plasma and THC.

[tcaudle]

Does LinuxCNC, when responding to up/down signals from the THC (eg Proma), does it incorporate acceleration ramp into those moves?

OK, I got this question answered on the LinuxCNC forum.

Linux step generator handles all stepper movements, no matter if THC or otherwise.

So ALL moves include acceleration ramps as configured.

That sounds to me like a big tick for LinuxCNC over Mach3 regarding plasma and THC.

I think I would stay with MACH3 if you goal is to get cutting any time soon. LINUXCNC is a 'rool your own" general CNC control program and it is not a trivial deal to interface things like a THC using the UP and DOWN (from something like a Promo). Yes, the motor control has velocity and accelertaion settings but its not a feedback loop (PID) type control either. It just puts in the set acceleration so you can ramp up to higher speeds but You have to think about what that means. The external THC is just going to send an UP or DOWN and the axis is just going to move within the confines of its tuning. The UP down commands from the Promo are already relatively slow because the ouputs are relays. Relays do not turn on or off nearly as fast as something like an Opto isolator. The lack of acceleration in MACH on the THC channel is there because most of the moves are very short and in opposite directions (lots of upand down the faster you go) There is little time for accleration unless the move is longer. You just can't make simple UP and DOWN behave in a way that will make it stable. Converting the voltage to frequency then doing the same UP and DOWN improves very little . It HAS to be proportional so the more deviation from the Preset, the faster you move and vice versa. You are building a servo feedback loop and the speed and response in every link of the loop defines the overall loop response. Translated: if anythng in the loop is slow or introduces latency then the entire loop is effected. It can never be faster than the slowest component. I think you will be surprised how little the improvement is

LINUXCNC is a rich and powerful system but the vast majority of users are NOT gantry style routers and even fewer Plamsa tables. You pretty much have to figure out how to hang all of it together. Its not as simple as just picking two inputs and coupling them to an external THC and off you go. You have to find or write a HAL component to know what to do with the inputs There are just a lot of things it does
not support. It uses a toally different set of G Codes. Download the documetation and the G and M code manuals. You cannot simply write a script in VB or LUA and access internal functions. What about things like a "touch-off" sequence? Homing is done completely different than MACH. What about showing the status of the inputs? If you are used to running a THC that only has the UP and DOWN on it, those issues become less important
but LINUX and "simple" probably don't need to be in the same sentence.

I suspect that MESA as a board design and manufacturer, is not setup to do support to individuals trying to roll their own system who are not already LINUX savy. Selling board products to a novice DIY crowd is a totally different process than selling to an OEM with programmers and patience. At some point you have to decide what your target market is. MESA needs to double the price of their individal boards and hire a fulltime support person or simply refuse to sell less than 10 boards.

The THCAD card is just a simple voltage to frequency converter that can then be read by the FPGA. What you do with it is then up to the programmer and the FPGA interface into LINUXCNC . It is NOT a THC. It is no different than the PWM module that takes analog torch volts and converts it to Data (PWM) to be used by the real THC circuits. Its just one rather minor piece of the puzzle

[beefy]

Is the THCAD dependant on using a Mesa IO card?
I was under the impression that the LPT would be OK with it.

The THCAD default output frequency is 100,000 to 900,000 hz representing 0 - 300 volts, but there are jumpers/switches on board that can be used to divide that frequency down. So it's a case of is the parallel port fast enough to read the frequency and does Lcnc have a means of interpreting that frequency and using it.

I'd ask the question on the Lcnc forum. I'm sure someone would know there.

Keith.

[Greolt]

Thanks tcaudle for taking the time to write.

All good info. In my particular case the way LinuxCNC does up/down movement MIGHT just be enough to get over the line from inadequate to adequate. Remembering mine is a home made table only for hobby use. (and on a hobby budget)

I will pursue THCAD info over on the Linux forum

[vmax549]

A really good PLASMA setup would be the UC100 running under UCCNC software. Their THC UP/DOWN runs from inside the hardware controller and THAT is about as good as it can get. Very fast and responsive.

Now what would be realy nice is to actually RUN the entire THC control from inside the UC300 chipset as it has analog inputs and outputs so you could route the arc volts and arc ok signals directly into the controller. Then the presets and adjustments could be made on the fly directly from UCCNC across teh USB/ETH com. This would do away with ANY outside THC boards. All that would be needed is an interface conditioning board OR a simple COM interface to the HT plasma units CPC port and Machine Com port(modbus ASCII)

THAT would be the Cadillac of Plasma controllers . (;-) Investigated that right now.

There ARE other options out there. (;-) AND it is ONLY going to get better.

Keep Your eyes on UCCNC. Interesting things are on the horizon.

(;-) TP

[Greolt]

A really good PLASMA setup would be the UC100 running under UCCNC software. Their THC UP/DOWN runs from inside the hardware controller and THAT is about as good as it can get. Very fast and responsive.

Does that mean those THC up/down Z axis moves include proper acceleration ramps? Unlike Mach3 LPT.

[tcaudle]

Now what would be realy nice is to actually RUN the entire THC control from inside the UC300 chipset as it has analog inputs and outputs so you could route the arc volts and arc ok signals directly into the controller. Then the presets and adjustments could be made on the fly directly from UCCNC across teh USB/ETH com. This would do away with ANY outside THC boards. All that would be needed is an interface conditioning board OR a simple COM interface to the HT plasma units CPC port and Machine Com port(modbus ASCII)

Piece of cake Terry! All you have to do is figure out how to read the single register for the analog inputs fast enough to have a usable sample rate then pump the thousand or so values per second back across the USB interface and not interfere with the motion commands coming the other direction. Unfortunately USB being a polled protocol is not bidirectional. Then all you have to do is take that data and write a .NET plugin to convert it to moves for the Z but that would have to go back down the single pipe. Then you need another plugin to be able to display and change settings in the UCCNC UI. for the THC settings.

As long as you work within the software loop of the controller runnign in Windows you have to deal with the inherent layers of overhead and in the end you have the same issue as MACH: The UP / DOWN of the Z is under its control and not the hardware. In the end all you gain is fewer external cards at the expense of achieving little or no dynamic improvement in response.

If you take the THC servo loop outside the control of software running in Windows and sending "jog" commands to a captive axis, the response improves ten fold and you have an opportunity to do PID. Now you might be able to do some level of PID in the plugin but I suspect the loop delay would be so high that it would be more like a random postion generqator or slow frequencuy oscillator than a true closed loop.

One statement in the programming manual for the UC300 is interesting: The motion transfer data is given such a high priority that all you need is a .1 sec buffer.....hum .1 sec buffer? So how fast do you think the inputs and any analog samples (remember you only get one reading of the register at a time) do you thnk we can get going back the other way?

Let us know how you work this out.

[vmax549]

HIYA TOM the idea is to NOT send anything back across the COm channel but handle everything INSIDE of the UC300 chipset in the controller itself. The only thing being transmitted across the COM will be the setvolts value at startup and any change of value of the SetVolts during a run. The UC300 is very fast but the newer generation chipset is even faster/larger yet and Ethernet driven .

SO the com really has little to do with it. The THC function would be running in realtime at the controller level. I believe that is the way you do it at the THC card level NO ?? The advantage it would have is it comunicates teh THCup/down internally so there are NO outside inputs to worry about .

The Z position updates would just be handled when it found a niche to update back to UCCNC or at the next M5 .
Early yet (;-) but VERY promising. Maybe it will and Maybe it won't.

I had talked to you about using YOUR THC card with teh UC300 under UCCNC but you did not seem interested in the project .

(;-) TP

[vmax549]

HI Greolt, The UCcnc does NOT do it the same way that Mach3 did it. It has a very unique way to make it all work and NOT skip steps. I am not sure I am at liberty to say HOW but I can say that it does work very well. (;-)

(;-) TP

[vmax549]

TOm just in case you are wondering the plugins for UCCNC can use any net2.0 language I use C# and it is very easy to work with. AND if I can do it and I can then to the rest of yall it would be a piece of cake.

The overal plan is to also include a modbus(ascii) link to teh hypertherm unit to do settings on the fly. Similar to how you do it.

Just a thought, (;-) TP

[tcaudle]

So if the plugin runs in UCCNC on the PChow do you access the internal FPGA code in the UC300 chip? I did not find any programming examples of how to directly access the firmware in the UC300 which is where a direct interface would have to be written. Maybe I missed something but I did not see where they exposed access to their internal firmware.

I have no interest in working with a USB motion controller OR anything that runs under Windows going forward. I have over 10 years of it with MACH and its been painful at times. Since my card (actually several) operates totally outside the control of Windows and even the plugins in a true servo control loop
its is only the physics of the mechanics that are the primary factor.

If you don't understand servo closed loop design and the things that cause instability , then it can be a painful journey.

[vmax549]

TOm I was NOT expecting you to have to do anything concerning the UCCNC project with your Card . I just needed to know the Com protocol you used to talk to your THC card I can do the rest. I can still do a sniffer on the comport to figure it out. It is NOT that hard just time consuming. Your THC assy would have made a great combination for plasma combined with UCcnc and their controllers . I was just looking for a good THC to pair up with UCcnc for teh DIY crowd that would not break the bank so to speak.

But with you selling off the low end side of THC it may be a mute point now. SO I have to look at other avenues

The Uccnc group does NOT seem to have trouble running windows ( all versions so far that I have tested ) and a controller at the same time and being that they build the hardware AND write the control software there have been ZERO surprises so far it all works exactly like they said it would. Every time.

(;-) TP

[vmax549]

Tom did I tell you that the UCcnc group also builds Servo Drives (CNCDrives) (;-) SO yes they understand servo loops and PID quite well,. ANd Yes you are correct it would use a new function in the firmware to run the THC function inside of the chipset.

(;-) TP

[Greolt]

HI Greolt, The UCcnc does NOT do it the same way that Mach3 did it. It has a very unique way to make it all work and NOT skip steps. I am not sure I am at liberty to say HOW but I can say that it does work very well.

Just out of interest I emailed CNCdrive about how Z axis moves are handled during THC control.

The UC100 and UC300 (their hardware) while operated with Mach3 does those moves the same as Mach3 via LPT. As in "bang bang"

The same hardware using the dedicated control software, UCCNC, does include an acceleration ramp on those movements.

How much that all means in practice, on my table, is yet to be seen. However I think in my case it just might be enough to make the difference.

Shame UCNC does not have a screen designer like Mach. Never understood how any user could put up with the standard Mach3 screensets.

[beefy]

Another update on that wonderful company Mesa, I'm furious with them:

http://www.forum.linuxcnc.org/forum/27- ... t=10#66481

[vmax549]

Hiya Greolt , UCCNC DOES have a NEW screen designer(GUI) AND a new style screenset framework(OpenGL) It has just not been released yet. (;-)

AND it does work very well . It is a bit different than OTHER screen designers(Mach3) but it gets the job done.

AND the old screens WILL run in the new framework. Also you have always been able to rework screensets BUT you had to do it by hand programming and THAT was not really that bad once you get used to it. It is like command line programming instead of a GUI.

The only thing missing for plasma is anti corner dive AND I think I can do that(ACD) from a simple plugin .

(;-) TP

[Greolt]

TP

Downloaded UCCNC and had a bit of a look at it.

Looks like it has most of the THC stuff. Should not be too hard for them to add a "disable THC when below x speedrate"

That combined with hysteresis programmed into the simple THCs should cover it............ Wouldn't it?

Screen designer would be most desirable. Could Klaus become interested? (just dreamin')

I'm thinking if it had Ethernet connection I would send off my money tomorrow. Been a bit burnt with SS USB constantly loosing connection.

Beefy

Sorry to have watched your disappointing experience with Mesa. Certainly makes me wary of pursuing that avenue.

Can't help wondering if it might have been different if you lived stateside.