Mega V Plasma/Router Custom Build w/Open Frame Bed

[bLouChip]

It took nearly 3 months to complete, but with help from a few friends in the shop, and several tips and recommendations from major suppliers, I have completed a custom build of Mega V Plasma/Router mounted on an Open Frame Bed.  This configuration makes the machine portable, primarily for the purpose of landing on a surface, the subject cut material surface, to either route or plasma cut. This makes the boundaries of a job limitless in terms of XY dimensions.  That scenario is not the only way to use this CNC machine, it certainly works as a static mount on a table as one would expect, you put the cut material into the work bed and clamp it, it works fine, in fact that is the scenario I expect to use it in 80% of the time. Secondarily, the Open Frame Bed configuration, being detached from a table or bed, allows for multiple storage options when the machine is not in use; I chose to store the unit on the ceiling of my shop!  The plasma water table then becomes my welding table with the simple addition of .250" steel plate, laid down in overlapping grid of 4 panels of .125" x 44" x 21" dimensions, each weighing ~30 lbs. 

For graphic design and gcode generation on plasma jobs, I'm using Lightburn software (https://lightburnsoftware.com/)built for CNC laser use. I post process the gcode using a VIM editor (https://www.vim.org/) macro to search and replace the M3 and M5 commands with my plasma torch touch off probe and retract commands, then I run/send the gcode file using UGS.  The end to end process is very smooth and quick.  I've made a feature request to Lightburn to include the plasma use case as it has other UI/CNC advantages that would eliminate the use of UGS if one so desired, same as in their CNC laser use case.  Here is a link to the request and further details, files, and the VIM macro on Lightburn's forum (https://forum.lightburnsoftware.com/t/update-on-plasma-use-case-success-albeit-post-processing-gcode/37258).  With LB, I was even able to resume a job cut after having a torch probe malfunction, which otherwise would have resulted in scrap steel.

After completing the build, I ran into several issues while debugging the system for the first successful plasma cut. Primarily the issues were EMI/RFI interference with the Arduino USB communication line, the symptom being lost comms with my computer as soon as the plasma arc began.  The only cutting that took place after that depended on the amount of gcode that was in the Arduino stream buffer, which amounted to about 16 characters or so. The solution was to apply largely just basic EMI/RFI shielding and ferrite chokes to communication USB cables, D25 signal cable, trigger and sense cables to the plasma cutter, to relocate the trigger cable and relay from the MR controller box to a panel 4 ft away, and to earth ground the plasma table and major CNC system components.  I have a dedicated ground stake 5 ft into the ground just 10 feet away from the table, grounding the table and components with 2/0 1000 strand copper welding lead cable.

The plasma cutter and torch I'm using is a PrimeWeld Cut60 (https://primeweld.com/products/cut-60-220v-110v-60-amp-plasma-cutter) unit with a Tecmo PTM60 coaxial shielded machine torch. Excellent machine for the price, and excellent customer service.

I have created a youtube channel, bLouChip Consulting (https://www.youtube.com/channel/UCdUAmFDK3Tw_ihA8WcPkb8g) and I'm posting videos there describing the custom build.  I also have several posts in a related thread in this forum here (https://millrightcnc.proboards.com/thread/3323/plasma-mega-set)   I'm happy to share information if anyone else is interested in a similar build or any subsystems.

I'd like to thank MR for great support throughout this build and for designing such a robust CNC machine that makes this possible.

Cheers,

Lou

[bLouChip]

I'm continuing to update my youtube channel with videos and instruction re. the custom build, and have joined the facebook group MillRight CNC Mega V Group to post short vids and to offer assistance.

www.youtube.com/channel/UCdUAmFDK3Tw_ihA8WcPkb8g

[bLouChip]

update:  I recently studied the problem where, during plasma cuts using a THC, and with 20 or more piercing cuts that require a torch touch off with the cut material,  the THC induced Z machine error can cause a soft limit alarm either when a torch touch off is about to occur (grbl probe command) or when a post cut/M5 retract (Z+ movement) is about to occur.   The THC takes control of Z axis motor without knowledge or monitoring of grbl (CNC control box). Even though the THC is only moving a mm or less on each cut, after 20 cuts if those movements are all in the same direction (worst case) then eventually such soft limit alarms/errors can occur.  Also consider if your THC set voltage is just 1 or 2 volts off of normal, then this worst case condition is likely to occur.  More on why this occurs, how to fix it, and how to precisely set the THC monitor voltage, see this post and thread in the LightBurn software forum: forum.lightburnsoftware.com/t/update-on-plasma-use-case-success-albeit-post-processing-gcode/37258/6

If you don't have LightBurn software, the post is still informative, and here is the THC Z error diagram it references:

[Bruce]

Thanks for the updates on using the plasma cutter on the Mega V. This is great information.

[bLouChip]

I've had a couple of questions sent to me recently re. the Mega V Plasma CNC machine setup in general, so I'm posting a short update here.
The key modifications and additions I made in this regard are as follows, although some of these are also necessary and/or beneficial to the Custom Open Bed Frame design.

1) Mount the control box to the table, or somewhere such that its in a fixed position relative to the table and moves with/if the table moves. In my case, the control box is mounted to the Custom Open Bed Frame because I want it to move with/when the CNC and frame moves on and off the table. As for grounding the control box, decide if you want to keep it isolated/insulated and separately grounded from the table OR if you're going to intentionally ground it to the table. There are engineering arguments either way. Net point is that you'll want to eliminate or minimize electrical ground loops, they contribute to electrical noise and otherwise unreliable signal communication and control design, and of course computer interference. However, there are so many potential cross ground connections in a CNC machine, one or more ground loops are inevitable if trying to isolate the control box.

2) Grounding general. The CNC machine is a giant antenna, in fact multiple antenna, receiving EMI/RFI under any conditions, and then couple that it's also a giant transmitter when plasma cutting. Thus I found using a common star ground buss design with dedicated earth ground has worked out best. That is also what Hypertherm, the gold standard in plasma machines, recommends. Here are a couple of links to Hypertherm's recommendation for CNC EMI/RFI mitigation, its an interesting read.
1. updated link:  www.hypertherm.com/en-US/archives/learn/articles/how-to-nullify-noise/externalContentView/0e1f4b28-34dc-4e44-8958-5e459d25d228?preview_generated=True
2. www.hypertherm.com/Download?fileId=HYP103900&zip=False
The only thing in the recommendations that I don't like nor do in the manner they suggest is hard wire (bolt) my plasma cutter + work lead to the CNC table ground buss. Rather I use 2 clamps on my work material; 1 clamp is the usual plasma cutter work lead and 1 clamp is a short jumper to the table earth ground buss. I concur that grounding the plasma work lead is a good thing to mitigate EMI/RFI. In my method, I don't start cutting a job unless I see 2 clamps on the work material, it's that simple; but if one or both clamp is missing, no harm no foul, it just doesn't cut or the job is potentially RFI interrupted. If I were to set this up per Hypertherm's recommendation, THEN WHEN I FORGET to clamp my work material, then very likely I would fry the small gauge grounding wires from the star ground buss to other components of the machine, computer, or control box. That won't be a good day. I suspect Hypertherm's recommendation in this area is more of a dedicated setup where the plasma machine is dedicated to the CNC table; not the case in my shop where I use the plasma cutter manually with a hand torch just as often as CNC use.

3) EMI/RFI Shielding. I found that MR did not thoroughly ground their shielded signal wire used on homing switches; one end of the cable should be grounded but it's not, and control box paint insulates the male connector but I was able to scratch the paint to get a connector ground, although I don't believe the shielding drain wire is grounded to the connector. Luckily, I have not yet had to rewire the cables and ground them; live'n on the edge I suspect I don't notice the motor control cables grounded either, I don't even recall if they are shielded. I'm prepared to rewire these and properly shield them if necessary, but so far so good. Managing the excess signal and control wires is just as important. Coiling them near the control box is clumsy, messy, and not conducive to EMI/RFI mitigation. Rather, I discovered that the excess wires loop nicely into and out of one end of an 18 inch long 2x2 inch metal tube secured and grounded right behind (or next to) the control box. This makes for a shielded and clean solution in this excess cable problem.

4) Plasma trigger and THC voltage sense wires. Mount the THC and plasma trigger relay at least 4 feet away from the CNC control box, they are just too noisy for Arduino and USB communication line. Use the DB25 cable to send the signals to the remote mounting location. I grounded the even pins of the DB25 buss and cable, used only the odd pins for the signals. The plasma cutter is another 10 feet or so away from the table, more EMI/RFI mitigation.

One more thing...  perhaps the best modification yet for CNC Plasma control...  I add a simple SPST toggle switch in series with the trigger relay, and I named it the "Plasma Arming Switch".  The purpose of this manual control switch is to simply "disarm" the firing of the plasma arc very easily and very positively from the CNC control panel area, I don't have to walk over to the plasma cutter to turn off the main power.  This is useful to easily conduct dry runs and to safely disarm the plasma during other times while working on the torch or setup. I can also use it to delay the refire of the plasma torch when Resuming a Paused job since there is a 4 second delay in XY movements after the M3 command is issued during Resume command.

5) More grounding. I found that Arduino is powered by the connected computer through the +5VDC of USB communication cable. At the control box, neither the USB jack nor Arduino is chassis grounded to the control box. So with a laptop computer, the +5VDC USB is floating relative to ground. Yet another EMI/RFI receiving antenna. Grounding the negative voltage buss of Arduino solves both problems; I grounded to the control box, adding a wire to ground the 48VDC power supply and 5VDC to a common box ground screw inside the box.

6) RF Chokes. I added RF chokes to several signal wires, this was perhaps the single most important EMI/RFI mitigation step of all: 1 each end of the 6 ft USB cable connecting my computer to CNC control box; 1 on the USB pigtail inside control box to Arduino board; 1 on the DB25 cable nearest the control box; 1 on the signal and power wire bundle at the THC top buss; 1 on plasma trigger and THC voltage sense wires nearest THC and trigger relay board.

7) Plasma machine torch magnetic mount. I used the stock MR sliding/probing magnetic mount with a few simple modifications. I added a small steel plate (.250 thick) to the Z plate by tapping the 2 router mount holes in the Z plate and drilling/tapping 2 more making a 4 bolt pattern. This added plate extends the Z plate down by 3 inches and relocates the router mount and plasma magnetic mount area down by 1.5 inches. On the steel plate I welded a horizontal .375 square steel rod as a shelf and a .188 square vertical rod as a left side corner. The 2 welded rods make a positive, repeatable, and easy location corner, horizontal and vertical, for the magnetic torch holder, and a horizontal shelf/locator for the router mount. I remove the router mount when using the torch holder, I don't like passing the torch through the router mount, it can break the torch if a collision were to occur. See my youtube video showing and describing this setup. www.youtube.com/watch?v=QAnQKW7SjpQ

8) CNC / GRBL Pause/Resume and Magnetic Torch Mount Breakaway Interrupt Switch. See the video in previous item 7.
A side effect of this circuit, specifically the Pause/Resume buttons, is that I can easily pause the plasma cutting job at any time, but most importantly after about 10 pierces and cuts. Pausing allows me to check the accuracy of the THC voltage setting which controls the torch cutting height. An incorrect setting will cause an eventual Z movement soft limit alarm, either during retract after each cut or during the torch touch off probe prior to each cut. The soft limit alarm occurs because Z accumulates positioning error due to too much THC movement in the same direction. GRBL doesn't know that THC is moving Z, and Z will never be exactly in the same position after THC control as it was before it took control. Hopefully, the Z error is neutral over time, but I have found that as little as 1 volt of incorrect setting can accumulate enough Z error to cause the alarm after 25 or so cuts/pierces. So in order to check the accuracy after 10 or so cuts, I compare the latest probe depth (reported in the console log) to the first probe depth, the difference is the accumulated Z error. A positive error means the THC voltage is set too low, a negative error means the voltage is too high. So simply adjust the THC voltage accordingly, maybe even a little more than ideal so the Z error corrects in the opposite direction, and resume the job. Repeat this pause and check every 10 cuts or so until happy that the THC voltage is tracking well.

9) Plasma machine torch. Get a quality torch, preferably one that uses coaxial shielding. Tecmo makes one, PTM60. I got lucky and acquired one without asking for it, and by comparison to the China made off brand torch, its night and day different in quality, mechanical movement of the electrode, longer consumable wear factor, quality of cuts, and most importantly, with coaxial shielding for much improved EMI/RFI noise reduction. The mechanical movement (retract) of the electrode is key to being able to run low PSI (sub 55 PSI range) when cutting thin gauge metal and using 20-30amp tips.

Cheers,

Lou

[bLouChip]

CNC plasma cutting process can be improved if the torch is adjusted to a precise height for both piercing the work piece to begin the cut  (Pierce Height PH) and then lowering it to an Initial Torch Height (ITH) as the XY axis begin to move and thus making the cut.  If a Torch Height Control (THC) is also used, then the cut gets even better since the THC can maintain the ITH dimension as the cut progresses.  All of this adjustment per cut is desirable since the work piece will warp due to heat during the cut, AND the best cutting takes place in a narrow zone of +/- just a couple of tenths of a mm.  But with or without a THC, the PH and ITH can be precisely set using the MR MVP standard magnetic torch mount and the Probe input circuit to the GRBL controller. The limit switch on the torch mount is wired to the Probe input circuit.  As the GRBL probe command executes (G38.2 Z-25 F800 ;for example), the torch tip (torch off of course) will gently collide with the work piece and the torch mount gravity fed slide will engage, allowing the Z axis to continue downward until the probe limit switch is made. At that point, the movement stops, the next instruction retracts Z by the known amount of the gravity fed slide travel (I call this the Torch Probe Travel TPT). The torch tip is now effectively "zeroed" precisely on top of the work piece, and can now be moved again to the PH and then eventually the ITH.  The speed of the probe command will influence the accuracy, its just like homing, so use a relatively slow speed like F800, and then do it again at F100, and you'll get great accuracy.  This whole process is called the Torch Probe Touchoff per Cut.

Here is the Torch Probe Touchoff gcode that I use.  I search and replace the M3 and M5 commands in laser process gcode generated by Lightburn software. I use the VIM Editor and a custom macro to perform the search and replace.  The macro text is posted down below. The whole process takes place in milliseconds in the VIM Editor, then save the gcode file, and run it !

================================

;begin probe for Pierce Height (PH) and Initial Torch Height (ITH)
; Torch Probe Travel TPT=9.2mm
G21 G91               (metric and relative)
G38.2 Z-24 F800  (probe fast from safe retract, 10mm+2xITH+TPT rnd)
G0 Z8                   (fast retract off probe switch)
G38.2 Z-10 F200  (slow probe -2mm+retract)
G0 Z9.2                (fast retract to TPT)
G0 Z3.2                (retract to Pierce Height PH, 150%-225% of ITH)
M3                        (light it up)
G4 P1.0                (pierce delay, make THC delay => this+2.0)
G1 Z-1.5 F800      (transition from PH to ITH)
G1 f2350              (restore feed rate, move mode, and units as needed)
;end probe ITH

;various G1 XY movements to perform the plasma cut...

;begin safe retract and ready for probing next cut ITH
M5                        (put it out)
G4 P1.0                (wait for THC to give Z control back to GRBL)
G21 G91
G0 Z10                 (fast retract to safe height)
G1 f2350              (restore feed rate, move mode, and units as needed)
;end safe retract and ready ITH

;repeat for each plasma cut in the job...

=================================

Here is the text of the VIM macro to perform the search and replace of M3 and M5 for the code above...

===================

" macro template to replace M3, M4, M5 with plasma torch probe touchoff code
" use vim find and replace command...
" TPT + PH - ToITH = ITH
" TPT + 3.2 - 1.7 = 1.5
" TPT + 3.4 - 1.9 = 1.5
" TPT + 3.6 - 2.1 = 1.5
" TPT + 3.8 - 2.3 = 1.5

%s/\(M3\|M4\)/\r;begin probe for Pierce Height (PH) and Initial Torch Height (ITH)\r; Torch Probe Travel TPT=9.2mm\rG21 G91 (metric and relative)\rG38.2 Z-24 F800 (probe fast from safe retract, 10mm+2xITH+TPT rnd)\rG0 Z8 (fast retract off probe switch)\rG38.2 Z-10 F200 (slow probe -2mm+retract)\rG0 Z9.2 (fast retract to TPT)\rG0 Z3.2 (retract to Pierce Height PH, 150%-225% of ITH)\rM3 (light it up)\rG4 P1.0 (pierce delay, make THC delay => this+2.0)\rG1 Z-1.5 F800 (transition from PH to ITH)\rG1 LDCrestore (restore feed rate, move mode, and units as needed)\r;end probe ITH\r/gc

%s/M5/\r;begin safe retract and ready for probing next cut ITH\rM5 (put it out)\rG4 P1.0 (wait for THC to give Z control back to GRBL)\rG21 G91\rG0 Z10 (fast retract to safe height)\rG1 LDCrestore (restore feed rate, move mode, and units as needed)\r;end safe retract and ready ITH\r/gc

promptrepl LDCrestore

===================

Cheers,

Lou

[bLouChip]

update: I rewrote the VIM script I’m using to post process LightBurn laser generated gcode into plasma process gcode. I cleaned it up quite a bit, made it much less cryptic, more man readable, and taylorable, I hope.

You can download it from the Lightburn forum here since we still can't post files to this forum. 

forum.lightburnsoftware.com/t/update-on-plasma-use-case-success-albeit-post-processing-gcode/37258/10

Also, while testing the new script and gcode, I made a plasma specific interesting discovery too; I noticed I was loosing 0.4mm of the 1.6mm Initial Torch Height (ITH), thus I had an effective 1.2mm ITH unknowingly, due to the speed (200mm/min) that I was probing the torch touchoff. Basically this is overrun of the probe micro switch, that is Z moving another .4mm by the time the CPU could handle the switch close interrupt and stop Z steps. I tried different slower speeds and found that 100mm/min was about as accurate as anything slower, so 100mm/min during probing it is. Good to know.

Enjoy.

Cheers,

Lou

[bLouChip]

another MegaV Plasma process update in general and also on previous posts:

re. torch touch-off and retract gcode prior to each plasma cut:  here is example of the gcode...

==================================

;begin torch touchoff probe: isRampZ=1, ITH=1.5mm, PD=1.1s, PH=3.3mm
G21 G91 (metric and relative)
G0 Z-20.000 (Z down, undo previous safe retract)
G38.2 Z-12.100 F1000 (probe fast to find limit switch)
G0 Z5 (pull off limit switch, ready for slow probe)
G4 P0.400 (delay for probe switch debounce)
G38.2 Z-5 F50 (probe slow for accurate make on switch) <== double probe to find switch and correct height, where the 2nd probe is at 50mm/min to reduce overrun.
G0 Z8.600 (retract TPT, to top of material)
G92 Z0 (reset temp Z0 in case of absolute mode gcode file gen)
G0 Z3.300 (retract to PH)
M3 (light it up)
G4 P0.660 (partial pierce delay; set THC DLY=>1.8) <== this is only 60% of the pierce delay (PD), 40% is used during the ramp Z move in next stmt.
;end probe ITH

G1X77.969Y-0.015S356 G93 F136.364 Z-1.800 (ramp Z, remainder of PD) <== the XY moves are from the design file and represents the lead-in in this cut.
G94 (restore distance units move mode)
F2400 (restore feedrate)

...  XY moves to perform shape cut...

;begin safe retract and ready for probing next cut ITH
G21 G91 (metric and relative)
M5 (put it out)
G4 P1.000 (wait for THC to give Z control back to GRBL)
G0 Z20.000 (fast retract to safe height)
;end safe retract and ready ITH

==================================

re. VIM script to modify the LightBurn laser generated gcode, making it plasma process gcode...

This is the script which produces the gcode shown above, for each M3 and M5 in the LB laser generated gcode file. The script was further improved for ease of use and modification if you desire.

Get the latest script in this LightBurn forum thread: forum.lightburnsoftware.com/t/update-on-plasma-use-case-success-albeit-post-processing-gcode/37258/13

re. plasma cutter unit...

The PrimeWeld Cut60 unit is much improved for CNC process, they have redesigned internal component placement and wire routing and harnesses as well as made some improvements to circuits.  PrimeWeld shipped me a new unit to demo and it was significantly better (reduced effect) in EMI/RFI emissions. Since I had already mitigated most EMI/RFI noise through grounding of the CNC table and components, use of RF chokes on critical comms and signal wires, and distance separation of plasma control and sense wires, the most significant improvement I noticed with the new plasma cutter was in the THC management of the Z height during the cut and the displayed arc voltage. There was much less fluctuation of arc voltage and apparently much less noise on the arc sense wires and as a result, the THC did a much better job when moving Z, it did not overshoot the movements and there were significantly less Z corrections. This all translated into a very smooth and quality cut.  I even ran tests to begin with a warped sheet of steel to force THC to adjust Z as it cut, and it did great!  Not one torch collision with the steel or dross.

All in all, I am very pleased with the reliability, repeatability, and quality of the MegaV Plasma CNC right now. 

The price/performance ratio is fabulous considering all components: MR CNC system, PW plasma cutter, and LB vector graphics design/CAM software.  I have also experimented with LB and SheetCam combo for plasma process, it too is very good but also requires some plasma specific programming for torch touch-off.

Cheers,

Lou

[bLouChip]

The latest update to this CNC system is a NEJE A40640 15W laser. It's quite the improvement over the 2.8w laser on my smaller CK machine.

See some details and photos here: forum.lightburnsoftware.com/t/new-addition-neje-a40640-w-air-assit-working-well/61268

I'm renaming my system the MegaV XL 4-Axis Custom Tri-CAM w/ Open Frame Bed 

It's time that I make some updated videos of the system, I'd like to share the 'how to' in case someone would like to use this design as a base to build another, make improvements, that's how we progress this hobby. I hope to have the videos posted a few weeks.

The Z plate overlay/extension is the key to a quick change of CAM process heads; meaning I can switch between router, plasma, and laser in as little as the time it takes to stow the cutter away and retrieve the next one in the case of plasma-to-laser and visa versa, since both process heads use magnetic mounting. In the case of routing, it takes 5 mins to attach or detach the router mount to the Z plate overlay/extension.  Well, it also takes 10 mins or so to fill the water pan in the case of plasma process.

I use the stock MR MVP controller. I keep the software use simple also, using only Fusion360, LightBurn, and UGS; broken down into use cases of - CAD software: Fusion360 for mechanical design and some sheet metal; LightBurn for artsy/crafty graphics and text design for all 3 processes. For CAM software: Fusion360 for routing/milling; LightBurn for laser and plasma. For CNC control:  latest version of UGS for router and plasma, LightBurn for laser control.

Cheers,

Lou

[kevin]

Those are some amazing mods to your Mega V Router/Plasma hybrid.

[bLouChip]

machine and plasma process updates recently: 

1) Alas (sigh),  I have shelved MyPlasmaPP() VIM script for converting Lightburn laser gcode into plasma process gcode.   

It was fun while it lasted, and certainly a low cost entry into this once unknown realm of CNC plasma cutting for me.

Tip:  don't f' around with other means, buy SheetCAM, its great.

That said, I still had to modify the SC post processor code to tune the process and gcode to my liking and for the MegaV w stock controller box (w grbl 1.1i MR version).  I'll share it with anyone that wants it. Compare it to the stock version by the same base name "GRBL plasma.scpost" to see what's modified.

Of course you still have to design/draw the parts/job using a vector graphic or CAD sw, and for that aspect I still love and use Lightburn for the task, and I still use it exclusively for CNC laser process, and for graphics and text in routing process. So with LB, I "export" my design to an .svg file, and "import" the .svg into SC, and then its a few simple SC specific UI maneuvers, visually verify and simulate the toolpath, and I'm ready to cut ! Its a very fluid LB to SC to UGS work flow and easy to iterate as I may make changes to either phase of the end to end process.  LB is great for vector graphic design for any purpose, and it's toolpath gen and review for laser process is suitable, but not efficiently suitable for plasma process given the number times a leadin needs to be positioned and the importance of cutting in the correct direction and on the waste side of a shape line, not to mention several other key aspects of the toolpath gen and review that SC facilitates well. 

I'll be checking into SC for routing/milling also, given the struggles I tend to have with con-Fusion360 if I don't use it frequently

2) Proma THC SD:  this is a weird device. So it has a few side effects; and unfortunately MegaV stock configuration of Z motor control is a triggering factor for one of them.  In a second case, PrimeWeld Cut60 plasma cutter is the triggering factor.  In yet a third case its choice of arc voltage monitoring, that's simple, wire it to your plasma cutter to use Raw Arc Voltage not 50:1 divided.   So here's the deal...

In the case of the Z motor config:   The THC has setup config that includes naturally the Z motor direction it must dictate when attempting to adjust the torch height during the plasma cut, that's it main job of course. So the THC DIR setting is binary: 1 or -1. The stock MegaV config requires this THC DIR value to be -1.  It turns out that the THC has a little quirk when this DIR value is -1 in that - read this carefully and perhaps several times -  [the first time the THC attempts to adjust Z, AFTER a power on event, it will drive Z in the wrong direction!; depending on your THC monitor voltage value this will cause the torch to either crash into the metal being cut or to move so high off the metal that the arc will eventually fail to cut.] Digest that for awhile. It drove me nuts for months trying to figure this out whereby I discover the quirk!

So I wrote Proma recently to inquire if I was actually experiencing what I thought I was seeing, and they confirmed it, eventually, after initially replying "What??" "is this right, is that right, blah blah blah" Then, "oh yeah, that's a problem, someone else may have reported that behavior too; here's what you do..."

So the fix is:  change the MegaV Z motor configuration and wiring so that is runs in reverse of the factory config.  Its simple.

step 1) power off the MR control box, but leave the USB cable plugged into the box as this will keep power to the micro-controller board and grbl program.

step 2) on Z motor driver (black box gadget that the Z motor wires connect to), swap the B+ and B- terminal wires; trace these wires to the Z motor 4 pin connector on the back of the box to be sure you have the correct cable and thus driver gadget. This swapping of wires will reverse the motor direction. DON'T drive Z motor again until you complete step 3!

step 3) in UGS, on the command line, change the grbl configuration value for motor direction ($3 step direction invert mask). The command to enter is: $3=6

Now enter command: $$

the grbl configuration will display on the console window and you can verify that now $3=6

$3=6 means Z and Y axes direction are inverted.  Y is factory inverted already, now you have made Z invert also.  No change necessary for X. read about it here: github.com/gnea/grbl/wiki/Grbl-v1.1-Configuration#2--step-port-invert-mask

By changing this grbl config value, you have now reversed the Z motor again, thus making it rotate in the mechanically correct direction again! Very important!

So to recap - step 2 reverses the motor via electrical wiring, step 3 reverses the motor via grbl configuration control. Reverse + Reverse = Normal

step 4) power on the MR control box and home the machine, it should work normally now. If not, review the steps carefully above.

step 5) power on the THC. After the display shows "------", press the up+down arrow keys simultaneously until you see DIR on the display, now make that value 1.

That's it.  THC now works first time every time after a power on event.

In the (second) case of PrimeWeld Cut60 and perhaps some other plasma cutters:  The THC monitor voltage calibration should be adjusted.  I switched a couple of months ago to monitoring plasma Raw Arc Voltage, not 50:1 divided arc voltage. I should have known better a year ago but I was concerned about the high voltage; basic rule of thumb is to not use high voltage if you don't have to.  However, in this case raw arc voltage has a much greater signal:noise ratio than does any divided arc voltage, since the noise amplitude is constant as it is radiated onto various wires such as the arc voltage monitor cable and wiring internal to the plasma cutter. Electrical noise = EMI/RFI; not good for electronic circuits, micro-controllers, and circuits and computers in general. So the greater the amplitude of the actual circuit signal (arc voltage in this case) is when compared to electrical noise, then the better the receiving circuit will be able to filter the noise and operate correctly. Its not complicated. Its like needing to speak loudly in a room full of noisy people.

Now, it just so happens that PWCut60 has raw arc voltage pins on the CNC Port connector on the back of the unit. If you use these pins, then as the user manual says, there are 100K - 150k ohm resistors in series behind each pin, so as to protect the otherwise high voltage (w available high amperage) pins from melting if they are shorted to anything. You don't have to test this, but if you do short them together or to ground lets say, it won't matter because the 100K ohm resistors will limit the current to only 1 milliamp or so, in theory anyway, but don't test it.  So, back to the THC... the raw arc voltage input on the THC has an impedance (resistance) of 384K ohms, so the THC is only seeing roughly 1/2 of the raw arc voltage or 50VDC or so. The greater the voltage that the THC "sees" then the more accurate it can respond to minute voltage changes. So, let's "recalibrate" the THC to "see" a higher voltage; this is done by a process more intense than I can explain, so watch this video:  www.youtube.com/watch?v=hOb4YRvx-0s

The CAL value that I use is 180, which is the max. I was a little leary of setting it to the max, so using a value of 170 is fine too, either is much better than the default of 100.  These values are a ratio where 100 = 1:1 of actual voltage:read voltage.  So a value of 180 = 1:1.8, making 50VDC look like 90V to the THC, which puts its voltage compare circuit in the sweet spot of optimum response and adjustment on the Z motor control. 

Cheers,

Lou

[bLouChip]

Hey, 2 updates in one week !  I've had some spare time lately so I thought I'd catch up on implementing some CNC machine improvements such as the previous post, but also today's exciting successful implementation and test results.  One of the most nagging issues I've had with CNC plasma process is the Proma THC SD propensity to drive the torch down (Z move) on every XY cornering move (aka abrupt change of direction). In theory, the THC should then also raise the torch once its through the corner and back up to programmed feedrate, but it doesn't always have the "runway" to do that if another corner, and another corner, and so on just keeps occurring. Most cuts are closed shapes, so the torch is doing a full 360 deg of corners and arcs and then some if the shape has left and right corners; cutting the outline of text is a great example. So there becomes an accumulation of down moves to which upward moves just don't cancel, and thus after each shape cut, there is usually a 0.2 to 0.8 (tenths) of mm of "Z error" that the THC leaves when returning Z control to the CNC controller. If  the per cut error = 0.5mm on average, then 40 shape cuts could accumulate 20mm of Z error, then add 20mm as the usual post cut Z retract before moving to the next shape cut and eventually you're going to get a Z limit alarm on retract or on the subsequent Z torch probe in prep of the next cut. Very annoying. You mitigate this somewhat by paying attention to the THC up/down LEDs, or the Z motor coupling rotation, and/or the THC voltage display vs. its track to go to the set voltage; and even adjust on the fly the set voltage to attempt to compensate, but all this is getting way too attentive, mentally taxing, and just not fun. BTW - the THC, being isolated from the CNC controller so as one does not know the other is in the circuit, the THC performs this Z down move in corners because grbl has to slow down XY speed to navigate the corner, while the plasma jet continues to consume metal as if it were going full speed; so with less metal to consume, even for a short time of 50 to 100ms, the plasma arc voltage spikes and the THC does it's job, it drives Z down assuming the arc is vertically too long, when in fact its too long in 360 degrees !   So, how do we keep the THC from driving down in corners and essentially over reacting when its not necessary to do so ? 

Here's a TTL circuit that will do the job  

The solution premise is that we can use the grbl Spindle Pulse Width Modulation (PWM) signal to disable THC Z control/movement when the CNC machine is otherwise conducting a plasma cut and XY is in the process of abrupt cornering, abrupt enough that grbl has to decelerate XY to navigate the corner and then accelerate back to programmed feedrate. When in Laser Mode ($32=1) and using M4 command as Spindle Enable control, then grbl manages S-PWM duty cycle proportional to the feedrate; where 100% feedrate (F) : 100% Spindle Speed (S), or any other 1:1 proportional linear value thereof where realtime feedrate dictates the percent of total. I was going to use S-PWM to simply AND gate the Z STEP signal from the THC, however it was even easier in that the MR motor drivers have an ENA-able pin, so I just feed the digital circuit to that pin and all is fine. Grbl command M7 (mist signal) is used to gate enablement of S-PWM, as is a toggle switch I named "Plasma Mode"; so those 2 conditions have to be "ON" in order for S-PWM to have any disabling effecting on Z motion.

I bread-boarded the circuit today, ran many tests, and it worked great!   I cut 5 plasma jobs with many acute angle corners and short < 10mm straight line segments and the Z error was the least its ever been, ever !  It never exceeded an accumulation of 2mm total in the jobs !   The cut quality was improved somewhat also since I did not have to over compensate THC target voltage to reverse accumulated Z error, and thus the torch height was consistently 1.5mm gap and arc.

Cheers,

Lou

[bLouChip]

Updates:

1) New improved THC Z STEP Suppressor Circuit...   The original circuit in the previous post relied on the Spindle PWM signal's duty cycle to suppress THC Z STEPs, so that meant that some THC Z STEPs "leaked" through to the Z motor, basically when the PWM was High, aka the duty cycle period. This new circuit latches the motor disabled when the the PWM duty cycle is 67% or lower, and does not re-enable the motor until the duty cycle is 88% or greater. So this circuit absolutely disables the Z motor when the XY slows down between this range while plasma cutting. The new circuit also eliminates the "chatter" I'd hear in the motor which was the rapid on/off switching going on at the 1Khz frequency of PWM.  I eliminated the "Plasma Mode" toggle switch which was AND'd with M7 signal to enable the circuit; I just decided I'd dedicate M7 signal to this purpose, thus no need for a toggle switch AND'd with it.

The customized SheetCAM GRBL plasma postprocessor (.scpost file) is in the photo gallery shared folder.

For the SheetCAM job file, a graphic of the parts layout, and example of generated gcode from the post script, see the "shape test.*" files

  in the shared folder- drive.google.com/drive/folders/1Kqt8RInPSwdsv4s-5aUB3uAAHuvQLFic

MR folks should consider modifying the GRBL code to handle this function of THC Z STEP Suppressor; they've already modified it to include the 4th axis (rotary axis) support so they maintain their own thread; they could use M7 AND'd with a toggle switch (Plasma Mode) if they'd like to preserve the M7 Mist function; wiring M7 & Plasma Mode sw to the Z driver +ENA terminal would be an excellent solution; the GRBL modification would involve raising M7 signal IF M7 was issued AND the PWM signal was <=70% or some other value they may choose to use; they might even make it a GRBL config value ($nn=-FeedRate% to Disable THC)

update 10/1/22:   NOTE:  see an update/correction to the THC Z STEP Suppressor Circuit in the post of 10/1/22

and updated link to the circuit diagram- drive.google.com/file/d/1UKgOIl-9XRypoTO2jaWg7nsvEHAYx1RG/view?usp=sharing

update 3/4/23 and 4/2/23:  Decel and Accel timer Threshold resistor changes as noted in updated circuit drawing. Tip- add a 10K pot in lieu of fixed resistors and you can fine tune duty cycles to your liking.

These resistor changes were necessary due to the PWM frequency being 1.9kHz instead of 1kHz on the Mega V Atmega2560 controller.

See this post for explanation- millrightcnc.proboards.com/post/25739/thread

2) I thought I had dry compressed air for my plasma cutting until a couple of weeks ago I cut a large job over several days where each unit in the job was: 32"x24" overall size, 10ga HR mild steel, 65 pierces (not much), but 22 meters of cut length!; it took 21 minutes to cut one unit and there were 12 units in the total job. So my band new father's day gift Kobalt 4.5 SCFM Ultra Quiet 70db Air Compressor was working overtime to keep up, but it did just fine as I was using a 40A (0.9mm) tip. However, the runtime of the compressor being so high, and the summer air here in NC being so humid, it basically turn into a water pump after 30 minutes of runtime.  The 26 gallon tank was warm, and the pint sized desiccant dryer was warm also, it was likely expelling water at that point not absorbing it. So needless to say I arc welded a couple of torch tips while I diagnosed the issue due to wet air. The fix was to split my unit job into 10 meter segments and change the desiccant beads at that interval, but after cutting 3 units even that scheme failed. So I needed to get a better compressed air dryer; so I had been considering adding a condenser/after-cooler to the compressor anyway and decided to just build it. It works great!  I have dry air now, cost only $200 for the 16 pass tube and fin coil (truck tranny cooler), 12VDC condenser fan, 12VDC 10A power adapter, AN fittings, copper tubing and fittings, and a ball valve for the drain tube. So for $600 ($400 for the Kobalt Ultra Quiet Air Compressor and $200 for the condenser system, I already had the $70 HF desiccant dryer) I have a very dry air system, vs. $2,200 for California Air Tools similar capacity system w $300/yr maintenance. The desiccant beads that I use are $30 a gallon, it takes me a year to use a gallon and they are recyclable by drying them in the kitchen oven for 1 hour at 200 degF.

Here's a link to the condenser and a customer review that I wrote explaining the installation- www.amazon.com/review/R2DJXMDNZGGE9T

Here's a link to some pics- drive.google.com/file/d/1Kh8V0KRslsMgf6AOc18Z7sc4UODk4Aj1/view?usp=sharing

drive.google.com/file/d/1KgFvmv8cJsUfxdR_cL1QWDOGAQdAkChC/view?usp=sharing

I hope you find this info useful.

Cheers,

Lou

[bLouChip]

update-  New video posted on bLouChip Consulting youtube channel - www.youtube.com/channel/UCdUAmFDK3Tw_ihA8WcPkb8g

Router to Plasma Process Conversion on the MVP bC Custom Tri-CAM CNC w Open Frame Bed.   Parts 1 and 2.

I've also started sharing CNC build and how-to photos in this google drive folder: drive.google.com/drive/folders/1Kqt8RInPSwdsv4s-5aUB3uAAHuvQLFic

Enjoy.  Cheers,

Lou

[bLouChip]

update:  I posted yet another video on my youtube channel the other day, that's the 3rd in a week, holy cow!  This one is titled Plasma Cut Narrated, something I've been wanting to do for awhile which was to explain the plasma cut process preparation and setup and then go right into a cut.  The cut results (parts) were the best ever that I've had, only +/- 0.003" from the design drawings, using a Tecmo style torch PTM60, 60A shielded tip (1.1mm), on 12ga HR steel, 1.3mm kerf, 3.2mm PH, 1.5mm ITH, 1.0 sec PD, 52Amps, 70 PSI, 1800mm/m, 6mm arc leadin, PrimeWeld Cut60, SheetCAM toolpath.  Even small 0.100" design holes cut at 0.110" actual, no leadin, crazy ! 

So what changed over the past year as it relates to plasma process ?  Many things as noted in the "blog" written in this forum thread.  But most significantly in the past 2 months I'd have to say are 1) dry compressed air.  2) adding the THC Z STEP Suppressor circuit, 3) monitoring raw arc voltage with the THC, rather than 50:1 divided voltage, 4) fixing the THC first cut bug. The fun and efficiency factor went way up when I began using SheetCAM, the workflow process is quick, easy, smooth, and largely automated and error free.

So my observations of many plasma cuts comparing before and after the "dry air" fix is that dry air drives many factors in terms of cut quality, accuracy, consistency of operation, less wear on consumables, edge squareness, on and on.  You don't have to be blowing up torch tips to realize you have wet air, I only did that when I ran a 20 minute job due to the compressor storage tank becoming warm from continuous running and it being impossible to dry the air down stream at that point. But I suspect the air was wet enough even with much shorter running jobs and I got inconsistent quality of cut and overall inconsistent results but didn't know it was caused by moist air at the time, since my desiccant dryer was catching moisture and I was maintaining silica bead changes on a regular basis. Still, there was moisture getting through and although it didn't blow up my torch tip, it was leaving deposits on my electrode and that should have been enough of a clue for me.  I don't have those deposits now with the dry air system. Also, over time, prior to the dry air system (the compressor condenser), I was getting an accumulation of water in my air line hose as 90% of it sat on the reel, those loops around the reel would collect condensed moisture in the air, which would likely result in a "burp" of water droplets every now and then. So bottom line,  if you are CNC plasma cutting, save yourself the grief and make sure to use dry air, which involves some sort of after compressor condenser before the storage tank.

Changing topics-  Recently I purchased Vectric Aspire for the purpose of creating and cutting 3D models.  I'm just getting started with it after studying many tutorials. I hope to make my first 3D wood carve within the next 10 days or so, I have a bday gift deadline approaching for "the boss"

Cheers,

Lou