IT sure would be nice to hear their comments on both methods (;-).

(;-) TP

Sorry for this being a very long post but wondering what the experienced plasma guys thoughts are, thats you TP, Ian, Rob, Dave and any others

 I have been messing around with the servo drive to see what my options would be.
So here are a few things you can do in this drive.
You can have 2 modes of operation, so for example you could have :-

1. The main mode as  Step/Dir then the override mode as Preset Velocity (RPM) .
2. The main mode as Preset Velocity then the override mode as Preset Position (encoder counts)

You can also set up an input to start the motor homing and you can have various methods of defining home. One option is a switch, another is motor current value etc etc.
Also for the Homing you can set a back off amount if desired, that would mean when the home sensor is seen it will back off a desired amount and then set the encoder count in the drive zero.

Another nice feature is you can define an input to have multiple meanings, for example I have set one up as Operation mode override and also Preset Position 0, another I have set up as Operation mode override and preset position 2.


Now that I have mentioned some of the options I have in the drive here are a couple of ways I could work this axis.

1. I could use Step/Dir as one mode and Preset Velocity as the second. I could then command the drive as the Z axis from Mach then when I want THC to take over I just use the override input and either up or down and these would get operated directly from the THC.
2. I could use Preset Velocity for one mode and Preset Position for the other and that would mean I do not use the Z Axis at all in Mach.
3. Just have Mach control the whole lot and have the THC control the axis via Machs UP/Dn commands

So in example 1. above what would happen is I command a home move, the axis would move to the surface (under the drives control) reach the current value then tell Mach it has homed and Mach  would set the Z Axis as zero. Mach would then back off to pierce height and XY move to position and the torch would start, after the defined time then Mach would lower to the cut height and then pass control over to the THC and the THC would control the servo directly via the Preset Velocity inputs. Once that part of the cut was complete the THC would hand back to Mach, Mach would move to safe Z, move XY then tell the drive to home and the cycle would repeat.


In example 2 above the Z axis would not be controlled from Mach at all. A signal to home, probably from a M Code, would be sent to the drive, drive would move the motor down until it touched the material and the motors current reached the set amount. It would then back off the preset amount, call this safe Z, then Mach would be told it can continue.
Next X Y would move and then  the Mode Override/Preset Position Input would send the axis to the pierce height, once pierce time is elapsed the next Input would be used to move to cut height and then control of the drive would be handed to the THC  by using the Preset Velocity Inputs for Up/Down. Once cut is complete the axis would be sent back to the safe Z height and then the operation would repeat for the next cut.


I can see both 1 and 2 working and the second method would be more configurable with respect to accell and decel, as they can be set individually in the drive, but I can see that the pierce height and cut height would be fixed values in the drive. That could be overcome by having more than 1 input for  pierce heights and thus I could have 2 or 3 different pierce heights. I think the cut height is normally a fixed height (1.5mm or so) after which the THC will take over and vary accordingly.

What do people think?
Any pitfalls that can be seen in any of the options?
Is this worth pursuing or am I better just letting Mach handle the whole thing?
 The advantage I see with the THC controlling the drive direct is it cuts out the middle man (Machs Up/Dn), so it should be more responsive, but would it make much difference?

At the moment I am thinking Option 1 would be the best but.....

Hood

I would hook it up as a normal Mach3 thc setup , USE it for a while . I would bet you will not have any problems cutting what you cut.  I do it about every day with a slow stepper setup (;-).

(;-) TP

Och man, that is just boring

Ok I can probably do that but I think I will use the AB drive rather than the Samsung I originally intended, just in case

Also I reckon the drive homing would be a good way to touch off, so I could use that part. It would allow me to have the Z Axis as a rigid light set-up and not have to have a floating torch.

I home the Beaver Mill with the servo drives except I have it configured for Switch and Index pulse. I also home the lathes turret via the AB drive, on that I use the motor current method, they both work great and I can see it working equally well on the plasma.

The only concern I do have with driving the UP/DN via Mach is I will be using the CSMIO, there may be more of a delay there than would otherwise be if using the PP.
The inbuilt THC in the CSMIO would probably be faster but I suppose time will tell.

If I can wire/configure the panel up so  that I can easily swap to different options that may be the best way.

The only problem I see with using the AB drive for Z is the motor is small, only 50W but if I put it through the 4:1 gearbox and keep the Z light I think it should be enough.

Hood

Hood, thanks for the too highly rated complement, I'm really just starting out on a very long journey.

I know zero about servo drives, that being said, I like your options, they look very interesting and may have advantages of driving the THC directly.

However... If you drive the THC directly, the z axis DRO will have zero idea where the torch actually is until the next touch off (g28.1).... I don't think this would be a problem with my setup, as I touch off before my torch refires every time.

Another problem you may have if you control the servo directly via the THC.... You won't be able to turn the THC off when blended tradjectory is below x% of the feedrate....

With regards to the rest of the control, I'll try to explain my setup (right, wrong,  works for me).  I use sheetcam, and that goes care of my touch off (I use g28.1, but you can use g31), and also via snipetts, I turn my THC on and off for small circles and corners etc.

Really, consider my setup a little further, I use an ESS, and turn my THC on and off by making and breaking the THCOK signal back from my THC, given THCUp and THCDn are ignored when THCOK is not present.

The THCOK signal is broken by feeding the THCOK feed via a relay which is controlled via m10px and m11px as these are the only mcodes that don't cause a divot and tradjectory to slow for the mcode execution.

The THC is a proma (actually its Dave's old one...) (I am awaiting a new product on the market, .... Since last sept..!!, hence I nearly built my own and have a few times nearly bought the  neuron... If they dropped the pricing back to their introductory price I'd be there like a shot!). I have wasted a great deal of time exploring everyone THC that I could find on the market to try to understand what each offered.

Hence its crude (only settings are hysterysis, torch transfer voltage and target voltage), and its slow... It uses good old fashioned relays... Optoisolators would have been way faster....I would not recommend them, although for the cost it has met its needs as a very entry level thc. (Read:cheap!)

Anyway, the THC is set with a very close hysterysis (2v, the minimum), and an arc transfer voltage of only 5 volts above my target.... The reason behind this is because its acting as an anti-dive function.... Hence the THCOK signal will drop out (or not be enabled until the torch voltage is within 5v of the target voltage, and any spikes the THCOK signal will drop out.   From all the info I've found on the net, 1torch volt is about 0.32 to 0.64mm of torch height... Say 0.5mm per volt.... Hence 2v will be around 1mm of cut height (+/- ~ 0.3 mm).

I leave the THC to control the torch height.

I leave sheetcam to insert my code snippets (which I may hand adjust onscreen) to turn my THC on and off via m10px and m11px, via the post processor.

I leave sheetcam to inset my pierce pause by setting up the relevant tool settings  (pierce delay, pierce height, initial cut height and do my touch off before every pierce)

Code snippets in sheetcam reduce my feedrate for circular cuts to 60% of the straight line feedrate automatically also for circles less than 1.5" dia (most of my metal is under 4mm)

Hope some of that is for use to someone

Rob

OH boy Hood are you in for an edumacation. With plasma you HOME z to find teh top of the material not to establish a real Z home. There are 3 main sensing methods ohmic tip, Remote offset switch, and floating Switch on Z. AND 2 possiblities of motion with each, G28.1 and G31

Now that is with the standard Mach3 methods. THere are actually many more but that is another story.

  TP

Oh... I use floating switch...

Have a look at this table, you can use bits of the idea too.

http://m.instructables.com/id/CNC-Plasma-Table/

Rob

Whatever you use, i would not use motor current sensing for the IHS method, unless you cut 12mm plate all day it will buckle the sheet and give bad readings for z-zero. Since tweaking my system up with an offset switch probe and increased z motor tuning, the cuts have certainly improved and I am fairly confident it can follow pretty much any warp I will see.

I still have the floating z setup, connected to my limits chain, i would not build without  this as its a good safety feature in case of a dive, a mag-break torch holder would probably be a good replacement though, i will be wiring my z limit to the e-stop circuit soon as the limits are ignored when doing a g28.1 and if my probe switch fails it will plant the torch heavily into the sheet, there is enough torque in my stepper/screw drive that it will not stall the motor and just keeps on going! The failure point is that it rips the delrin nut out of the z-axis body which means stripping it down to replace it.

I admire your work on this though, I love R&D work, used to de plenty of it once in the day-job, only used servo's once though and that was step/direction to the servo drive and then closed-loop with encoder/resolver on the servo motor, this was a feed system for a stamping machine.

Proma did have an upgrade version of their THC and it took over direct control of the Z motor, switching back to Mach when stopped, I nearly tried it but splurged on the MP3000 instead. I managed to get the loop response times from Proma guy and it was certainly slow, ok for cutting thick stuff but I found it failed on thin sheet due to the speed needed.

Hood, thanks for the too highly rated complement, I'm really just starting out on a very long journey.

You know way more than me on the subject, so that is reason enough to ask your advice

I know zero about servo drives, that being said, I like your options, they look very interesting and may have advantages of driving the THC directly.

However... If you drive the THC directly, the z axis DRO will have zero idea where the torch actually is until the next touch off (g28.1).... I don't think this would be a problem with my setup, as I touch off before my torch refires every time.

Don't think that would be an issue as I too would be touching off each new cut

Another problem you may have if you control the servo directly via the THC.... You won't be able to turn the THC off when blended tradjectory is below x% of the feedrate....

Not too sure about this, the THC has an input that locks it out, so I could likely use that. Not sure however in what circumstances it would really be needed, there should be very little slowdown at corners, then again maybe it doesn't take much. With servos I will have high acceleration (hopefully ) and the code will hopefully be such that even a sharp corner the cut will roll round the corner at the radius of the kerf thus never actually having to come to a stop. Maybe I will end up with a slowdown but if that is the case then I can probably integrate the Lock function.
The THC itself has Anti-Dive in it but how well it works I don't know.

With regards to the rest of the control, I'll try to explain my setup (right, wrong,  works for me).  I use sheetcam, and that goes care of my touch off (I use g28.1, but you can use g31), and also via snipetts, I turn my THC on and off for small circles and corners etc.

Really, consider my setup a little further, I use an ESS, and turn my THC on and off by making and breaking the THCOK signal back from my THC, given THCUp and THCDn are ignored when THCOK is not present.

The THCOK signal is broken by feeding the THCOK feed via a relay which is controlled via m10px and m11px as these are the only mcodes that don't cause a divot and tradjectory to slow for the mcode execution.

The THC is a proma (actually its Dave's old one...) (I am awaiting a new product on the market, .... Since last sept..!!, hence I nearly built my own and have a few times nearly bought the  neuron... If they dropped the pricing back to their introductory price I'd be there like a shot!). I have wasted a great deal of time exploring everyone THC that I could find on the market to try to understand what each offered.

Hence its crude (only settings are hysterysis, torch transfer voltage and target voltage), and its slow... It uses good old fashioned relays... Optoisolators would have been way faster....I would not recommend them, although for the cost it has met its needs as a very entry level thc. (Read:cheap!)

Anyway, the THC is set with a very close hysterysis (2v, the minimum), and an arc transfer voltage of only 5 volts above my target.... The reason behind this is because its acting as an anti-dive function.... Hence the THCOK signal will drop out (or not be enabled until the torch voltage is within 5v of the target voltage, and any spikes the THCOK signal will drop out.   From all the info I've found on the net, 1torch volt is about 0.32 to 0.64mm of torch height... Say 0.5mm per volt.... Hence 2v will be around 1mm of cut height (+/- ~ 0.3 mm).

I leave the THC to control the torch height.

I leave sheetcam to insert my code snippets (which I may hand adjust onscreen) to turn my THC on and off via m10px and m11px, via the post processor.

I leave sheetcam to inset my pierce pause by setting up the relevant tool settings  (pierce delay, pierce height, initial cut height and do my touch off before every pierce)

Code snippets in sheetcam reduce my feedrate for circular cuts to 60% of the straight line feedrate automatically also for circles less than 1.5" dia (most of my metal is under 4mm)

Hope some of that is for use to someone

Rob

Thanks for the insight, time will tell how my THC works, looks decent on paper but what it will be like in real life is another matter.
I will likely just use BobCAD for the code as I already have it, how good it is for Plasma I have no idea but the PP are quite easy to alter to suit so I should manage to customise it if I find it is not suitable from the start.

Hood

Yup the Proma is a bad choice... but it was "cost effective" and a starting point on a long learning road ("if I knew then what I know now...")

Basically anything that relies on relays is a bad idea for THC for a start!

the Proma "SD" being direct drive [pass through until Torch Voltage Sensed, then Proma Controlled until the end of the cut) is just an improvement by way of some motion controllers not having THC programmed into them (Chinese USB ones or the USB-SS for instance [started there!, read line 1 above]).... but still uses relays it "looks" like.... (if you read the manual closely).

The SD does not offer the THC Inhibit if blended trajectory drops below "x%" of the feedrate...

And it also does not offer M-code inhibit for corners to turn the THC on & off...
http://www.planet-cnc.com/faq/THC/THCSD.png

If it detects torch voltage.... it controls !




Ohmic has always interested me.... but if you have a working floating head.... is it a gimic or does it really improve the cut?   
All depends I guess on what you are cutting (thickness, surface rust / coating etc) and how you are cutting it (water table).

Could also be considered a "double tap".

I did not know G28.1 ignores limits....   what about if you tie your home and limit inputs together using the same pin numbers?

And put the floating head (or ohmic) on the "probe" input (which is my arrangement)?

Rob