Pokeys PlasmaSense Torch Height Controller Mach 3 Problem

[Konstantin]

Hi,

A few months ago I built a plasma table 2000x3000mm for personal use but I am struggling with getting the Torch Height Control to work as expected.

My setup is as follows:
1. CNC controller - Pokeys57CNC
2. Plasma source - HELVI PC EVO 126
3. Torch Height Controller - Pokeys PlasmaSense
4. I also have another Torch Height Controller laying around - Proma THC 150
5. Mach 3

Now I use SheetCam to generate the G-Code files.

Let me explain how the system is wired. The plasma source has an integrated voltage divider 1:50 that is wired to the voltage divider input of the PlasmaSense. The torch is fired with an integrated relay on the PoKeys57 CNC controller. The Arc OK signal, THC On in Mach 3, is wired to the outputs terminals of the plasma source. The reference height voltage in Mach 3 is a manual set value on the PoKeys plugin page. All these signals are terminated to a plug with a few pins on the back of the plasma source.

For sensing the metal sheet an ohmic sensor is used.

The reference voltage, the cut height, the pierce height, the cut velocity and the pierce delay time are taken from the Hypertherm manual since my local HELVI distributor asked the manufacturer and they told him to use the cut tables from the Hypertherm manual.

What I am doing is setting the pierce height at 6.4mm, the cut height at 3.2mm, the pierce delay at 0.5mm, the cut velocity at 2500mm/min, the amperage at 50A and the torch voltage reference at 129V to cut a 3mm carbon steel. I might be wrong about these settings but the Hypertherm cut tables start at 6mm carbon steel. Before enabling the THC a cut with THC disabled was executed and the reference voltage from the plasma source was around 129V. Then the G-Code is imported in Mach 3 and I press the THC button on the plasma screen in Mach 3.

There are a few scenarios what happens now, I will try to summarize.

1. Managed to get a cut with the Z Axis moving around +- 0.2mm. I assume the THC works.
2. The torch fires and stays at M03 line, after some time the arc is off and waits to toggle the torch on off button to resume the cut - only dry cut.
3. The plasma source reports an error when triggering the torch. According to the manual the input voltage of the plasma source dropped drastically. The mains voltage is 3 Phase x 400V. In my area we do not have the best voltage but I measured the voltage drop on each phase when running a 5.5kW screw compressor the voltage dropped from around 400V to 385V. Do you think that little voltage drop could cause problems?
4. The torch fires, tries to reach the desired amperage but gets stuck at around 19A, assume 50A is the setpoint.

My first thought was that I have to enable the THC when I am at the cut height, therefore I wrote macros M101 and M102 with the StartTHC() and EndTHC() function respectively but the same situations appears. Mach 3 even gets stuck at the macro line and waits until execution is done. I read on the forum that when cutting circles the THC should be off that is also another reason for writing those macros to control the THC on and off at G-Code but some people say that this a bad idea since this slows the motion control processor. My THC controller does not have an enable input and a delay for neglecting the THC.

Then I my mind decided that when the metal is pierced and the Arc OK signal is received the THC should start working. I pressed the THC enable button in Mach3 deleting the macros I talked about above but still the same not consistent results.

What would recommend to try in this situation? When to enable the THC and does it need to be disabled at all? Any settings that I miss in Mach3? What would be your approach?

Best Regards
Konstantin Kolev

[weldguy]

Oh man, you have a lot going on there! I think going back the the basics and confirming all your systems are functioning properly 1 at a time will go a long way.

Can your cnc table complete a dry run start to finish with no problems?

Figure out your plasma input voltage issue and then confirm you can cut simple straight 12-24 inch lines with it manually. Will it cut error free?

Next cut some 12-24 inch straight lines under cnc control and keep your THC off for now. Are you able to cut these with no issues and with decent cut quality?

Also your pierce height, cut height seem height seem high to me and not settings I would expect to see in a current Hypertherm cut chart. What Hypertherm manual are you referencing?

[Konstantin]

Oh man, you have a lot going on there! I think going back the the basics and confirming all your systems are functioning properly 1 at a time will go a long way.

Can your cnc table complete a dry run start to finish with no problems?

Figure out your plasma input voltage issue and then confirm you can cut simple straight 12-24 inch lines with it manually. Will it cut error free?

Next cut some 12-24 inch straight lines under cnc control and keep your THC off for now. Are you able to cut these with no issues and with decent cut quality?

Also your pierce height, cut height seem height seem high to me and not settings I would expect to see in a current Hypertherm cut chart. What Hypertherm manual are you referencing?

I managed to make some experiments as you suggested.

About the test cuts, I did some 500mm lines straight cut with the settings in my previous post - 5mm pierce height, 3.2mm cut height, 1s pierce delay, 50A, NO THC, 4000mm speed and the material is 3mm mild steel. Everything is fine, apparently my mains voltage somehow drops a lot when the torch fires and the compressor triggers. I asked a neighbor to use his power and he agreed. Now this problem is fixed.

The next is the cut charts I reference. What happened is that I got a file from my plasma source distributor that has only cut chart for 105A consumables. My thought was that since no other chars were given to me I have to reference only the 105A cut chart no matter what consumables I use. I got suspicious about that the cut charts should be specific to what consumables are used when you asked what manual I reference. Since I use 65A consumables there should be other cut chart to follow. The charts I found actually are on the forum. plasma cut tables
According to the 65A consumables chart the settings are as follow - 3.8mm pierce height, 1.5mm cut height, 0.2s pierce delay and 5000mm speed. When I use this cut height the torch dives into the material assuming NO THC. That is why I did the test cuts with the previous settings I mentioned above.

Then I enabled the THC in Mach3, it tries to do some adjustment but still dives into the material or PlasmaSens E-1 error is shown in Mach3 status bar. The voltage reference is set to 125 as stated in the cut chart. The manufacturer of the PlasmaSens does not describe what is this error about in the manual.

And my assertion that when at pierce height the voltage is higher than when at cut height still holds, at least for me. In my opinion I should somehow make the THC to be enabled when at cut height and there should be possible to enable disable the THC when cutting circles. As I described in my previous post I tried with the M101 and M102 but I guess it is too slow to be run with motion

[acourtjester]

It seems your timing for the THC is the problem of the torch diving. I have worked with a Proma 150 but not the other one you have, so that is where I will talk about. Maybe the other THC has similar settings that you can use.
First off the pierce delay and the THC on delay are 2 different things. The pierce delay is to allow the plasma to cut through the material. The THC delay is to allow the arc voltage to stabilize. There needs to be enough time to the pierce the material and start moving at cut distance and then the arc voltage to stabilize before the THC come in to control. You are able to make cuts without the THC fine, so it falls back to the delay of when the THC is activated. Attached image is of the setting on the Proma, the D--T is for the arc voltage stabilizing delay, your other THC may have a similar setting, I would use about 1.5 seconds. Get the table cutting error free before fine tuning and you'll be happy.

Proma delay.JPG

[weldguy]

The next is the cut charts I reference. What happened is that I got a file from my plasma source distributor that has only cut chart for 105A consumables. My thought was that since no other chars were given to me I have to reference only the 105A cut chart no matter what consumables I use. I got suspicious about that the cut charts should be specific to what consumables are used when you asked what manual I reference. Since I use 65A consumables there should be other cut chart to follow. The charts I found actually are on the forum. plasma cut tables
According to the 65A consumables chart the settings are as follow - 3.8mm pierce height, 1.5mm cut height, 0.2s pierce delay and 5000mm speed. When I use this cut height the torch dives into the material assuming NO THC. That is why I did the test cuts with the previous settings I mentioned above.

Ok, so reading above you are not at the point where you even want to begin thinking about using the THC. Forget about that, it will be a project in itself to get it functioning the way you want it is what I am guessing.

You are referencing Hypertherm cuts charts, good starting point but not set in stone settings so start with them and work with it cutting the straight lines. if you are using the 1.5mm cut height and it begins cutting at that height but then dives intot he material then your THC must be ON and cauing the torch to drive down. If the THC is not on then I suspect your telling the software you want a 1.5mm cut height but your not actually at 1.5mm cut height. You will want to test for accurate cut height values so the 1.5mm you set comes out at 1.5mm when you measure the distance between the torch and the material with a feeler gauge.

I would get that sorted out first so you can cut the sample lines at 1.5mm cut height sucessfully and then move on to the next step.

Tom has some great suggestions above with regards to your THC but again I would revisit that later once your doing really really well without it.

[Konstantin]

I appreciate your replies. Let me explain what further experiments were made today.

Perhaps the word dive was not the right one. I meant that if 1.5mm is used for cut height the torch hits the material as if it is not flat, the Z axis does not move in coordinates neither up nor down.

Then I tried some experiments with the THC. I decided to wire the Proma 150 instead of the PlasmaSens. The ARC OK input signal is wired directly to the plasma source output, therefore not using the Arc OK signal from the Proma 150. The THC settings are as follow: HYS - 2V, d-t - 2s, H-U - 200V. Since at 1.5mm cut height the torch can hit the material because of its flatness, I decided to pierce at 3.8mm and cut at 2.5mm. I did a 500mm straight line cut with no THC and observed the voltage on the Proma 150. It was I the range of 120V. Then the set voltage was 120V and the THC was enabled. But after the ohmic sensing procedure that is implemented in G Code it stays at the M3 line and waits. I can see that the plasma source gives the ARC OK signal but the motion does not proceed. It seems that it waits for ARC OK from the THC. After turning the torch off from Mach3 the motion resumes and I fire the torch again, during motion. Motion in Z Axis appears but at random points during cutting the gantry stops moving, maybe waiting for Arc OK but the arc seems fine as the plasma source is outputting the signal.
One thing is that the Torch On LED in Diagnostic screen is flickering. I tried to short the input pin for Arc OK but it still still flickers.

[acourtjester]

I is good you are watching the signals on the diagnostic page to what is happening. It seems you may have a wiring problem in the Arc OK circuit. I would use the Proma's output for this. When in the test mode you should see the up/down and torch on lights working in sync with the Proma's display. I don't remember which one but changing (maybe the HYS) value will make the light on time longer so it is easier to see them. Here is a video showing the way to check torch height setting and voltage checks when you get to fine tune things. The THC settings should be the same for many brands.
https://youtu.be/EyZJVMtzKzY

[Konstantin]

Let me make an update from what I experimented today with the plasma. I played with the Proma 150 THC and the result is as follows. The first few attempts to cut with THC on were with the following Proma 150 settings: HYS - 4V, d-t - 1.5s, H-U - 200V. But with these settings, as I mentioned in my previous post, the torch fires at 3.8mm and for some reason waits there at M3 line in G-Code, not going to cutting height, motion is halt. When turn off the torch manually with the Mach3 button the motion resumes and a dry cut is executed. This is with the Arc OK signal from the Proma 150. I am starting to think that perhaps the M3 macro is wrong - I have this function there: DoSpinCW(). Do I need to use ActivateOutput(OUTPUTx) there?

I read some similar topic on the forum about Proma 150 settings and Mach3 and reduced the H-U value to 150V, all other settings are the same. Now when the torch fires, the pierce delay expires, then the d-t time expires, the torch goes to cut height and motion starts. A few cuts were made with these settings without motion halt after pierce. One thing that bites me is that why the torch goes to cut height after the d-t expires? Should it not go to cut height after the pierce delay is expired? Perhaps a little bit elaboration the this H-U will be great.

The next try was to use the Arc OK signal from the plasma source, all other settings the same, but with that signal the motion does not start after pierce as in the first case. Then I recovered the Arc OK from the Proma 150.

Another action that rankles me is that when I have motion the the THC on, the Y axis motion is not smooth, I mean it seems it stops for a fraction of a seconds and then resumes. It is not that the motor stalls but more like the machines wait for something perhaps Arc OK signal. What would cause this rough motion?

Another thing to mention is that, in the PoKeys Plugin configuration, there is a Gain in the THC settings an I played with it from 0.1 to 1.0. This is the explanation for the gain in their manual.

Code: Select all

Gain – correcting motion gain (reduce the gain if the torch height is oscillating)

I will add that I did not get the Proma 150 to work in test mode, the device sends the signal to Mach3 but no movement in the Z Axis.

Well, there is some progress but I am still far away from what I expect.

[robertspark]

it is waiting for the arc ok signal before moving

the sequence is
touch off
zero the z work coordinates
retract to pierce height
fire the torch

now it has a choice depending upon your settings in the configuration.... if you have enabled "plasma mode" in either the configuration screen or possibly the mill tab (it's been a long time since I used mach3)

if you enabled the plasma mode it will wait (for ever) for the arc ok signal to be received, one it is received it will then start the pierce delay timer,
once the timer has elapsed,
it will drop to the cut height
and motion will begin.

if you have not enabled plasma mode then Mach 3 will not wait for arc ok
it will just start the pierce delay
drop to cut height
and motion will begin

check you have wired up and assigned the arc ok signal in mach3 and that it is getting the signal (check the ports and pins)

I hated the proma THC

it lacks settings and adjustment, yes it will cut steel for you but not well and not to a high quality finish, it will (or did for me) mean I spent ages getting it to work.

the transfer voltage setting in the proma is the but that you need to adjust.


if mach3 just flies off the handle and cuts an odd direction to ruin your material it is because you have a fake license key as artsoft put a bit of code that checks the licence key against known fake license keys and adds a bit of random code to ruin your work or material as it serves you right for not buying a proper licence.

I started with mach3, then tried mach4 then used uccnc which is great (not the best as it does have a few issues) and then needed something for my lathe and now run linuxcnc.... I will not go back to windows and motion controllers and being reliant on some but of code that I have no control of or how it works or not having enough I/o's or any other hardware issue or limitation..... but I appreciate that Linux is not for everybody and linuxcnc can appear daunting from a pure windows plug and play background

good luck

[acourtjester]

One of the problem with the Proma is sometimes it is wired wrong. the 3 signals going to the controller for up/down/arcOK. The comm connection on the Proma needs to go to the controller's comm connection not the Proma's power negative. Power needs to be completely separate from the signal connections.

Proma comm.JPG

[Konstantin]

Hi,

There has been quite some time with no update on the topic but I decided to upgrade the machine in terms of resolution and accuracy of closed loop stepper motors.

I did not mention it in the thread yet, since the other problems with the normal run of the plasma and the THC have higher priority in my opinion.

But when I tried to cut some circles ranging from 5mm to 30mm I got some really terrible results. The circles were not even close to smooth. It seemed that you can see the steps that the motors did while cutting. The arc of the smaller circles was awful.

I decided to read through the Internet about this problem and found out that my resolution - mm per step was really high. Since I had no reducer and the pitch diameter of the gear was 50mm, the length moved for 1 revolution was 157.07mm. According to the forum and other resources the optimal distance traveled per 1 revolution of the motor should in the range of 15 - 25mm.

Therefore I decided to order a 1:10 planetary reducers for all the axis except the Z. I got a table calculator for the reducer and the motor torque where I referenced my calculations.

Well, now the circles improved drastically, they are very close to perfect. Perhaps, there should be some more work done in SheetCAM for the lead in pattern.

I did my trials with no THC, this will be the next step.

This was a little introduction to what I upgraded on my plasma cutter.

One thing that is still missing is the downdraft table. But I am aware of what centrifugal fan should be chosen. After an Internet research, I got some reference values for the fan needed. And wanted to see your opinion about that or perhaps, another suggestion about the fan.

Basically the plasma table is 1500mm x 3000mm and I read, more than one place, that the fan should be centrifugal with backward inclined fins, 8000 m3/h, 2.2kW and the inlet duct should be in the range 350 - 500mm.

The downdraft design is mine and resembles a trapezoidal hopper for both halves of the plasma table - each half is 1500mm, no separation in the middle. There will be a drawer at the bottom of hopper - the short base of the trapezoid. On one side of this hopper, I plan to have to holes for the ducts that will go the a Y connection and then to the fan.

Do you think this fan is suitable for the machine scale or is it exaggerated?

Best Regards
Konstantin Kolev

[tcaudle]

The secret to downdraft is to spread the area of negative air pressure evenly but to limit air flow (CFM) . The objective is to actually cool the heated air and suspended particles so gravity csues them to drop vertically out of the air. Spreading out the intake ports across the whole surface but controlling the size of each port becomes the challenge;. One giant port takes a HUGE amount of negative air pressure to do the job .

The best design I ever saw was a shop that had experience in commercial air conditioning and tuning heat exchange systems by volume. They had two main pipes (about 305mm in diameter on each side) and cross pipes that were triangle shaped with holes about every 200 mm and about 50mm in diamter. it was designed so good that virtually NO debris or dust went into the pipes and it all fell down into the collection hopper in the bottom. It just not possible to get even or sufficient air flow across the entire grid unless you have a HUGE *an loud) vacuum fan.

Some downdraft is better than none but "tuned" air floe Downdraft is pretty amazing.

[Konstantin]

The secret to downdraft is to spread the area of negative air pressure evenly but to limit air flow (CFM) . The objective is to actually cool the heated air and suspended particles so gravity csues them to drop vertically out of the air. Spreading out the intake ports across the whole surface but controlling the size of each port becomes the challenge;. One giant port takes a HUGE amount of negative air pressure to do the job .

The best design I ever saw was a shop that had experience in commercial air conditioning and tuning heat exchange systems by volume. They had two main pipes (about 305mm in diameter on each side) and cross pipes that were triangle shaped with holes about every 200 mm and about 50mm in diamter. it was designed so good that virtually NO debris or dust went into the pipes and it all fell down into the collection hopper in the bottom. It just not possible to get even or sufficient air flow across the entire grid unless you have a HUGE *an loud) vacuum fan.

Some downdraft is better than none but "tuned" air floe Downdraft is pretty amazing.

Hi,

Thank you for your detailed answer.

As far as I understand separating the intake is not generally a good idea, rather have one larger intake port that goes let us say to a tube with evenly distanced smaller holes.

Would you elaborate a bit more on the design you described with the cross triangled pipes?

Best Regards
Konstantin Kolev

[Konstantin]

Hello,

Let me make an update on the topic.

I managed to get the plasma machine working. Now the machine covers all points mentioned in the posts above. It can cut straight lines with the CNC control. No more power source failures, the grid is stable now. I have cut a lot orders so far mainly in thickness ranging from 5 to 15 mm. All cuts that I performed were without THC.

I have been trying to get the THC working but no success. The first THC that I bought is PlasmaSend from Polabs. I can see Z Axis movement when sending voltage to the THC via power supply for testing but when I start cutting everything is really bad.

Then I bought Proma 150, the one with Arc OK, UP and DOWN signals. Actually I am still trying to get the Proma 150 running. I have the following setup:
1. Arc OK signals is received from the CNC interface of the plasma source.
2. HYS = +-8V
3. dt = 1s
4. H-U = 200V
The analog voltage from the plasma source is 1:50, therefore 0 to 6V.

Now, when in tst mode I can see the UP and DOWN signals in Mach 3. During test mode of Proma 150, I use it's own Arc OK signal. It can be seen in Mach 3 as well.
I have the following problems when cutting:
1. Not sure, still debugging, why the Arc OK signals flashes in Mach 3. In the Pokeys57CNC controller the input does not flash or perhaps, the flashes cannot be seen. Now in the start of program I get materials initial height via ohmic sensor. Then go to 5mm in Z, fire the torch and then machine waits with Arc OK flashing very fast. I manually turn M3 off, the machine resumes but no Arc available. It runs in dry mode.
2. After turning M3 off, the machines starts lowering to cut height - 3.2mm. Then, as fast as possible I manually turn M3 on. The THC takes place. But the movement of the machine is quite jerky, as if the machines stops, waits for Arc OK(not sure), and resumes. This happens very quick and as you can imagine the cut line is awful. The THC is correcting the height ,though.

I remember that you did not like the pierce and cut heights I use, but at those height I get good results. The plasma source is HELVI PCEVO 105.

Actually, I tried a lot of settings but cannot get the THC to work as intended. Would anyone provide any hints of how to tackle this phenomenon?

Best Regards
Konstantin Kolev

[weldguy]

It sounds to me like the jerkiness of the motion could be caused by hysteresis with your THC. Can you try a wider seting for this? I am not familiar with your THC but I see you have setting HYS = +-8v so I expect if you were make that number much bigger you may see the jerkiness slow down or go away altogether. Curious what you find out.

[Konstantin]

I will try what you suggested during the weekend and give feedback.

HYS=+-8V is in the middle of the controllable range of Proma 150. My desire is get it down to +-2V in the future.

What about the problem with Arc OK signal? I receive Arc OK from the plasma source and Mach 3 does not allow motion. The after some seconds the arc is lost due to no material contact. This flashing of Arc OK LED in Mach 3 concerns me. I have set up an exterbal 24V LED to indicate the same Arc OK signal and it does not flash, the signal is pretty stable.

And I am not quite sure if the H-U=200V is correct. As far as I understand the manual, this setting is to detect Arc OK signal from the Proma 150. In my opinion, it is better to use the Arc OK signal from the plasma source.

[tcaudle]

Noit sure how this is all interfaced BUT the Promo150 simple UP and DOWN THC needs to be used whit Parallel port MACH3 . With USB or Ethernet there is a delay (like an echo on a phone line) that will cause overshoot and jitter.
The Promo150 is not he fastest to start with then add the loop delay from the interface to MACH though a non-paralel port and you get a LONG loop that is poor response and unstable.

The Promo150 is basically a Z jog design that depends on the jog commands being near real time and not delay though buffers.,

[Konstantin]

The Pokeys57 controller is wired via Ethernet to the computer.

There is an LPT interface on the Pokeys57, do you think this will be equivalent if I wire the signals through that LPT interface?

The communication between Pokeys57 and Mach 3 will still be via Ethernet even if I wire the Proma 150 signals to the LPT interface.

[acourtjester]

I have no experiences with the PoKey controller but from other controller I have worked with the DB-25 interface is an output not an input. Meaning it is to connect to another BOB type card, not for a THC signal input. the image is from their manual.

DB 25 connecter.JPG

[Konstantin]

I apologize, got confused with all that mess about the THC. Off course, the DB-25 connector does not have inputs, this option drops out.

tcaudle mentioned about the delay that causes jitter when USB or Ethernet is used. I can switch to LPT to try but my moving velocity will probably drop a lot. Perhaps I did not mentioned what is my mechanics. The steppers on the Y axis are NEMA 34 2x12Nmm the X axis stepper is NEMA 34 9Nm, the Z axis is NEMA23 low torque. There is a rack of module 2 and a gear with 25 teeth, the pitch diameter is 50mm. The circumference of the gear at the pitch diameter is 157.078mm. The gear is attached to a planetary reducer with a ratio of 10. Therefore 10 revolutions of the motor will result in linear move of 15.7078mm. I set the mircrostepping of the stepper drivers to 51200. The maximum pulse frequency of the Pokeys57CNC is 100kHz. The steps per mm is calculated as 3259.527. Having all the information above with current setup I can barely run the machine at 2m/min, this is at around 108kHz. Would the LPT be capable of the 100kHz that is written in the Kernel settings?

I am also trying to be able to increase the maximum velocity to at least 4m/min since when cutting think material it is better to run faster.

Perhaps the reducers are overkill having that 12Nm of each side of the gantry but steppers are known to significantly decrease the torque in higher RPMs.

These days I contacted the manufacturer of the PlasmaSens THC. They ask for wiring pictures etc. The PoKeys57 controller is designed to work with the PlasmaSens and I still wonder why it is that difficult to get a descent THC results.

[Konstantin]

Let me make an update on that topic, perhaps it will be useful for someone.

Now I am trying to use the PlasmaSens THC which is part of the PoKeys57 CNC controller. I decided to give it one more chance since it is designed to work with the CNC controller and it is manufactures by the same company.

Attached you can find some pictures of the test I performed. The results are far from good. The cut sample is a square 40mm from a 3mm sheet. I follow the Hypertherm cutcharts for pierce height, pierce delay, torch-to-workpiece distance and THC voltage. I cut at 2000mm/min instead of 5200 mm/min. As I explained in my previous post, this is the limit speed of the machine due to motors microstepping and reducers. The hysteresis voltage is +- 4V, this is the Deadband setting in the picture. THC speed in Mach 3 is set to minimum - 5%

As you can see from the cut sample picture, the result is awful. There are areas where the metal is not pierced.

Would you share your opinion about the cut results and the settings used in the experiment?

Cut Sample 40x40x3mm.jpg

Plasma Source Settings.jpg

PoKeys Plugin Settings.jpg

Mach 3 Settings.jpg

Cut Chart.jpg

[tcaudle]

Not sure what some of those settings mean or what values they are in, but cutting mm per min and reducing federates by 50% throws you voltage settings all off. The charts are designed to be used with ALL of the values and if one is off by that much the results will be off. One of the problems you have is your table simply cannot perform at the level it takes to cut thinner material so you are going to have to change some things.
Make your cuts with he THC turned off and watch the voltage as it cuts and THAT will become your new target (Preset) voltage. Consider that every torch volt is equal to .005 of height. (.127mm) If your optimum cut height is 1.5mm .then + or - 4 volts is + or - .5 mm.
All that being said: What has been ignored is the DELAY in the THC loop . Imagine driving a car at 120K/h remotoly wit a 5 sec delay in the video.... In your THC control, Because the signals have to be converted from parallel to serial Ethernet/USB then back again and feed through buffers, I don't see how the MACH THC logic can be used . In order to get real time THC you have to move the voltage sensing and up/ down control outsde MACH into its own loop and just let MACH give you status and ON/OFF .
So If your THC vendor does not understand the delay issue, it time to switch vendors . You also need to look at your mechanical design and realize that with plasma cutting there are specific parameters you need to be able to hit within the limitations of steppers . Buying components from various vendors (franken-build) and making them all work together can be a major engineering project and support is nonresistant.

[weldguy]

Again, forget about the THC. Everyone wants to throw that in the mix and it will send you on a wild goose chase most times. You need to get your machine cutting those squares successfully with no THC and as Tom mentioned while you are doing that, look at your arc voltage to determine your new setpoint voltage but until it is cutting perfect with no THC, don't even think about using it. Find your amps, find your speed, perfect the cut, then introduce THC.