You may remember several months back that i was atlking about a $100 DTHC system which is controlled entirely by Mach, you may have also seen some discussion about it on the machsupport forum, well things have changed a little since then, like moving from England to California.

Anyway i decided not to market the product but instead to try to help out the people like myself who cant afford a commercial system but want something that works ok for a hobbiest by writing down how i made my system and publishing it on here for all to read, well ive eventually gt round to starting and the first draft is attached, its not perfect but it will give you an idea and ill be able to see if anyone is interested by the response. There is one other guy who has recently completed a similar system and so we plan to go ahead and use this as our area to chat about the system, we have been doing this offline but decided it would be more constructive if we talked on script so others could read about it, if you have any questions or constructive comments please feel free to join in.

I would like for this to become a popular subject so that we can maybe get some of the functionality the system uses included in Mach in the same way that spindle drive feedback and other things are and then possibly a manufacturer including the other hardware in their breakoutboards to have a cheap easy plug together system for those who want it.

I am also working toward a DTHC system but with a Galil board for Camsoft.
Mach can also use the Galil board. I'm still welding on the mechanical table, only thinking on the THC at this point.

From what I've read, in additon to voltage dividing; filtering is a good idea to reduce noise. have you looked at capacitors? Also an opto isolation device would protect your equipment. I've been looking for help on the detailed design of this concept.


Interesting project, I'll watch this thread closely.

Karl

Hi Karl,

This is an ideal first post, someone else on my wavelength, i tried a few capacitors, i forget the values off hand but they were the ones recomended for the schematic in the mach documents which tom c donated, they didnt really seem to make any difference, so i tried a really large value just to see if that did anything and it smoothed the signal to such an extent the thc didnt work, so i took it back off and never replaced it, i havent found a way of testing the system accurately enough to be able to tune this sort of thing to perfection, any ideas on how to test it?

As for opto isolation i agree it would be nice to isolated the signal but i didnt think that was possible on an anlogue signal is it? Maybe someone who is cleverer than me could clarify that?

Thanks

Matt

I'd sure like to see that schematic. I looked for it over at the artsoft site and didn't find it.

As to capacitors, you know what's too low and what's too high. The best way is trial and error. if you can collect digital voltage data at high speed, calculate the short term standard deviation. The optimum is where adding capacitance doesn't make a significant reduction in variablity. You're almost certain to have situations where your system is unstable if you don't take the time to do this up front.

A EE type told me I should add an opto isolator like its no big deal. I see digikey has 100s of them but I'm not knowledgeable enough to make a selection and design the circuit.

This is great, now im learning, i presume by a means of measuring voltage data at high speed i would need some sort of sensor and then a data logging system? I have used similar with strain guages which i guess is just voltage so that would work wouldnt it?

On the opto isolation the only type of opto isolators i have used from places like digikey are essentially an optical transistor, i used a bunch to isolate the inpu / outputs from my pokeys device, all they do is use a different ground to avoid noise i think? So they wouldn work with an analogue signal, next time you see htat guy ask him how you do that id be interested to know.

As for the schematic il have a look a minute...

here you go, section 12 shows a schematic for an analogue thc similar to the thc 300

http://www.machsupport.com/docs/Mach2Mill_6.11.pdf

Our MP1000C (circa 2007-2008) was the first model with total analog isolation. The new DTHC module is isolated. Doing analog isolation and keeping it linear and affordable is not a simple.

Mixing the workclamp (which is not ground) and the logic ground (PC ground) can have interesting effects. Hook up with a HF-HV start unit (a lot of older and larger units can jump an arc over an inch) and your voltage divider will be pumping enough voltage into the upstream circuit to fry everything. The output if most plasma units have noise and the newer ones are PWM. The loop gain characteristics of a THC using feedback to the control software (MACH) are important to understand. Make it happen too slow (too much dampening) and you have a low frequency oscillator and the torch bounces down the table in a vertical wave pattern. It will be interesting to watch as you attempt to use the slower serial input to close the control loop. We were never successful with that approach but maybe we just didn't stay with it long enough.

Not having some primary test equipment (DSO, Isolation transformers, etc) will be a challenge and a learning experience. Be careful and don't make it a painful one.

Yes, I did the initial design of what is now the THC300. It has gone through several minor but important design changes by Jim Cullins (Sound Logic). It was me that worked with Art to put THC logic in the core of MACH way back when. I helped Les add plasma features to SheetCAM. All of it was selfish at the time because I wanted it for my own plasma cutting business.

Your call for integration of a DTHC to the important breakout functions has already been heeded. Our single port solution for all the I/O (UBOB) has an expansion slot where a low cost DTHC module just plugs in and becomes part of the PC to table interface. It can be upgraded in the field from a simple 4 axis interface to a fully isolated Plasma control unit. Using our board level products, You can build a complete powered, buffered,isolated BOB + DTHC with all the cards and isolation for less than $500 bucks. That includes the plug-ins, screens and profiles for MACH3.

We have a unit running with EMC as well.

TOM Caudle
www.CandCNC.com
www.CandCNC.com/UBOB.htm

Great introduction and expanation of the basic design of an arc voltage THC (or DTHC) system for controlling steady state torch to workpiece distance with plasma cutting! Hopefully this will get some experimenters tinkering and improving the methods of controlling torch/work distance....and absolute necessity in order to get the best consumable life and cut quality from any mechanized plasma.

I wanted to add a few comments from my 31 years of experience at Hypertherm......we have been designing and supplying stand alone and integrated torch height controls for high end industrial plasma systems for over 35 years.

- Research shows that controlling the height accurately during steady state plasma cutting is critical to cut edge angularity and dross formation. Older, analog type THC devices controlled torch to work distance within a + or - 10 volt deadband range...which effectively maintained height to within about .020" of the desired height. Improvements over the years in plasma torch and power supply technology dictate that the torch height should be held to much tighter tolerances...most of todays THC systems control torch height to within + or - 1 volt ....holding torch to work distances within about .004" of the desired height.

- Equally important to consumable life and cut quality is the ability to find the surface of the plate acurately before every cut cycle.....and to retract the torch to the (manufacturers recomended) proper pierce height for the particular material being cut before the pierce is initiated. It is critical for a full function THC system to have an accurate IHS (Initial Height Sensing) function in order to get the best performance in terms of cut quality, and especially consumable life. Many methods have been used to find the plate over the years.....micro-switches, inductive and capacitive proximity sensors, fluidic pressure sensing, ohmic contact, z axis stall force sensing....are the most common, and each has its advantages and disadvantages under certain usage conditions. IHS is necessary for cnc plasma systems.

-There are quite a few tuning factors to consider as well....while I'm not totally familiar with the Mach capabilities....I'm quite sure many of these tuning functions are available: kerf crossing anti dive, collision sensing, torch collision protection, pierce delay, tip up avoidance, height freeze, partial retract, full retract, current reduction, arc transfer (aka arc good, machine motion, arc on, etc), asymetric z axis motion. These are a few items that are often used to fine tune a THC to a particular plasma or cutting application....many are background setups....but some must be adjusted/tweaked for individual part shapes with difficult geometry.

With todays technology all of the above can be easily achieved....the biggest area with need for improvement is in maintaing and controlling accurate, repeatable pierce height. Making IHS work reliably is the single most important factor that makes for automated, repeatable plasma processes at the lowest operating cost (long torch consumable life)

Let's keep this thread going...it will be a great discussion....maybe we'll all get some good ideas on improving THC for plasma.

Jim Colt Hypertherm

I can see Tom and Jim have forgetten more about THC than I'll ever know. Too bad the market for a Galil/Camsoft THC is so small - I'd buy one in a second.

Looks like Tom has shown how to read torch voltage in that Mach schematic. I'll start there but ask a EE to design in isolation for my high frequency start. Don't want to burn up a $2K galil card.



Matt, You need a scope or high speed logger to do any serious investigation. here's a high speed logger for cheap http://www.dataq.com/products/startkit/di148.htm I'd start with Tom's Mach circuit to read voltage. Compare the performace of it to yours. Then adjust some caps and coils, repeat. A quality circuit here is the key. its relative child's play to tell the axis to move.

Karl

One of the issues we have run into over the years is in the voltage divider circuit....if you are designing it for one particular plasma it is relatively easy.....unfortunately, there are a lot of different power supply and starting system designs.

Most of the power supply designs today for entry level plasmas are inverter based....and operate between 10 and 20 khz (power supply switching frequency.....and most of these use contact start torches ....with no high frequency starting and essentially no open circuit voltage as the output is at a dead short (nozzle to electrode) until air pressure reaches the torch to separate these components and create a short circuit spark.

Hypertherm's industrial plasma power supplies are of transhformer design with PWM output to control current.....all of these units use a high frequency discharge (approx 15kvolts at 2 megaherts) to ionize the plasma gas...this high frequency is applied to the nozzle and electrode.....the same location as the DC current. Needless to say....15KVolt can wreac havoc on a voltage divider circuit, as well as low voltage inputs on microprocessor controlled equipment.

Hypertherm's voltage dividers have become far mor complicated that a simple resistor type voltage dividers....with provisions for filtering and conditioning the DC signal as well as providing high voltage relay isolation for the 15Kvolt high frequency...and providing circuitry for ohmic contact IHS functions. What used to be a simple circuit has turned into a more complex circuit in order to cover all applications!

Jim

Jim,

Thankyou for your input, i must say i have great respect for your knowledge, you set me thinking actually, a relay that disconnects the voltage divider from the plasma until the arc ok signal comes on, that would remove a lot of the problems with hf start surely? Of course it would have to be a high voltage relay but that could work.

Thankyou sir

MATT

HI MATT,
sorry I haven't piped in sooner, I think your write up is great, it gets you thinking as you have done to me in the last 6 or so months.
Yes everyone I 'm the one that has the second working edition of Matts DTHC, and I do say it works quit well, and for $100 it works perfect,(for Oneoff or small run things) everyone must remember what Matt is doing is for the Diy guy that wants to be challanged, and with the way the economy is today who isn't willing to try something new, this guy didn't have the $500 for Tom C's unit (Not that I wouldn't of loved to have it) so I did have time one my hands and I tried it, very carfully as I know that a plasma cutter has very high voltage, and If you are scared of that then this is not for you.
Tom C and Jim Colt in the most respect, as Karl T said you two have probby forgoten more that we will ever know, and thank you for giving to us what we have now.
Matt I think we need to move some of our off line chats over here because I think it could help out, help out the Diy guys like us.
Dave Farver

Of course you can buy industrial THC/IHS systems....you will spend a lot of money but 30+ years of thinking, design and field experience has already been accomplished...these systems are reliable under harsh 3 shift conditions.

There is no reason that a DIYer cannot build pretty much all of the features that are in the proven industrial systems.

Much of the cost of commercial systems is in additional features that allow them to interface to hundreds (actually probably thousands) of different combinations of plasma/cnc controls and cutting machine designs.....with different operating software and up to 8 plasma torches that need to have synchronized starting. Add to this machines that have full rotating bevel heads that can countour parts at high speed with up to 50 degree beveled edges (height control gets real complicated when beveling as you are now in the 3D world!). So cost is in the design and support.....but also the z axis slide/torch lifter assembly for industrial machines is generally very robust...and designed to handle splashback of plasma cutting slurry from water table applications...as well as carrying the weight of the heavy industrial torch and lead sets.

I'm certainly excited about helping the DIY segment build full featured height controls....and I would like to see more focus on the IHS/pierce height segemnt of this type of development as it is the best next step to improving the cut quality and consumable life of the plasma process.

Jim

This is great guys letstry to keep it rolling,

on the IHS i feel i should give my experience, i built my machine witha floating head design or a touch off if you like, i have never used a different system but having used the floating head with sheetcam i would never want to, i never have to worry about material thickness itjust finds the top and off it goes, if the plate isnt perfectly level, so what between the ihs and the dthc it doesnt matter, i also get what seems like decent consumable life with nearly no effort, i have never had any instances where i get consumables that look like the ones some people post pictures of asking what am i doing wrong, i am really chffed with that part of it, i can see how a professional shop where time is money could want to economise on it a bit to save time but for me half the reason i built it is because i love to watch it work!!

I agree with jim, dont make a plasma system without some kind of ihs you may well be dissapointed i think.

Which conversations do you think i should put up dave?

For those who dont know what we are talking about me and dave have been talking via email for a couple of mnths about the system and decided that we should put some of the corresondance on here to help others out a little so if there are some randm heaps of text from me thats what it ll be

matt

Karl

Thanks for the link, i never knew data loggers could be so affordable, the ironic thing is it could probably be used to measure the voltage real time for the dthc if you had a mach plugin, theres thinking actually because then you could buy one item, design your voltage divider and optimise it with it then use it for the system at less cost than most of the modbus devices!! I might just be trying that then!!

Thanks

Matt

...

There is no reason that a DIYer cannot build pretty much all of the features that are in the proven industrial systems.
...

I'm certainly excited about helping the DIY segment build full featured height controls....and I would like to see more focus on the IHS/pierce height segemnt of this type of development as it is the best next step to improving the cut quality and consumable life of the plasma process.

Jim

Let me brag up myself for one second... I worked 20 years as a manufacturing engineer upgrading all sorts of equipment to new controls, more production and 24 X 7 reliablity. I changed carreers but I miss the challenge. So, I refit old machines to "better than new". The plasma project is machine number five with Camsoft . I did a couple Ahha and one Mach before that. Anyway, this refit will be professional quality but probably not cost effective.

The torch voltage circuit will be designed by a retired Senior Fellow level EE. Anybody can learn from this guy.

For initial torch height, I have a concept that should work extremely well on this machine. The machine has a shoe on an air cylinder that drops to the work and rides on five roller balls. The Z torch height ball screw will drop with the assembly. A linear encoder will be installed on the air cyclinder travel. To determine intial torch height, just drive Z down till motion is detected on air cylinder; retract to initial height. Now, if you want to run without the rollers on the material; engage a stop pin and only let the shoe drop to one inch above the work. same routine for initial torch height.

The dedicated Galil motion controller will handle THC. It will run this little program in the background at 3 Khz. THC isn't even a minor task for this professional motion control system. The key here is good input data. In addition to voltage, the controler will both set and monitor current plus any I/O that might be needed.

When the whole thing is running, I'll plan on designed experiments to determine the scientific relations between current, voltage, height, and velocity for several materials.

Karl

Sounds like a nice design. You will need to make the air cylinder motion sensing (for sensing the surface) very light so that it does not deflect thin sheets....or better yet sense loss of motion while the torch is rising to the pierce height. You want to start measuring distance for pierce height after you have left the surface of the material.

There are many "plate rider" type plasma height controls on the market for cutting materials thinner than 1/8"....just an air cylinder controlled slide assembly with the torch mounted in a housing with 5 or six rollers.....the torch is fixed so that it is the correct height off the plate when the rollers are riding on the plate. These are usually used in low tolerance applications such as cutting ductwork for HVAC systems. The rollers tend to scratch the sheet being cut and even move it. As well.....there is no means for initial height (pierce height) for the plasma torch....so you are limited to 1/8" and thinner as the torch can pierce this thickness at the same height as it can cut it.

and with the way the economy is today who isn't willing to try something new, this guy didn't have the $500 for Tom C's unit

Remember that price includes the UBOB (buffered & isolated Breakout) and external cards using our new Port Expander technology, so you don't need a second parallel port for added inputs and outputs. It also includes the THC Sensor card with the torch relay, two Arc OK methods and the divider/filter components. Those costs have to be factored into your real costs.

The problem with using a plug-in and reading analog direct is it's just not fast enough to be effective for real time control (same problem as using USB or any serial feedback protocol) Plug-ins run in an outside loop and are not priority. So it does not matter how fast your "data logger" reads it if you expect to use the integrated MACH THC logic. If you introduce a delay in the loop then your servo action is over dampened and the response curve lags way behind real position.

There is absolutely nothing wrong with using the analog control approach (AKA THC300) . It's analog your starting with and it's fast, low component count and it works. The primary purpose of going to digital is all of the other decision and presets you can have.

Tom Caudle
www.CandCNC.com

Ok so i think i may be about to show my ignorance,

Surely both my systemand the other systems out there where mach still controls the z axis rely on the update rate of mach to respond to signals, whether they be picked up through the parrallel port or through a bus? If this isnt the case could someone correct me? and i mean with numbers, i know mach updates at 10Hz what rate does other stuff work at?

matt

Voltage Transducer

Tom and Jim,

I really hate to custom design when a standard product is available. Do you see any reason the voltage transducers on this Flex Core catalog wouldn't work well to sense the plasma torch voltage?

I'd select the VT7 for 50 micro second response, 0-250 volt input, galil card likes 0-10 volt for the output. This product provides 2500 volts of isolation.

Karl

I'm not sure if they would work or not.....it says that the frequency range is 10khz......I'm not quite sure if that will work with inverter based outputs which are generally in the 16 to 20 khz range. (pwm output)

Jim

I'm not sure if they would work or not.....it says that the frequency range is 10khz......I'm not quite sure if that will work with inverter based outputs which are generally in the 16 to 20 khz range. (pwm output)

Jim

Hmm... Are there no capacitors present? Matching a voltage sensor to the PWM out would be extremely difficult if you have no caps. Is it a trade secret as to how to sense voltage?

Karl

Karl, I don't believe there are any trade secrets.....I just am not the electronics engineer that design's hypertherm's voltage divider circuitry. Of course there is an output inductor and capacitors on inverter based power supplies.

I would think that as long as the (transducer) voltage divider reduced the voltage proportionally to the arc voltage....to a level that was useable...then it should work just fine.

Jim

Ok so i think i may be about to show my ignorance,

Surely both my systemand the other systems out there where mach still controls the z axis rely on the update rate of mach to respond to signals, whether they be picked up through the parrallel port or through a bus? If this isnt the case could someone correct me? and i mean with numbers, i know mach updates at 10Hz what rate does other stuff work at?

matt

Think about it. 10HZ is 100msec On a table moving at 200 IPM (3.33 IPS) you are reading a solid .33" (almost 3/8") inches behind where you are cutting. If the Z were fast it would over shoot in both directions. You would be force to put in severe dampening.

Analog to digital is all about the sample rate. The higher the rate the more accurate it is. You can use a 10 bit A/D and sample and at over a MHZ with modern chips, but if you can only get an update every 1/10 second all that speed is wasted.

Parallel port reads in MACH are many times faster ( in the usecs range).



Tom Caudle
www.CandCNC.com

GLOAT

I just scored a really nice DC current sensor
http://cgi.ebay.com/NK-Technologies-DC-Current-Transd-50-75-100-Amps-4-20ma_W0QQitemZ150354068187QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item2301cd02db&_trksid=p3286.c0.m14&_trkparms=65%3A10%7C66%3A2%7C39%3A1%7C240%3A1318%7C301%3A1%7C293%3A1%7C294%3A50

Now I can try some of the things torchhead (Tom) talks about in his thread on THC current sensors.

Karl

Ive been asked by a couple of people to post up a schematic of exactly how my voltage divider works, ive attached it underneath, i have also included a relay activated by the arc ok signal to connect the system to the plasma after the pierce has taken place.

Matt

...
i have also included a relay activated by the arc ok signal to connect the system to the plasma after the pierce has taken place.

Matt


I'm curious, haven't had a chance to look into "Arc OK". What conditions close this contact on the cutter? Also, what conditions open the contact?

karl

Hypertherm calls it "arc transfer" or "machine motion". When current is sensed through the work lead.....a relay closes in the plasma system.....this indicates that the plasma aec has transferred to the plate. If current is lost through the work lead....the relay contacts open.

Jim Colt