Printable View
Wow thanks for all the grounding tips. I have a small pile of ferrite rings lying around I think I'll get him to try it. The ONLY digital signal we have (besides the steppers) is the relay to the torch on/off. If the signal is messing with the ground on the relay side, I can see that causing all sorts of reference problems with the driver board :O
I'm not sure it this will help but on your table but I attach the plasma cutter's work ground clamp directly to the metal I am cutting.
Reducing any gaps that the current needs to jump will reduce the noise in the system.
I have used both Everlast PM256 (plasma HF start) and a Longevity Force cut 80I on my table.
http://www.youtube.com/watch?v=WdG9_ZQSxCg Everlast PM 256 with supplied torch.
Have fun
Tom
I haven't done much in the last couple weeks. It's been a combination of being tired of getting cold working in the garage and travel for work.
Today I just got the Rev C shields back from fab.
Attachment 9870
After I sent it off I realized that I didn't run the "torch on" signal lines on the DB-9 to a solderpad so it can be wired to the relay. As soon as I got it, I saw that the power jack footprint was wrong.
This rev of the board goes back to the ground plane configuration used in the first rev (that worked). New to this rev are:
- DB-9 connectors for cabling to the CNC and PP50
- 3.3V or 5V supply jumper
- Power jack w/ power supply regulator
- Standard shield layout
- no support for a UI or buttons (will rely on control interface application on the PC or integrated into Mach)
The goal with the standard shield layout, the on-board power and 3.3/5 volt jumper is to allow the shield to work with both the Arduino and the Freescale Freedom board (32-bit ARM that costs $12).
I hope to get this build and testing this week.
Progress has been painfully slow.
One big time waster was messing up wiring of the DB-9 connectors I added. The connectors make board assembly and connection to the Plasma/CNC a breeze. The issue is that I reversed a lot of wires (use the wrong side for pin one). It resulted in the board's DB-9 connectors not matching the cables I already had for the CNC and Plasma. That's taken a couple weeks to fix.
This rev of the board removed all of the LCD software. It now runs much faster (I need to a delay loop to keep the control at 1 ms. timing). But, when ripping out the LCD, it took a while to get the software "fixed". I had lots of problems with the Ardunio Eclipse plug-in I've been using. I finally gave up on it and starting using the Arduino editor.
To support operating with no LCD, I put on new faster opto's for the serial port allow 119.2K baud. It took a while to debug that because of a bad opto and not using shielded cable. Shielded cable is mandatory or you can't really communicate. I also added message sychronization and checksums to deal with communication interference.
I finally got to the point of working on the voltage filter circuit just this afternoon. That is really wacked. I had one "wrong part" installed on the voltage divider. I'm not sure if this damaged components. Figuring out the voltage filter circuit is the last hurdle to get it working to the point to be able to test.
Just got the last problem figured out to bring the board up. It was a software error in the PC control app. The app currently looks like:
Attachment 9906
The plan is to integrate the interface program functionality directly into Mach when I have it all working. (I've started playing with the Mach interface code. It looks do-able.)
I finally got a successful data capture (broken wire issues). It has almost two volts of noise on the signal.
Attachment 9907
After magnifying the graph it was easy to see that the noise is at 60 hz, on the nose. On this rev of the board I added the two pole filter to the back end of the op-amp filter. I'll just need to tweak values to see if I can improve the noise response.
If I can, then this is a pretty solid design. I already have a Rev D design that is identical to the current one with the only major change being the addition of a torch-on relay on the board (I currently have it on the side).
The current board looks like this:
Attachment 9908
The bottom left hand corner is missing the power jack and the voltage regulator. Not necessary, but put there so it can be used with the Freescale Freedom board.
Having the torch connector in the upper left corner is a problem. If you push down on the board, the pins from the DB-9 short on the metal housing of the USB connector. For Rev D I moved it to the botton of the board.
The layout of the Rev D board is done, except for fixing the footprint of the voltage regulator. It's also a little wider.
Attachment 9909
With my son's help, I figured out what was doing on with the 60 hz noise. I hadn't realized that the frequency I entered in the R-C filter calculator was where the filtering started not the frequency that was eliminated. So, a 60 hz filter only started filter at 60 hz. An 8 hz filter reduces 60 hz noise to pretty much non-existent.
I went from a 10 uF final R-C filter capacitor to a 40 uF to a 100 uF. (The 10 uF and 40 uF were tantalum, the 100 uF was electrolytic. I have some tantalums on order to see if they hold the voltage less time than the electrolytic.)
The graph below shows the 40 uF and the 100 uF. The 40 uF has a span of 3 to 4 analog counts (1/2 of a volt on the input). The 100 uF has a span of 1 (1/7 of a volt on input). I'd like a little more filtering than the 40, but not the poor response of the 100.
Attachment 9992
After removing the LCD and push buttons, the code size, RAM usage, and processing overhead all dropped. This should now work on an Uno (About $15 on eBay versus $25 for the Mega.)
I have already sent off the Rev D board that integrates the torch control relay. It should be in by the end of the week. After that one, I think I'll be done with the hardware and focus on integrating the control of the THC into Mach. It's usable now with the PC app.
After I have the Rev D board working, if I get enough interest I'll write up a guide on building and using it and then get some boards made to sell (probably about $5 + shipping - priced to get it out there versus make a profit).
Looks very good Nest. The 100uf signal looks damn clean.
I would imagine even the 40uf signal would suffice for THC
Can you just hook to of the 40uf caps in parallel to test?
The concern with the 100 uF is that kerf detection won't work that well. With the noise on the 40 uF being about 1/2 volt, I can probably only control the torch height to 3/4 volt (probably good enough, but I'm trying to get it as good as possible.)Quote:
Originally Posted by sportbike
To get the 40 uF I had to use (2) 15 uF's and a 10 uF. I ordered some 47 uF's and 68 uF's (only off the shelf values) to see how they work.
You'll appreciate that it took my son the Mechanical Engineer to straighten out my electronic filter design :)
Of course....Mechanical Engineers can figure out anything if we need to. :cool:Quote:
Originally Posted by EmptyNester
Glad to hear the filter got worked out. There's always a trade-off in everything you do, so now you just have to decide if the values you have are good enough to work, or if its worth the extra time and experimentation of really pinpointing the exact values you want. You do realize you're going to have to package all this together and start selling them to board members soon right?
Yeah. Part of the object of this was to come up with a DIY solution for brave souls. I figured once I was done I'd order boards from SeeedStudio (China) because I can get lots of 10 for about $5 a board. Since I'm doing it for fun, I figured it best to sell bare boards to discourage people who want a packaged solution.Quote:
Originally Posted by agent4573
I'm anxious to get other people working with it and helping refine the behaviors. I'd love to see someone modify it to work with plasma units that don't have a CNC interface.
I have been working on tuning the final filter stage. The graph below shows the difference between 40 uF, 68 uF and 100 uF used in the low frequency filter (for 60 hz).
Attachment 10045
While the 100 uF signal is really clean you can see the response time suffers. The 40 uF capacitor gave a swing of about 4 counts (a little more than 1/2 a volt on the torch). The 68 uF gave a very similar response but only had a swing of 2 counts. That's the ticket.
I made a couple changes for the Rev E boards I just sent out:
1. Added a spare capacitor on the voltage divider (not necessary but allows easier customization)
2. Added a zener diode on the "Arc Good" signal from the plasma to protect the Arduino from misconnection
3. Fixed the package layout for the torch-on relay
4. Fixed the package layout for the power jack
I ordered 10 boards from SeeedStudio four days ago. I got a email today saying they shipped. So, I expect I'll get them in about two weeks. It was about $50 for 10 boards.
I've started writting the manual to accompany the board.
After I get the boards and build one to verify it, I'll offer the others for sale to folks in the US ($10, $5 for the board, $5 for shipping). I don't want to make money, I want some folks to build it and play with it.
I expect it will be at least a month before the new boards have arrived, and one has been built and completed testing. If you're interested in one, PM me with an email address. I'll email the manual first, when it's more complete, so you can look at that and decide it you are still interested after seeing what is involved.
As I'm waiting on the new rev of the PCB that addresses the last couple layout mistakes, I decided to try an Uno board. It's a smaller and cheaper board than the Mega. Without the LCD display, the software is small enough to fit on the Uno.
I found that the plasma voltage signal on the Uno was very noisy compared to the other boards. So, I did a comparison of all the boards I had (1 Uno and 4 Mega's).
The Mega's were very consistent but the Uno was lousy. I don't know if it is just a "bad" Uno board, or it's a fundamental/design issue with the Uno.
The graph below shows the a comparison of all the Mega's and the Uno. It's interesting to note that the same THC shield resulted in different cutting voltages with different Arduino boards. (I use touch and go, so they should be pretty consistent.)
Attachment 10060
It's about a 5 volt (plasma torch volts) difference between the highest and the lowest.
Still waiting on the next rev of the printed circuit boards w/ the minor fixes.
I started working on making the THC talk with Mach. It took a while to find all the info I needed, but it wasn't too bad. The screen design in Mach is another story.
There are a bunch of different screen layout tools, and I think they all have a respectable learning curve. And, that's not something I'm really that interested in doing. So, I took the standard plasma screen and got rid of some controls I don't use (never bothered to even learn what they did). I've put the THC controls on the bottom right of the screen:
Attachment 10077
The top left of the panel shows the current THC mode of operation and has the button to switch modes. The green LED to the right of the "Arduino THC" label is a status LED. It's green if it is talking to the THC and red otherwise.
The display can be either in volts or counts. It shows 2 volts as the current voltage due to noise. I had code to compensate for that but took it out for debugging since a non-zero value helps.
The arrow buttons will adjust the set point. There is an LED with each button. They show if the THC is sending a torch up or torch down signal.
The bottom right has a checksum error counter. If the serial cable isn't shielded, you can get so much noise the system doesn't work. By showing the counter, you can tell if you're having communication problems.
I still have to do the cruise control button. I can also put the tip life tracking in, but space is getting tight.
I finished debugging the Mach interface and went to test it with my table. I got a Mach error because I need to add the .NET 3.5 upgrade to XP. Unfortunately, when I "optimized" my XP for use with Mach, I apparently killed the network adapter. So, it may take a while to rebuild the PC so that I can do a live test.
Assuming the Mach interface works as well live as it has during testing, I still have more work to do on that. It's currently hard coded for the serial port, so I have to add some configuration ability to it.
I got the PC boards back from the fabricator (SeeedStudio). I now have two versions. The one "flavor" I'm been working on all along that I call the "Shield" and a new configuration that uses an Arduino Pro Micro that I call the "Micro". Both are Rev E. I refer to them as "Rev EM" for the Micro version and "Rev ES" for the shield version. Both have identical filter and interface circuits, they just use different Arduino's and have different configurations for the Serial to USB converter.
Both work very well and the ones I just completed construction on had very clean cutting signals (+/- 1 count, or +/- 1/7th of the cutting voltage). Both are shown below:
Attachment 10239
The top is the Shield version and the bottom is the Micro version.
The Shield version is "clean" and doesn't require any "white wires". This is the first pass of the "Micro" board and it needs two white wires (I actually did it by scrapping off a little solder mask and putting down a solder bridge in two different places.)
I have to say that I like the Micro version a little more. Overall it's just a little smaller and shorter. It doesn't have the issues with the Shield version where you have to put electrical tape on the USB shield to prevent it from shorting connections. It also mounts the serial to USB port right on the board so you don't need to make a custom shielded cable, you can just use a normal Mini-USB that is shielded.
I have a draft of the construction manual but have to update it with all the new pictures for the new boards. The board software works. The stand-alone PC app works and the Mach plug-in "mostly" works.
When you order from SeeedStudio you get 10 boards. So I have "spares" and if anybody in the US wants to build a THC for their Power Plasma 50, let me know and I'll send you a board for $10 ($5 for the board, $5 for shipping). At this point I don't plan on changing the hardware for a while because I want to focus on the software and cutting some steel (and tinkering with the hovercraft).
Looks very nice.
Were the seed studio boards assembled or were they blank PC boards?
Great project by the way. I love DIY stuff like this.
Thanks. They were bare boards. It just cost too much to get small quantities of boards made.Quote:
Originally Posted by sportbike
Is there a place to down load the Adriano sketch for the THC?
I updated the manual with directions. But, you have to download that :) The updated manual is on Git Hub. But, I've copied that section of the manual below........ When you download the zip file from Git Hub (as directed below), you get a 5 subdirectories:Quote:
Originally Posted by acourtjester
1) "ArduinoSource" (the Arduino THC source code that can be compiled and downloaded)
2) "HardwareDesign" (Eagle files for both versions of the board)
3) "Mach Plugin" (Mach files to talk to the THC w/ a display screen for the interface)
4) "THC Manual" (The instruction manual for the boards and software.)
5) "WindowsSource" (The source code for the Windows interface program to run the THC.)
The section from the manual is:
7. Programming the Arduino
- Download the Arduino Tool Set
The Arduino compiler and tool set are constantly being updated. The currently supported version is 1.05. The software should work on new versions, but changes in the compiler can sometimes cause compile errors that require minor changes.
To download the Arduino tool set:
1) Go to www.arduino.cc
2) Click the “Download” label in the top title bar
3) You can use either the “Windows Installer” or the “Windows (ZIP File)”.
a. If you use the “Windows Installed”, run the program after downloading
b. If you use the “Windows ZIP file”, open the file after downloading and copy to a convenient location (usually “C:\Arduino-1.0.5” (or something similar)
Run the Arduino IDE and make sure it works.
If you’ve not used the Arduino before, try some of the example software to get the hang of the software and downloading.
- Download the Source Code
To build the software and load it on the Arduino, you need to have the source code. All source code is maintained on GitHub (www.github.com).
1) Go to the Git Hub site
2) Enter “regeg/ArduinoTHC” in the search box on the top menu bard of the window.
The Git hub name you search on is case sensitive! You must match the case: “regeg/ArduinoTHC”
3) Once the project page comes up, click the “Download ZIP” on the right size of the page.
4) Save the ZIP file somewhere convenient
5) Open the zip file and copy the directories to a convenient location
- Building and Downloading the Source Code
Now that you have the source code downloaded, you must build it and program the Arduino.
1) Start the Arduino development environment
2) Enter the “ArduinoTHC” directory (in the location you previously downloaded it)
3) Enter the “ArduinoSource” Directory
4) Enter the “THC_REV-EM sketch”
5) Open the “THC_sketch.ino” file
6) The project should now be opened with a number of tabs across the top (all the files in the project)
7) Select the “Platform.h” file.
8) Set the “define” for the hardware you are using
a. If you are using an Arduino Micro Pro, make sure the line with “#define MICRO_THC” IS NOT commented out (a “//” denotes a comment
b. If you are using an Arduino Mega, make sure the line with “#deifne MICRO_THC” IS commented out.
9) Plug in your Arduino (if using an Arduino Mega, unplug the shield before programming)
10) Select the “Board” Menu
11) Select the board type you are using, either
a. Arduino Mega 2560, or
b. Arduino Leonardo (for the Micro Pro)
12) Select the serial port that the board is on
13) In the tool bar, click the icon with a check mark. The software should compile without errors.
14) Click the icon with the arrow pointing to the right to download the code to the board
Once programmed successfully, you should be ready to go.
Thanks Empty
Man you have tons of time in this.
Have fun
tom