[mmurray70]

Hi Guys,

I've had a lot of people ask me details on my Kflop retrofited 1994 Fadal 4020. I decided to make a thread about my machine and post some things ive learned and some of the wiring information i have figured out about Fadals along the way. Hopefully it makes things easier for someone. So heres my story...

I bought my machine last summer and the previous owner had already started the Kflop conversion. He had the three axis moving, spindle somewhat working, waylube and drawbar working. Basically got the bare minimum done to actually make a part and then stopped. Bit of a sloppy job as well, but i had a good head start with the machine. Here are some pics of the machine when i got it:

[mmurray70]

This machine uses DC servos and the original motors and drives are being controlled by Kanalog. The original motors had tachometers and resolvers and velocity mode servo drives. The previous owner removed the Tachs and resolvers, added encoders and modified the drives to operate in torque mode with Kanalog controlling them. Looks like he bypassed the velocity loop and soldered signal wire directly into a specific point in the amplifier farther down the line. This is working well, no issues with motors or drives so far.

When i got the machine I cleaned up alot of the wiring and traced nearly every wire in the machine and got rid of anything not needed. Nothing at all worked in the original control box on the front of the machine so i stripped that and added a LCD display, keyboard, buttons and overrides and an MPG so this machine could be operated like a normal VMC instead of fiddling with a laptop. I had to add a Konnect board to get the number of inputs and outputs i needed.

[mmurray70]

Controlling all the outputs on a Fadal is a piece of cake, couldnt ask for an easier machine to control. Fadals use a power disribution board that has built in solid state relays and fuse protection for almost every output. The SSR's are powered with 5v from this board and the relay (and output) is turned on by connecting the other pin to ground. All these relays are controlled through a ribbon cable connector and winford sells a breakout board to connect to it: https://www.winford.com/products/brk2x17.php

There are two of these relay boards, one in the side cabinet that is connected to the breakout board, and another in the rear cabinet that is factory connected to the first board with another IDC cable. The breakout board is only connected to one relay board, but it will control both of them through this other cable. Its as simple then as getting Kanalog or Konnect opto outputs or relay outputs to connect a certain pin to ground and a particular function (coolant, drawbar, waylube, ATC motions, etc) of the machine is turned on. See the notes ive attached for which pin controls what. Please note that my Cable is most likely backwards! I bought the ribbon cable and ends from winford and made my cable this way so the wiring was neater. If you use a factory made cable the pin numbers will probably be all different. See pic below, it needs to be the same as this without a twist.

The spindle is controlled the same way, but it doesnt go through the breakout board. There is a small three pin connector at the bottom of the board (its hidden by a cable in pic sorry). The middle pin connected to ground turns spindle on and it will shut off when disconnected. The right pin is for reverse and the left pin is not used.

All of the sensors for the tool changer I connected to my Konnect board. They were powered by 5v and some had a 220 ohm resistor inline as well. See the notes I made for details on this. Sorry its not neater but dont have time to type it all.

[mmurray70]

Heres a pic with Konnect added and everything in the control box finished. The machine has WAY more stuff connected and working in this picture then when i got it, and yet it looks much cleaner and neater then before.

So thats about it for now. The project was alot of work, but its a great machine now and it didnt break the bank. Great learning experience too. I dont have any electrical background. Im a mechanical guy with basic electrical knowledge but was always interested in electrical stuff. I wasnt sure how to control a simple output like a coolant pump when i started and I had zero experience with C programming. After watching a few videos, looking at example programs and some help on here im fairly comfortable with it now. And its awesome to have so much control over what your machine does with the C-programming. Really happy with the Dynomotion system overall, and the Fadal ended up being a great machine to retrofit. Hope this info is a help to someone and thanks to everyone who helped me along the way.

Also please verify the information I posted for yourself, a few things may have been moved around on my machine. Be careful and good luck with your retrofit.

Mark

[COlger]

Mark,

Nice work! Thank you for providing the details of the retrofit. Are you using the Index marker for homing the axis?

Chris

[mmurray70]

Mark,

Nice work! Thank you for providing the details of the retrofit. Are you using the Index marker for homing the axis?

Chris

No unfortunately my encoders do not have an index mark. I wish they did and would recommend anybody sourcing encoders to look for models with an index. I hate to admit it but I still haven't added any home switches yet. Im able to use the mill fairly well as it is so the home switches got moved to the back burner until I get some more free time. When i get my lathe done ill finish the home switch issue.

For anybody not used to Fadal, they dont actually use home switches at all. They have a visual "cold start position" marker on each axis and you have to line up these marks and run a cold start proceedure which looks for the index mark and thats it. Position and limits are controlled by the software after that.

I've been shutting down the machine at the cold start position, and visually double checking all 3 marks on startup for now. Not the best setup, but good enough for travel limits and toolchanger for now. It actually works surprisingly well like this, but its obviously not the right thing to do and i don't recommend it.

[COlger]

Mark,

This might be a question for DYNOMOTION, but is the KANALOG board setup to use the Index channel? The schematic looks like it only uses AB?

Chris

[m_c]

Mark,

This might be a question for DYNOMOTION, but is the KANALOG board setup to use the Index channel? The schematic looks like it only uses AB?

Chris

For the index channel, the usual way would be to use another encoder input. The way the way the Kanalog encoder inputs are designed, is although the A/B inputs connect directly to a hardware encoder counter, each A/B channel can also be used an individual/standalone input.

To connect an index channel to a Konnect, you would need to ensure it was capable of operating a Konnect input (you would probably need to use some kind of buffer, to ensure you get the required 4.75V needed to guarantee an on state, as a standard encoder may struggle to reach that voltage), and you would have to home slow enough to ensure the index pulse is long enough to guarantee the Konnect input stays on for long enough to be detected (Konnect boards don't read/updated by the KFlop as fast as KFlop/Kanalog inputs - Going by my notes, it works on an approx. 100uS refresh rate for a single board, so you would need to guarantee your index pulse was present for at least 100uS).

[TomKerekes]

Hi mmurray70,

Thanks so much for sharing. You do nice work. I especially enjoyed the completely empty card cage and the pile of wires

Regards

[mmurray70]

Hi mmurray70,

Thanks so much for sharing. You do nice work. I especially enjoyed the completely empty card cage and the pile of wires

Regards

Thanks Tom. My friends/family thought i was crazy to tear all those wires out of my newly bought machine, nobody thought it would ever work again lol. I wasnt totally sure at one point but it all worked out good.

[mmurray70]

I had a request for my C-programs so here is a copy of all my C-programs and the KmotionCNC configuration files. Not sure if i got them all, if theres anything missing let me know. Currently using 4.34i

Please dont try and use this code as it is. I would suggest creating your own C programs for your machine and maybe use this as a reference for ideas. Im not an expert programmer, might have been lots of better ways to do things. But my machine does work well now with this code. Every function is working properly, and every sensor is checked during toochange etc. Only thing missing is homing, sorry, see post above. lol

[Need TECH Help!]

Hi Murray,
Looks like a nice clean machine. Its something to see all the open space in cabinet after all the "old junk" was replaced with Dynomotion.

Troy

[kn6za]

HI Murray,

Nice job with the Fadal!

Thanks for making your C files available. I have used some great examples from your code.

I have a question about your tool change program.

You use Requested tool = persist.UserData[6]

In my setup KmotionCNC is using persist.UserData[9] to pass the requested tool.

How did you get kmotioncnc to pass tool data in 6

Andrew

[mmurray70]

HI Murray,

Nice job with the Fadal!

Thanks for making your C files available. I have used some great examples from your code.

No problem, glad to hear its helping you out.

I have a question about your tool change program.

You use Requested tool = persist.UserData[6]

In my setup KmotionCNC is using persist.UserData[9] to pass the requested tool.

How did you get kmotioncnc to pass tool data in 6

Andrew

When you configure the action for M6 in kmotioncnc you specify what variable the tool number gets passed too. See the attached pic. Good luck with it.

[kn6za]

Thanks Murray!

[hilo90mhz]

Hey Murray,

I just purchased a 1989 4020 Fadal! the control works but am not so keen on learning the ins and outs of a 30 year old text based computer system that I will likely want to replace at some time in the future....

I have been looking at the Kflop or the Centroid Oak boards.

The Kflop is MUCH less expensive and I like the free nature of the software without being nickle and dimed for every option.. I am familiar with C. So very much leaning towards the Kflop if I do retrofit the machine.

Do you think the Flop enhances the machine enough to justify the 100s of hours of work? If I keep the existing DC motors and amps the cost is very reasonable.. $500 for the Kflop + Kanalog and probably less than $1000 in parts total for the complete conversion. Centroid Oak with software options for tool changer would be about $2500 total. Resale value is not a concern as this machine is not going anywhere.. Mainly interested in usability for all types of CNC work, I want to do complex 3d stuff at some point but its not a priority.

Do you have any pictures of how the DC motor drivers were modified? I saw your note about a wire being soldered further down the control loop etc..

I am thinking I could save much of the software setup time by using your config if you will share Do you have the newest versions online somewhere?

I built a 4'x3' 130W laser cutter with Panasonic servos from scratch last year so I realize the amount of work it takes to set something like this up.
http://hilo90mhz.com/laser-cutter/

Thanks,
Chester

[mmurray70]

Im still very happy with the Kflop setup. It will be tons of work to convert as you mentioned but it is a great setup when its done. If you already know C, sounds like the way to go. Its awesome to be able to customize your machine to do whatever you want with C programming. I can share the latest revisions of my files if you want them.

Honestly i've never used the stock control so cant really say too much about it. People say it will get the job done, but doesn't seem like a real nice interface to me. And I doubt it would be as accurate as kflop at any speed. Most of my prior experience is with a late 90's fanuc VMC and a Mazak 510c II. The kflop controller is WAY ahead of the fanuc in every way you can compare it, light years ahead. And In many ways I like it better then the Mazak.

Im working full time with this machine now in a job shop and its been working great. Does everything I need. Mostly basic 2.5d stuff, some 4 axis and the occasional 3d part, whatever comes through the door. Only slight complaint would be that it doesnt seem to smooth poor quality 3d toolpaths quite as well as the Mazak I used to run, but I think Tom is working on an improvement and honestly its a very minor complaint. 95+% of 3d stuff runs just fine.

I think if you can afford the time its definitely worth doing. I would go the same way again for sure if I were to do it all over again.

[hilo90mhz]

Thanks for the insight! - I am probably going with the Kflop option.

Is this your machine on the Dynomotion YouTube channel? It sure looks awesome if it is! https://www.youtube.com/watch?v=qfY4WMXaTAw

I just found the Zip files you shared on the first page of this post that is quite a bit of info.

LinuxCNC +Mesa is the other option I am considering, but it seems like it would always have more latency in the servo control loop vs the Kflop all firmware system controlling motion.

The Kflop user interface is a little rough around the edges, I have seen the editor but there is only so much it can do it seems.

Do you have any pictures of where the analog torque control wire is on your axis drives? that would help a ton

Also a new copy of your Kflop software setup would be amazing

I am very impressed with your diagrams showing how you seem to have used all the hardware sensors on the machine - I don't think most people are as thorough in their conversions.

The code in your ToolChange.c file is very well commented and looks great!

[mmurray70]

Yes that is my machine on youtube. Glad you like it, thanks. Heres a pic of the signal wire on the servo drive board. Your drives may be different. Some of the older fadals had much finer pitch "inch" ballscrews and smaller motors and drives. Mine is a 1994 and has the larger motors, drives and much coarser metric screws.

The previous owner of my machine cut a resistor from the board and fed signal directly into the part of the circuit right after the loop gain potentiometer. The manual for the GA4568PA (my drives are actually GA4568EA) drive describes the loop gain pot on page 10 as being able to reduce current to zero if turned fully clockwise and the diagram on page 41 shows the loop gain pot as being close to the end of the pre amp. So it appears this point is used to bypass all the velocity mode preamp and feed signal directly into the power amplifier. Im not 100% certain but it appears like this might be the only modification needed to change the drive to torque mode.

I've heard good things about linux but for me that would be a whole new thing to learn as well. I like how kflop has its own dedicated hardware and just uses a regular old PC to send it instructions. I wouldnt look at mach or anything else to control a machine like this.

Im fairly busy at the moment but ill post all the current files someday for you if you go ahead with it. Good luck with it.

[mmurray70]

Heres some 3d parts I made today. These ran perfectly and turned out really nice. These are parts of an intake for a turbo charged snowmobile.