Deckel FP3A project

Hi,
I'm new here and new to LinuxCNC, so be prepared for some newbie questions.

I'm in the early stages of trying to convert a FP3A to run linuxcnc. It's a nice piece of German hardware, but also a bit of an odd ball, I think.
The original CNC2102 control is partly dead. Had a 5 volt overvoltage, so the mainboard if kaput.

Spindle:
Mech gearbox. 3 blocks of 3 gear options each, so 9 disc pins. Plus slow, fast, cw and ccw discretes. All 24 volt

Feed drive:
AC servo with DC feedback. 0-10 volt speed control
And here's the real odd part. Only one servo, but 3 brakes and 3 clutches.

And of course all the usual suspect. 24v disc out for Coolant, Hyd pump (for brakes and clutches I think) and disc input for endstop, spidle moving, E-stop etc.

Encoder:
Not quite sure about these yet. Probably quadrature ones, but there seem to be a bit more wires than I'd expect.
I wanted to attached the schematic for the input card, but it's too big. But if you find "Deckel FP3A Nom-Ar-Elec part1.pdf" on this linklink you will find a "NZP50" on pdf page 21. There is one of these "AchsenZählerKarte" (axis counter card) for each axis.
But is might just boil down to A, B and Index with diff inputs.

And now to all my questions:
1) Can linuxcnc even control such a weird machine, with just one servo and the 3 clutches+brakes combo???
2) Can linuxcnc (easily) control this gearbox??
3) Where would you guys "make the cut".
The control cabinet has a s#&% load of relays used for various logics. Including controlling the spindle gear change. Spindle slow start. As well as preventing things from happening, if various conditions are not meet.
I have the diagram of all this, so I would not be too afraid of using it.

Also, the pcb that currently makes the 0-10volt drive control as well as the discrete i/o, seems to be fixed now. But since the mainboard and therefore also the CRT, is dead, I have no way of using this now.
Would you recommend replacing all of this with e.g a couple of mesa boards. If so, which?? That's another jungle atm.

Instead of the mesa way, I have been troying with the idea of dropping an AVR/arduino/raspberry into that pcb, to replace the 6802 CPU. Would be fairly easy to have that set the spindle speed etc. And it would also allow easy access to scan the control box keyboard and switches. For manual feed & spindle control etc.
But where would one tied such a custom control into the HAL of linuxcnc, say it had a RS-422 or ethernet?? If I end up having to make drivers for it, then it's not a viable route. I currently know very little about linuxcnc (as you can hear) so I'd like to avoid making anything too special there.

Sorry for the lengthy post. Hope someone has the power to read through it and provide comments/recommendations.

Jesper

This is the reason the FPxA machines are worth far less than the true NC models.

If you keep the single axis drive through clutches and brakes, you will never be able to perform two or three axis coordinated moves. Kind of pointless to go to all the expense and effort to do a CNC conversion, and then only be able to mill straight lines, one axis at a time. You may as well just put hand wheels on it.

Do you have any photos of the clutch and brake unit and how it interfaces to the axis spindles? Can you remove this completly, replace the three axis drive spindles with ball screws, which you extend out the ends of the axisand put individual motors on? It is probably a fairly big job to make up the ball screw mounts for each axis.

MuellerNick did a Maho 700 conversion which also had a similar spindle gearbox ratio changing system, so that part is clearly doable.

Mark

Hi Mark,

Thanks for your reply.

My post was getting kinda lengthy, so I seem to have lost track a bit. I forgot to mention, that the aim is (of course) to make this a true 3-axis CNC eventually. Me and the boys in the rocket shop (yes, we build real rockets ) all agree on that. Some even dream of 5 axis and tool changer, but the phase 3.

As a first step, I'd like to get the spindle, encoder and drive feed running. In that order. And even if the drive is only one axis at a time with the clutch/brake at first, it will still be a huge help. The other week, I made 6 of these spars . Just drilling the million of dia 14 mill weight saving holes, took close to 2 hours for each spar. Something a singe axis FP3A could easily have done. And yes, I know it slow things down, that we don't have 2 proper vises and have to use the clamps.. but still.

So... I'm glad to hear about the other guys with the gear change. That's a must, to get anywhere.

The maschine already had handles for manual use, so the axis are free to be moved, without doing anything with the brakes/clutches. And at least the X axis has the shaft coming out on the other end, so it's fairly straightforward to add a motor there. Y-axis I dont know yet.
Was briefly considering keeping the current nice big Seimens servo drive and let that run the (heavy) Z-axis. But probably better to have 3 of the same type, right???

What would you guys recommend. Stepper motors or servos drives? Any links to good options would be nice. The current AC servo is 1.4Kw/5nM, so a decent powered one, I suppose. Hopefully the X & Y, don't need that kind of power. But I'm new to CNC, so what do I know...

Recommendations on a MESA product(s) (or other mfr, that might consider sponsoring a crowd funded rocket club), would be very welcome too.
We need something in the order of
28 24 VDC output, (24 might do)
22 inputs. At least. Would like to have another 30ish inputs to scan the existing console buttons/keyboard.
Plus what is needed for the 3 encoder inputs and the new motor drives. And maybe the 0-10 volt for the current servo interim.

Oh, and lastly. Off topic, but heres what the spars is used for .
6.6 meter rocket WIP .

Anybody dropping by Copenhagen this summer, are welcome to come and see first hand. And maybe kick some old German mills....

Jesper

The current single axis is controlled by a +-10 volt.

Also, I do not think the encoders are absolut ones. Unless you add the previous mentioned counter card into the mix. And I have no easy way of reading the cards. So the task at hand there, is to read a (3) quadrature encoders with differential input, including one for index.