Considering retrofitting small VMC -- Leadwell TDC-450

I have the opportunity to purchase a 91 leadwell TDC-450. The cost would be about $2200 including delivery. It seems like the most recent place used it for random R & D jigs incidental to their business, was working when stored, but not under power now. Manuals included, tooling included (making the cost maybe more palatable if I failed, loss wise, not sure), etc. The batteries apparently die frequently on these, but initialization settings are available in the manual. They used a g-code feeder over parallel. I might try to see if I can get it started as is to see what is actually OK or not, but unsure.

Rough images here, the last image is an example from the web, not the specific machine: imgur.com/a/IJyWXYY
 

It seems like linuxcnc + Mesa are pretty much my only clear path forward, but I am curious about considerations/cost/difficulty, and if I could get any clarity on how hard it would be to get certain bare minimums working. 

-It has an rf203C servo amplifier board, which I assume will be useless and I'm looking at ~$1200 for 3 reasonable servos + drivers that will not be a total pain to tune for linuxcnc (correct?)

- I was told it isn't perfect on spindle runout, but he is not sure what it was indicated at. I didn't know that I have to really crank the tool to get it to turn in the off state, so I didn't check runout. I can. The spindle replacement is a big unknown to me, if something were to happen. I'm not sure if I can skate by if the runout isn't awful, and if there's an option for a replacement that isn't exorbitantly expensive. I'm not sure what the cost for re-bearing and re-grind would be, if it came to that. My goal is to have a mostly solid, capable thing for playing in aluminum and learning. Are there lcnc friendly replacements if things go to hell, or other options?

-It seems to have an ATC that is rotated when the spindle reaches the top of the Z, which meshes with some gears to turn the ATC, then lowers to pick up the tool (cam releases draw bar). How hard would it be to figure out setting this up with linuxcnc, eventually? Could I get going with semi-manual tool changes somehow, while I figure out encoder feedback or something similar?

- On the issue of ATC/Encoder, are these spindle motors (3-5hp? I think) typically easy to drive with a VFD, plus linuxcnc somehow providing feedback to this type of motor to advance to the right ATC position, if this is all mechanically coupled like I am assuming?

- Sensors and the like, is there a path to getting the baseline movement/functionality, and perhaps ATC with a little bit of time, without a gigantic deck of extras besides X/y/Z/Spindle control, to start?

- I'll have to remove the Z servo and put it on a block to be able to fit under the 81.5" garage door, it seems I have about <79" total in machine height, but rolls on a pallet. Any ideas on how the ballscrew is coupled to the Z servo, because I definitely want to know it won't be a huge hassle to try to remove

(I get that things like linear hardware may need to be re-done, but that seems acheivable with chinese replacements like the diy gantry router I made with hgr20/2020 ballscrews from china for an OK price, I imagine)

I understand that it could be a decent investment to get operational, but the low initial price and the size has been appealing. I would like to avoid getting into a situation where I start it, spend a loooot of time and money on equipment that is specialized and worthless (I'm fine with servos, those are useful to me) trying to even get moving. Are some of the concerns about ATC/sensors/trying to run it manually adressable vi linuxcnc, or might I run into some obvious issues I'm missing?

I'm not looking for perfection at first, but happy to really make it excellent after I can clearly understand things are going "OK." It would be great to get things moving around within a few months timeline, if that's even feasible (sans some shipping waiting). Pretty much everything else I have seen has been about $8k+ and almost never the appropriate size, so this seems like a potential path to something I wouldn't consider otherwise. 

Any input on any things I should be thinking of in general or that I haven't thought of with respect to control scheme is appreciated!

       

some videos of other machines running:    

Aha, thanks so much for answering more servo related questions! I recently bugged ya via an email when I was trying to figure out STMBL current state, thanks again for helping with what you know here (and I saw STMBL comments on the hackaday industrial robot arm article recently too, haha).

It would be really fantastic (and would allow for some other things to go wrong and be OK) to be able to retain the servos and I don't mind playing around for a bit to see if it's potentially feasible.

this is mitsubishi MR-S3 3-axis servo amp. NC is MELDAS 300 i believe. the
unfortunately no analog control. parallel digital bus which nobody has implemented afaik. the stuff is old, released circa 1986...

the drive manual:
s3.amazonaws.com/Icarus/DOCUMENTS/Mitsubishi_Manuals_1053.pdf

aha! So looks like I'll end up with some more non-functional servos for the pile, if I go for it and retrofit.

Thanks for the servo manual, super useful to know the actual amp.

I think this particular one is Meldas 310, yeah, from the manual I saw on site

I'm pretty sure I recognize the spindle motor - also a Mitsubishi.  Basic, but high-quality, 3-phase induction motor.

If the Meldas spindle drive is also taking a proprietary serial control signal, there's a way around it.

The Mistu (tokugawa) resolver(?) can be replaced with a universal incremental encoder and the whole thing run off a VFD.  I went with a US Encoder optical, and they sell discs that match the shaft nicely.  You'll have to turn a thin mounting plate for the encoder pickup to sit on (there's a big hole through the motor end cover), but otherwise my Mitsu spindle motor re-purpose was pretty painless.

US Encoder, Automation Direct VFD, Mesa & 0-10v = profit

Photos, or it didn't happen...

Check the shape of the motor housing and cooling shroud... looks pretty similar, no?

 

It does look fairly similar, though there isn't a flange/flap at the front, I think from online photos:

 

I wouldn't have any experience, though

It is cool to know that encoders can be retrofitted if the current encoder is proprietary. I still certainly don't know exactly how the ATC works, but hopefully I can figure it out if I do go for it.

Edit: finally seem to at least have found a few machines with a similar ATC carousel mechanism, this is from a faunc robodrill t14ia:
 


 

Yep, that's a Mitsubishi spindle motor.  Identical to mine - just bigger.

In my photo - taken during the lobotomy - the fan, fan shroud, and OEM encoder/resolver/thingie covers are off.  The "tab" thing to the right of my photo is part of the motor frame casting and acts as a mount for the junction box.  The top-down photo you first posted shows the U-shaped junction box mounted to this tab.

The color and fan shroud shape are a dead giveaway.

btw, i believe the FR-SFJ spindle drive can be controlled without the bus interface card (SF-TLJ), with regular analog voltage and discrete i/o.
the SF-TLJ however contains the "magnet sensor" spindle orientation function and encoder output. maybe it can be left in place with the bus interface control disabled, but i'm not sure...

www.manualslib.com/manual/1253557/Mitsub...#product-FREQROL-SFJ