What's new
What's new

cnc retrofit

bigIron

Plastic
Joined
Apr 26, 2005
Hi everyone,

I have a Van Norman #12 ram-type (head pivots from horizontal to vertical, no quill) mill. I would say that it needs to be adjusted only, but for a 60 year old machine it's in quite good shape (origional paint, despite its fair share of blemishes.. no rust). I'm hoping if I adjust it correctly I can get some of the excessive play out of the X-axis handwheel, as well as make the knee a little less of a workout to raise and lower. Y axis is perfect.

Anyway, I'm a poor college student and I'm becomming addicted to machinist hobbies rather quickly in my spare time. I am considering retrofiting my VN to CNC as there is no chance of me getting anything more modern for a long time. My question for you gentleman is this:

Is it even practical to convert such an old beast to CNC considering its rare spindle taper (5V) and perhaps other undesirable aspects I am probably not even aware of?

here is a link that shows a #12 exactly like mine:
http://home.att.net/~JEKasunich/vannorman/VN_Home.htm

If it you guys think it is worth it, and asuming I'm successfull in correcting some of its issues, what kind of torque would I need in stepper motors? Do I have to buy something huge to lift the knee, or can I just gear down?

Thanks for reading my long winded question!
Scott

PS> Bought this mill at an estate auction with 5hp phase converter for $200.
 
Um, yeah - that could be retrofitted. I wouldn't use steppers, though - I would use servos. I make a post about it over here:

http://bbs.homeshopmachinist.net//Forum1/HTML/010644.html

Hmm... Looks small enough that taking off the table wouldn't be such a burden... I think that a ballscrew upgrade might be in order, as the first thing. As you have more time than money, see what you can scrounge at industrial surplus places in your area, and on ebay. Also, check this guy's site out:

http://www.homeshopcnc.com/

For a control, I am partial to a Mini-ITX motherboard, driving Rutex servo drives, all mounted in an electrical enclosure, and with a standard office-type monitor arm hung off of the side of the mill.

From a recent eMail about this that I sent to another member of this forum:

The reason that I like Mini-ITX boards is that they
are self-contained, yet common enough that I'm not
paying an "Industrial PC" premium for a "Single Board
Computer" (even though it's technically comprised of 4
PCBs, there is no card cage required - a simple 4-hole
mounting pattern and a set of standoffs will do).

Check out:

http://www.mini-box.com
http://www.logicsupply.com

For TurboCNC or DiamondBack, I would go with something
like this (total $190+tax & shipping):

http://www.mini-box.com/s.nl/sc.8/category.15/it.A/id.160/.f
EPIA 800Mhz C3 Motherboard w/processor, $100

http://www.mini-box.com/s.nl/sc.8/category.13/it.A/id.222/.f
DC-DC Powersupply card (takes in 12V, attaches
directly to the ATX power supply connector on the
motherboard), $35

http://www.mini-box.com/s.nl/sc.8/category.14/it.A/id.243/.f
CF Card Adapter (allows you to plug a CF card directly
into the IDE connector on the motherboard) - $20

http://www.newegg.com/Product/Product.asp?Item=N82E16820160010
256MB Compact Flash Card - $20 (after rebates, $12)

http://www.newegg.com/Product/Product.asp?Item=N82E16820136114
128MB PC133 SDRAM DIMM, $15

For Mach2, you will want Windows 2000 Professional, a
512MB DIMM, and a hard disk instead of a CF card
(Windows' swap file eats flash memory due to it's
constant re-writing of it). I might also think about
spending the extra money to get a Celeron or P4 based
Mini-ITX motherboard, instead of a VIA C3 (it adds
another $200 to $250 or so to the price for the
upgraded motherboard, CPU, heatsink, and power supply
- oh, and you will need to pay attention to keeping
the enclosure cool, also; something that you need to
worry about less with the fanless C3 - this is one
more reason I am not a fan of Windows in these types
of installations, as I have mentioned before). I would
avoid XP if possible; it's got quite a bit more junk
added on than 2000 Pro does.

Somewhere around here I've got a PDF with the mounting
hole locations for a Mini-ITX motherboard; I'll have
to look for it (I think it's on one of my external
HDDs, I know it's not on my PowerBook, cuz' I just
wiped the HDD and installed Tiger last week) - just
drill the appropriate holes in the sub-panel of the
electrical box, and you can dispense completely with
the PC-style enclosure...

Hope that helps!
 
Hammerhead,

Thanks a lot for this response. Exactly the type of info I was hoping someone could help me with. A few things..

Well I have to admit to being ignorant of ballscrews, but I would say that replacing the screw & nuts is a great idea, since I believe the leadscrew in my table has uneven wear along its length, so I don't see how I can really adjust it out. I know that the basic screw in the table is nothing special (I had it open a year ago when the autofeed needed some work). But what about the autofeed? I cannot recall the exact mechanism by which it turned the screw, but would I end up gutting this functionality (cnc makes it fairly useless anyway)? I wouldn't really mind doing that, as I could remove the autofeed motor and gearbox (huge). Would I then have a ball-screw cut to match the dimensions of the origional? Does one then simply have to find a way to fix the ball-nut in the saddle? Sorry for my ignorance here..

I study computer science so the computer part of it shouldn't be much trouble. I have piles of old machines lying about.

I am not familiar with servo control (I didn't even know servo's could spin continuously), so I have some reading to do on that subject.

A very cheery thanks from Michigan!
Scott
 
> leadscrew in my table has uneven wear along its length, so I don't see
> how I can really adjust it out.

You can't - it needs to be replaced.

> But what about the autofeed?

Remove it - it's redundant.

> Would I then have a ball-screw cut to match the
> dimensions of the origional? Does one then simply have to find a way to
> fix the ball-nut in the saddle?

Yes.

> Sorry for my ignorance here...

Everybody's got to start somewhere!



There are two reasons for going with ballscrews over acme screws: mechanical efficiency, and backlash. An acme screw is about 30% efficient at transferring torque to linear force; but a ballscrew is about 90% efficient. Also, a ballscrew can be preloaded to take the backlash out of the system without causing excessive wear and stiffness - something that an acme screw cannot really tolerate.
 
Hammerhead
Great info! Thank you! :D I am considering a retrofit for my Hurco SM1. I want a system that is closed loop and what I am finding is that there is no feedback to the program for the actual position of the machine. Is this possible for reasonable cost DIY?
Dale
 
Yes and no - the control program will not know the exact position of the machine, only the commanded position, but the servo controllers will, and will apply corrective action should it get out of sync with the commanded position. Functionally, the result is the same; with the exception that you don't get that little "in-position" flag on the screen at the end of a move.
 
Hmm, I see no reason a decent retrofit control won't have enocoder feedback to the control. I would look for one, the technology is only 25 years old. Most of the controls I am familiar with have tach feedback to the servo drive and encoder to the control.
 
bigIron:
Be very careful when retrofitting a CNC control to a machine with leadscrews. There's always considerable backlash in a leadscrew, where there is little or no backlash in a ballscrew. If the leadscrew is worn in the middle, then you've also got pitch error AND uneven backlash along its length to boot.

Many CNC controls have an optional software feature called "uni-directional positioning", which forces the X, Y, or Z servo to always approach the commanded position from the same direction. This is REQUIRED on a machine with leadscrews if you ever want to drill or bore a hole in the correct location. It does not make a leadscrew machine move like a ballscrew machine, however.

I once worked on a NITEMARE retrofit where some dummy who thought he was clever put linear encoders on a big gantry mill, then connected Fanuc servo motors to the acme leadscrews. The servo motor just occillated, jerking the machine back & forth in a rather alarming kibuki dance. The motor would take up the considerable backlash in the screw, then it would "bump" the slide about .100 inch or so, then react to the positioning error by "bumping" the axis in the other direction. This happened about 10 times per second! I had to tear out all the linear encoders and put in new motors with encoders inside them. That stopped the occillation, but we then had to use uni-directional positioning to make the machine go where we wanted it to go. The customer wanted a full 3-axis CNC mill, and he got a very large hole drilling machine instead.

If you use uni-directional positioning, then you're also going to need "pitch error compensation". This lets you compensate for the wear in the middle of the screws.

Backlash compensation can also be programmed along the length of each leadscrew, but you'll probably find that you can't use it. The AMOUNT of backlash compensation in a leadscrew will cause a servo to "jerk" the slides whenever the servo reverses direction. That makes endmills very unhappy.

You're much better off to fit new ballscrews, or just use uni-directional postioning and pitch error compensation to drill and bore holes. With leadscrews, you can forget about cutting circles and pockets like real CNC machines do.
 
keithg - remember, we are talking about a DIY control built with step & direction based servo controls (Rutex, GeckoDrive), not an analog servo control. There is no tach in the system, the encoder performs that functionality, and it connects to the servo control card.

Check out:
http://www.rutex.com/home/index
http://www.geckodrive.com/ (specifically, the G320 servo drive)

Dan Fritz - you are absolutally right about the large backlash causing oscillation.

bigIron, you should replace the leadscrews on both the x and y axis with ballscrews. Because the knee is pre-loaded, you can theoretically get by without going to a ballscrew system, but for machine efficency, I would do that axis, also (make sure that you put some sort of counterweight on it, though - perhaps some of those high-pressure gas-lift cylenders, commonly used to support the hood and trunk of a car?)

However, what I would have done with that gantry mill (had it been possible with the control in question) is used the encoders on the motor for servo positioning, and the linear encoders (assuming they are of good quality and accuracy) for axial position error correction.
 
hammerhead74000:

You're right that the way to go would be to use the encoders in the motors for positioning and the linear encoders for compensation. Back then (this was in 1980-82), we were working with a "state of the art" Fanuc 6M, which didn't have dual feedback. The Fanuc 9M did have this feature, but we couldn't get one. The 9M could use two feedback devices for each axis, and the position loop gain could be set much lower for the linear device. That might have worked out OK.

This customer already paid an unscrupulous retrofitter about $60,000, who I'm sure never answered his phone again. Naturally, they balked at the idea of fitting new 20-foot long ballscrews. More big $$ for another CNC control was also not an option. He settled for the hole-poker.
 
For 60k he got ocean front property with that right?

Hey I found some closed loop stuff. This is all new to me so I'm looking for all the advice I can get - let me know what you think. We have been talking about this in another thread see last link. Thanks for sharing your experience with us newbies.
dale

http://www.pico-systems.com/univpwm.html

"The advantages of providing encoder information all the way back to the CNC computer is that position can be monitored not only in CNC mode, but when manual movement is made, too. In that case, the CNC computer becomes a true DRO, showing actual position at all times, even when the Gecko drives are off."

http://www.pico-systems.com/gecko.html

http://www.practicalmachinist.com/ubb/ultimatebb.php?ubb=get_topic;f=13;t=000981;p=#000031
 
So it sounds like I need to spend some time looking for software. With the right software it would be a better solution, right? I wonder if Mach 2 would work with the components and just operate open loop. What I'm thinking is retrofit with the potential for closed loop once software is available. Or perhaps a simple program running concurrently as a DRO.

What happens with open loop if you move the table manually? Can you move it manually? I'm thinking like using an edge finder or something. If the table moves the program doesn't know does it?
 
The site daleroe pointed to is Jon Elson's site. He's about as knowledgable as anyone around when it comes to using EMC to control a machine, and it does close the loop in the computer or on the motion control board.

But, unless you're a computer geek or a glutton for punishment, getting EMC up and running isn't a walk in the park.
 
If you want true closed loop control which is PC based and allows the use of existing drives and motors, there is Camsoft and MotionMax from Parker. Both use dedicated motion cards and are capable of High speed maching, both comparable to full blown systems, another that allows the use of your own amps and motors is Fagor and Heidenhain, all the above provide professional results.
M.
 
>> computer geek

I qualify, and I wouldn't use EMC. It's a major pain to get going.

Systems that use dedicated motion control cards are not really open-architecture, and tend to tie you into one control program. Also, they tend to be much more expensive. Now, if that control does everything you could ever want, I guess that's OK, but just be aware of it.

Moving the table manually can be accomplished with the jogging function - it's kinda' like "fly-by-wire" - you turn the hand-wheel on your MPG or push a key (depending on your exact setup) and the table moves by a specific amount (which you can change) per click (or, hold it down, and it moves continuously).

An MPG can be constructed from an encoder with knob attached, and a couple of switches mounted in a suitable small box, wired to the control.

For what it's worth, I am setting up DiamondBack to use a trackpad or trackball as an MPG, however - they are cheap, and provide 2-axis control, also.
 
For what it's worth, I am setting up DiamondBack to use a trackpad or trackball as an MPG, however - they are cheap, and provide 2-axis control, also.
For me, when setting up a part I really use the handwheel for .click.click. feedback as I am approaching a part, tactile fedback helps alot to count the number of .001" or .0001" I've moved without taking my eyes off of the part. With a trackball, you lose the tactile feedback of individual clicks. However since you seem like a workaholic, you could add (optional)audio feedback for the clicks. ;)

Pete
 
Well, copley drives are 300 bucks a pop, and after that the control only has to put out +- 10v and read the encoders, not a complex concept. I wouldn't use any control that wasn't closed loop on any machine that was big enough to hurt me. Or cost me money.
 








 
Back
Top