PDA

View Full Version : My lathe retrofit



HuFlungDung
03-11-2003, 09:07 PM
As a point of general interest, I wanted to tell you all a little bit about a retrofit I completed last year. This is perhaps a little bit more expensive than the typical hobbyist would discuss, but as far as I am concerned, my machining business is just a "hobby gone wild" anyway :)

10 years ago I bought a Graziano lathe that was already retrofitted with a Bandit 1 controller. This was my very first escapade into cnc.

That controller was old when I bought it and I managed to resusitate it several times, but I finally bit the bullet and decided to redo the control with a PC based CNC called CNC Professional from CamSoftCorp.

This was quite the undertaking for me, as I live in a rural community in north-eastern Saskatchewan (Canada) and there is no technical help available in this community. I am the cnc guru here, I guess :D

Anyways, I got the rewiring done, I reused the existing servo motors, I simply replaced the X axis encoder with a high precision Sony magnescale ( .00005" resolution) to improve the positioning accuracy. The precision required on the X axis is much higher than Z of course, so I thought this would save me some money, just souping up the X axis.

The software I first started out with from CamsoftCorp was actually the cheaper version "CNC-Lite" they call it, but I later found that some of the features that I wanted to have required the more expensive CNC Professional package, so I upgraded.

I purchased a 3 axis 1832 Galil motion controller card and interconnect box from CamSoftcorp. I had an old 300mhz PC laying around which I used for the controller, I later upgraded that to an AMD 550 which is what it currently uses.

Then I got into writing logic for the user interface. Although their software packages ships in a "ready to run" state, it wasn't as safe or as foolproof as I wanted. The great thing about getting this PCbased cnc, is that I could rewrite and modify the graphical user interface to do whatever I wanted. This was a great treat for a tinkerer like me.

So I wrote logic and tested it at the machine off and on for about a year and a half, before I got everything close to perfection. I estimate I spent 500 hours on the project, but the results were worth it. I use the machine to this day, and I am very pleased with it. Best of all, it is fool-proof enough that it makes an excellent trainer lathe for a novice. I've designed every safety method I can think of into the logic, so its almost hard to make a mistake using it.

VooDooMan
10-27-2004, 01:13 PM
Hey man nice to see you posting your early years success............I am also clad to see some Canadians aboard!
I have not posted here much in posted in a long time, but since I just bought a new lathe, it is time to make it CNC capable!
I am over in Canada BC................Any way its a clausing 1500 15 swing 48 centers, LOVE this machine and since business is picking up real good on my end its time to do some serious production.!

Can we maybe sit down and chat some more on yur experiences, I am not new to CNC but I am definetly new to servo systems and such, Dont really have any idea where to start, leave my old lead scrws or change to ball screw, Every thing is money sadly :(


Thank you cya

HuFlungDung
10-27-2004, 03:57 PM
Hi VoodooMan

Yes, it can be a tough decision to decide whether to retrofit. When you are looking at buying things like precision ballscrews, and a Camsoft cnc controller and servos, it can easily cost 10 grand, plus your lathe. It will suck up a few of your working hours, too.

It might be a little more than you'e want to spend, but I'd take a look at the Haas TL1 lathe. If you are more concerned with being able to run the machine, rather than retrofit it, this one's ready to go, and still gives you a nifty measure of "man over machine" control. I had a guy demo one at IMTS and I was fairly intrigued by it. Its not all closed in like a full cnc, but for a guy who doesn't want to program every cut, I think this would be a good entry level machine. If I had some spare change, I'd buy one just to play with it and see how useful it would really pan out to be.

I think some other forumers have bought one of these machines, so I'd like to hear their feedback about it.

VooDooMan
10-27-2004, 05:15 PM
Very good point, but now sadly I am going to have to stick to retrofit, Quality isnt that big of a problem, since production isnt always going to be done, rarely on some plastics will I need CNC, but I love the freature!
The lathe was almost 4 grand and has paid it self 50 times over...........I still dont want to get in over my head with this retrofiting, but I am going to keep it under 1.5 thousand $, I have seen it done successfuly, with some time invested, searching for good deals is what takes up most of the time :)

Perhpas we can step into private chat through email, and I can pick your brains a little bit? What do you say to that offer heheh!

Thank you again!

Piobaire
11-04-2004, 02:36 AM
Hi there!
I'm just in an early stage of rebuilding an 1985 Harrison M400 CNC (8"x60") with a new PC-based controller from Ajax CNC. The old Anilam Crusader IIL have done its job a long time a go, that's why I got it quite cheap. A 4-position tool changer also came in the deal.

First of all, the cross slide table where worn as lack of lip seals. I'll make a new table and also increase the ball screw height in the slide to fit in a standard sized ball nut and better bearings. The servo is originally fitted in front resulting in that the operator is placed too far from the machine. I'll also change the x-axis servo to fit in the opposite side to earn space.

The servo for the z-axis is a bit in the way for the slide so the travel length in z-axis becomes a bit limited. Hopefully I'll come up with a nice idea to move that to, avoiding the two servos to collide.

Then I plan to build a new controller box a bit like the new Harrison Alpha series that could slide on the front with two hand wheels to run the machine in manual mode. I already got a 15Hp frequency inverter to control spindle speed.

Send rough pictures of "before" and "after" very fast sketched in SolidWorks.

Is there anyone with experiances regarding DC servos, DC servo drives, Ajax CNC, rotary encoder placement, lathe design for best performance and accuracy etc. pleas feel free to email me.

Best regards
/Thomas (Sweden)

VooDooMan
11-04-2004, 10:52 AM
any spare equipment you can let go from your old CNC, any thing working at least?


Thank you

r1656
05-01-2006, 03:14 PM
As a point of general interest, I wanted to tell you all a little bit about a retrofit I completed last year. This is perhaps a little bit more expensive than the typical hobbyist would discuss, but as far as I am concerned, my machining business is just a "hobby gone wild" anyway :)

10 years ago I bought a Graziano lathe that was already retrofitted with a Bandit 1 controller. This was my very first escapade into cnc.

That controller was old when I bought it and I managed to resusitate it several times, but I finally bit the bullet and decided to redo the control with a PC based CNC called CNC Professional from CamSoftCorp.

This was quite the undertaking for me, as I live in a rural community in north-eastern Saskatchewan (Canada) and there is no technical help available in this community. I am the cnc guru here, I guess :D

Anyways, I got the rewiring done, I reused the existing servo motors, I simply replaced the X axis encoder with a high precision Sony magnescale ( .00005" resolution) to improve the positioning accuracy. The precision required on the X axis is much higher than Z of course, so I thought this would save me some money, just souping up the X axis.

The software I first started out with from CamsoftCorp was actually the cheaper version "CNC-Lite" they call it, but I later found that some of the features that I wanted to have required the more expensive CNC Professional package, so I upgraded.

I purchased a 3 axis 1832 Galil motion controller card and interconnect box from CamSoftcorp. I had an old 300mhz PC laying around which I used for the controller, I later upgraded that to an AMD 550 which is what it currently uses.

Then I got into writing logic for the user interface. Although their software packages ships in a "ready to run" state, it wasn't as safe or as foolproof as I wanted. The great thing about getting this PCbased cnc, is that I could rewrite and modify the graphical user interface to do whatever I wanted. This was a great treat for a tinkerer like me.

So I wrote logic and tested it at the machine off and on for about a year and a half, before I got everything close to perfection. I estimate I spent 500 hours on the project, but the results were worth it. I use the machine to this day, and I am very pleased with it. Best of all, it is fool-proof enough that it makes an excellent trainer lathe for a novice. I've designed every safety method I can think of into the logic, so its almost hard to make a mistake using it.
Hi
Im interested in the parts you used to retrofit your sag 12 and as many photos as you can send so I can see what it is you did to accomplish this feat. Russ Please send to r1656@aol.com I sure appreciate it and it would make my project much simpler looking at what someone else did.

HuFlungDung
05-01-2006, 06:04 PM
Russ,
I don't have detailed pictures taken during the retrofit stage. Its all together now, and it is difficult to make out too much.
My lathe actually was a re-retrofit, so indeed, some previous owner had already gutted the quickchange geabox, apron, and installed the ballscrews and servomotors. So I started from that point, replacing the control was my main headache.

I'll take what pictures I can. A lot of the good stuff is buried under covers and such.... :)

Al_The_Man
05-01-2006, 07:22 PM
Hu, Did you ever get the Threading routine down pat? Was it using native Galil commands?
Al.

Geof
05-01-2006, 07:32 PM
......I think some other forumers have bought one of these machines, so I'd like to hear their feedback about it.

Could not be happier. I bought one to use for developing new parts and making tooling. Now I am reluctant to go to any of the manual machine even just for facing something. I have been impressed with the rigidity of the TL and have uploaded two pictures showing before and after shots of a shaft about 8 inches long done without a center or any support. The only minor complaint I have is that the rapids and tool changer are very slow so it is not really suitable for doing much by way of production involving all four tools.

HuFlungDung
05-01-2006, 07:55 PM
Hu, Did you ever get the Threading routine down pat? Was it using native Galil commands?
Al.

Al,
Nothing has changed in that department, since I'm all out of ideas to make it any better :D I do still run the threading via a Galil program directly on the card. It never mucks up tracking the original thread groove, no problem there, but sometimes I wonder if the thread tool is ever going to move, then suddenly the machine will see the encoder pulse and it will seem to take several passes just as quick as can be, with little delay.

I'm still thinking that it just a matter of synchronization, and the Galil card is just not looking for the input continuously, but is rather sampling the input. So in the short timeframe allowed, with the encoder running at quite high speed, it misses the right interval to take the sample and see the signal.

My reason for my theory, is that I am running the Galil card at nearly as fast an update rate as is allowed, plus I have the spindle encoder running at 1/4 of the spindle speed. This seemed to be the best fit for this hardware, although an even greater reduction in encoder speed might be beneficial. But, I thread typically at 800 rpm, so the encoder is only turning at 200 rpm now.

CATCH22
10-02-2006, 08:18 PM
HU im curious i just bought a sag12 and the transmission cluthes are slipping. do you know if you are still using the trans or are you just using the motor with a vfd. thanks catch.

HuFlungDung
10-02-2006, 10:54 PM
Catch,

I am still using the transmission, never did overhaul it. I figured when it got to that point, I would put the VFD on it.

There are a bunch of guys over on the practical machinist forum that have repaired graziano lathes. There might be some pointers there that will help you figure out how to fix it on your own, or perhaps to find parts, rewind coils, etc.

If you have 'heavy cutting' to do, it would be nice to have an option for at least a couple of gear changes along with the VFD. Maybe you can bypass the slipping gears and make do with the good ones, along with the VFD.

Adobe Machine
10-02-2006, 11:22 PM
Hu: Just a quick question: I'm converting a manual Pratt and Whitney Tool Room lathe to CNC and have decided to retain the full gear box ( it has a total of 32 speeds with the two speed AC motor ) instead of going the VFD route.I'm mounting an encoder to the spindle which will referance true spindle speed.Have you had any issues using the transmission in yours as opposed to going to a VFD ? My threading program relies on the encoder to act as a "Tach" only, coordinating X and Z as related to spindle speed. I have run a hand held , non contact tachometer to the chuck and can see little, if any rpm variation. Using the same tachometer on my retrofitted mill with VFD, I do see some small variations.
Thanks for your time

Adobe (old as dirt )

HuFlungDung
10-02-2006, 11:41 PM
Adobe,
If you are running a Camsoft retro, you should add a bit of logic into the timer.fil and take actual samples of your spindle speed as calculated from the encoder feedback. Because we are not using a VFD, we cannot force an accurate spindle speed to be reached, but we do need to know what it is.

The logic used within your threading gcode then needs to autocalculate the feedrate (in IPM) based on the actual pitch of the thread to be cut. This simple math is easy to include within the logic of your desired gcode.

For a complete threading cycle, I took one sample of the spindle speed and then forced the feedrate calculation to be saved and used for the duration of that threading cycle. This is because I did not want the feedrate to change because of the slight variation that could occur from a new spindle speed sample.

I have no issues with the tool tracking the original path as a result of taking these precautions.

Adobe Machine
10-02-2006, 11:53 PM
Hu: Thanks..so you have had no problems with late X axis pull out and breaking the threading tool on a shoulder..I see what you mean by "forcing" one sample as opposed to the controller trying to compensate for small variations in spindle speed..
Again thanks for the info

Adobe ( old as dirt )

HuFlungDung
10-03-2006, 07:17 PM
I can see that the X axis retraction could be a problem. The way I wrote my threading cycle, I included a radial retraction of the tool at the end of the thread. I do not call seperate G02 at the end of the thread because that could involve a definite hesitation and broken tool if you ran in single step.

Now because I wrote the threading routine in Galil commands, I was able to predefine the threading motion and the radial retraction as one smooth vector sequence. That means, set up the path first, then wait for the start signal (encoder triggered) and then the sequence executes as one fluid motion. You get the same type of action when you call for the Camsoft SMOOTH ON function. However, you must turn SMOOTH OFF after each such sequence and before the final move in the sequence so that is a bit cumbersome. Better to do it right in the Galil unless you can figure a way around any problems.

Here is an exerpt from my Galil program demoing how the cut is done. You will notice that the program is filled up with many 6 or 7 character variable names. You can pass information from your Camsoft interface to the Galil to feed these variables. So you can assume that mathematics is carried out elsewhere to get the missing values that you do not see here. I don't know whether that helps or hinders :D

REM Next will be the tool movement to the start position of the cut.
SP 37500,37500
PR BEGINZ,BEGINX REM start point for the cut
BG XY
AM XY
REM Next 5 lines define the threading movement including any radius pullout.
VM XY
VS VSPEED REM calculation is carried out for a combined axis speed so that taper threading will be on pitch
VP ZLENGTH,XHEIGHT REM this is the endpoint for this cut, before the radius retraction, next
JP #SKIPRAD,(@ABS[RADIUS]<10)
CR RADIUS,ARCBEGN,ARCROT
#SKIPRAD
VE
REM NOSTOP is a flag so that the threading pass cannot be interrupted in mid-cut.
REM However, this program will go into feedhold right after the threading pass.
NOSTOP=1
REM The next input signal 3 comes from the spindle encoder index pulse
AI 3
BGS REM this is when the cut happens, the Begin Sequence command
AM XY

Adobe Machine
10-03-2006, 08:57 PM
Thanks for the information. In fact, have the capability to "lock in" the rpm, so that small variations will not affect X and Z..X pull out is G00, calculated at 80 IPM. This has to be faster than I can manually do an X pull out , which is the only time I've broken an expensive thread profile cutter when I was a little slow bringing X out .
( lie, I dropped the whole cutter on the concrete not too long ago, the cutter end hit first, crappy luck)

Adobe (old as dirt )

Adobe Machine
10-04-2006, 11:54 PM
Wow..I can see why you are sucessful..What concentration and inguinity..I really compliment your methods and procedures. This is just gonna take me a few days to digest and try and implemint.

Hats off to Hu..

Adobe ( old as dirt )

weblogictech
10-15-2006, 04:42 PM
I own a mazak quickturn 10 with Mazatrol T1 controller. I am looking for an option to retrofit the same with PC Based maybe Galil Motion Card and Camsoft Professional Software.

I appreciate feedback from users who have done this in past, is it complicated or there are problems with it or is prop. controller like Mits, Fanuc, Siemens etc is better option for Lathes.

Regards,

Rizwan

HuFlungDung
10-15-2006, 05:27 PM
Rizwan,
I won't pull any punches, one of the standard proprietory controllers is likely to be more satisfactory right off the bat, because you will get a professional retrofitter in to do the job, and he most likely will know what is going on, and what needs to be done.

There is nothing stopping a professional from using Camsoft as well, but the interface will be a unique Camsoft interface, and the standard gcodes and threading and turning cycles may not work just like you would expect them to. If you have other people who need training to run the lathe, then you might be better off with standardized factory training materials, such as you would get with one of the proprietory controllers.

The strength of Camsoft, in my opinion, is that you can implement some nifty ideas that you will never find in any other controller. You can debug, and alter what you don't like. You can service the machine with relatively inexpensive components, from henceforth, without paying a king's ransom for parts.

But the Camsoft thing is more for those who really want to 'marry themselves' to a machine, or have a special machine that requires custom software anyways. Even having done a Camsoft retro myself, I'd be pretty askance at buying a machine that someone else had retrofitted with Camsoft, because I'd have to wade through their logic to find and fix any issues that I discovered. Even my own still has a few loose ends in it, things that I have partially implemented, or that need outright changing, and stuff like that is very tough to resell unless someone is going to go through and document how you set your controller up. That would take a hundred hours or so (at least) to write up how it all works.

Al_The_Man
10-15-2006, 05:34 PM
I know we have covered this before, but hopefully if I get the time in the near future I intend to experiment/work on this threading issues with Gallil using native code.
Is the problem with using Camsoft the fact that commands are issued one at a time?
For example, my experience with Galil, is that is is very powerful if a complete program or block of code is downloaded and run, which I believe can simulate the threading as is done on Mitsubishi, Fanuc etc, where essentially the z axis is synced to the marker pulse and then geared to the spindle encoder, so it allows for some variation in spindle speed and the z axis will compensate feed rate accordingly.
Al.

HuFlungDung
10-15-2006, 06:05 PM
Al,
You are correct in that the Galil is very powerful, and can execute a final part profile chain with excellent smoothness. The lathe is maybe not the prime example of a machine where this quality is needed, but I have actually witnessed toolmarks on a turned part (at the end of each segment), containing about 4 different 'counter-tapers' to try to turn a long, slender surface straight in one pass. As soon as I turn on the Camsoft "SMOOTH" ON, those same marks completely disappear from the surface. The manner in which the SMOOTH command works, appears to me to be the equivalent of feeding a sequence of vectors into the Galil onboard memory, then executing that sequence. It works very well, and I can imagine that it would be really nice on a mill doing 3d contouring.

But, if there remains any issue with lathe threading, it is not the manner in which Camsoft does it, it is the matter of getting the encoder index signal to do two things:
1. Start the motion reliably and repeatably after the encoder index is detected.
2. Detect the encoder index within one revolution.

Now, the first one I dealt with, trying to find reliable triggering methods from the Galil card. There are a number of ways that you can use Galil's commands to trigger motion, but all of them (except one) have a variable length of time delay between reading the index, and initiating the motion. On a lathe running at a few hundred rpm, there is really no room whatsoever for any 'window width' about when the Z has to start. Its got to start exactly the same after every index, or the tool will not pick up the same path as the previous cut made. I am not saying that it even must be 'super fast' at starting the Z motion, but it must be 'super repeatable' on whatever time delay is built into the hardware, firmware and whatever.

The second issue, I've heard that guys have built detection devices using 'off-board' high speed I/O to pick up the encoder index. Although I have not added a high speed I/O board into my computer, it does sound like a reasonable explanation for a lot of waiting around for the Galil to detect the encoder, and the faster the encoder is turning, the worse the detection rate using Galil's onboard I/O.

Al_The_Man
10-15-2006, 10:48 PM
I was considering using an input with the AL = Arm Latch, this stores the position of the axis within 1&#181;s of going low, so I assume it is a fast interrupt.
The status of the latch can be read immediately in order to trigger the Z to gear to the spindle encoder.
Al.

HuFlungDung
10-15-2006, 11:49 PM
Al, should I bias your thinking? :D

It's been a few years since I went through all that testing, and I think that was one of the tested methods. But you try it and see what you think. The acid test is running a single pass on the lathe, and then watching the tool rerun exactly the same path (no additional cutting depth) and watch whether or not the tool ever takes another chip on subsequent passes.

I don't have a clue what the Galil is doing that causes a variable time delay with some of the methods I tried. The speed of response to the high speed latch was not the answer to a consistent restart that was in sync with the encoder position. However, using the AI command to wait for the encoder index seemed to be very consistent and that is what I ended up using.

sanjiv
10-16-2006, 11:25 AM
have u listen that some one had made edm machine at home? can u find him

jeep534
10-19-2006, 01:53 AM
HU im curious i just bought a sag12 and the transmission cluthes are slipping. do you know if you are still using the trans or are you just using the motor with a vfd. thanks catch.

Some-one is selling rebuilt electric clutches. I just don't remember where I saw it. Maybe on The P.M.board
archie =) =) =)