Converting a Boxford 250B

More
11 Sep 2013 02:31 - 11 Sep 2013 02:33 #38643 by Einar
A few years ago I got hold of a fully functional Okuma LC20 twin turret lathe for free. After it was placed on my yard I discovered it needed 240V/125A. I only have 64A, bummer! It went for scrap this summer, all 6,5 tons of it.

then I just bought a Boxford 250B CNC lathe born Feb. 2. 1995. Just the right size and only needing 32A.
It is in full running order. So I figure I just have to do something about that. :P

The user interface is just terrible! So thus a retrofit is needed.

It has steppers on X, Z and the 8 position tool turret. The drivers are old, but standard with step & dir inputs.
There are some kind of home switches (sensors?) that I have not figured out yet. And an adjustable limit switch to avoid chuck crash.

The spindle drive is AC ansync motor driven by a IMO Jaguar VFD and voltage input speed control. The drive does have an RS485 port but Google cannot find a manual for it, so forget that. It also have a multi hole optical reader and an optical index reader.

There are a coolant pump and some safety switches too.

I tried downloading "linuxcnc.org/docs/LinuxCNC_Integrator_Manual.pdf", but that page is still blank after a couple of hours.
And several URL's under "Lathes" here "wiki.linuxcnc.org/cgi-bin/wiki.pl?Case_Studies" are dead.
So I'll poke around and ask here as I go.

Q1: I already have a USC Sn:0029 from Jon Elson. I understand this is an old version not supporting threading. But can it be used on a lathe not threadcutting?

There seems to be several options regarding interface boards. But as I remember Jon's customer support to be the best back when I used that board with my mill, that will be my preferred choice. I have to get a new one for this conversion, just want to know if I can start out with the one I have.

First task is to document the wiring as it is. Maybe also bring everything out to a terminal block to make life easier in the future.

The tool turret runs forward only. Then backwards until hitting a pawl (?) where the stepper stalls out. I assume it also have some means of reading the current position as I do not need to home it. For controlling it I guess the easiest option is to hook up a small PLC.
Q2: Can LinuxCNC output the tool change command on a serial port (preferrably) and halt until receiving acknowledge or fault from a PLC?
I see there are possibilities of a built in PLC in LinuxCNC, but I have several PLC's on the shelf. And I already know how to make them do what I want. ;) I can't say that about the LinuxCNC ladder program. Ladder is not my language of choice anyway.

My camera is with me in this. I'll hang up some pictures on my website and link them in. I'm sure you hate pictures, but anyway. :whistle:

Ahh! It seems I can add images right in here, so I'll annoy you immediately with some pictures from the seller.



Attachments:
Last edit: 11 Sep 2013 02:33 by Einar. Reason: Linking pictures

Please Log in or Create an account to join the conversation.

More
11 Sep 2013 05:32 #38649 by andypugh
Replied by andypugh on topic Converting a Boxford 250B
You might find this useful for driving the tool changer:

wiki.linuxcnc.org/cgi-bin/wiki.pl?Contri...oolchanger_component

Please Log in or Create an account to join the conversation.

More
11 Sep 2013 14:17 - 11 Sep 2013 14:18 #38655 by Einar
Replied by Einar on topic Converting a Boxford 250B
Yes it does. Thank you!
Yesterday evening I found out it probably does not know where the tool is. Just if it needs locking.
And I found only 1 homing switch, plus an inductive sensor at the end of the ballscrews. It obviously use these in the homing sequence.
These are just indexes so it seems this one switch is able to give a position with better than one pitch accuracy for both axes.
How it can do that is beyond me, but I'll just treat is as a fact. My aim is that this be an electronic only upgrade.
If the mechanical and electrical parts can be used without changes, I will.

Next I will hook up an analyzer to figure out the PW, max frequency and max accelerations used for the steppers.
Those will come in handy as a baseline when tuning LinuxCNC.
Last edit: 11 Sep 2013 14:18 by Einar.

Please Log in or Create an account to join the conversation.

More
11 Sep 2013 14:44 #38656 by ArcEye
Replied by ArcEye on topic Converting a Boxford 250B
HI

I have a Boxford 240F (originally a Fanuc control), for which I wrote the toolchanger component Andy pointed you to.

Mine had had a voltage outrage of some sort, a lot of the low voltage sensors etc were fried

If you use the toolchanger component, you don't need a tool index. At start up it prompts for the current tool number and there after moves are based solely upon the distances between the tool positions with a lock back move against the pawl.

If you wanted to use a tool index, say for number 1 tool, there are 2 other toolchanger components I wrote in the same wiki, which do just that and could be adapted depending upon how you do it.

The inductive sensor is under the chromed plate on the Z axis and matches a void to register, it may also match one on the X axis which crosses it, I assume the axes homed seperately with the original software.
Since mine was blown I have retrofitted physical index / home switch pairs on each axis.

I have repaced the DC servos fitted to mine (and sold them financing much of the retrofit), with large steppers. With my computer running 9K max heavily loaded base thread jitter, I can set the base thread to 30K and get 3600mm/min rapids with 100mm/s/s acceleration without step loss
(I have them set to 2400 for everyday use, the other figure was achieved in testing to push the limits and see what could be done)
This however is using 85V leadshine drivers and a 76V PSU on modern steppers, I don't know how good the original steppers and the Parker drivers were

If I can assist, let me know

regards

Please Log in or Create an account to join the conversation.

More
11 Sep 2013 14:57 - 11 Sep 2013 14:58 #38658 by Einar
Replied by Einar on topic Converting a Boxford 250B

The inductive sensor is under the chromed plate on the Z axis and matches a void to register, it may also match one on the X axis which crosses it, I assume the axes homed seperately with the original software.

Yes, I found that sensor using one of those "camera on a snake" thingies. And I was assuming it is like you say. Thus the inability of the original controller to home in Z only. It must home in X first, then Z. The thing I don't understand is how does it do it? There is quite some distance from this switch to the cross slide. There may be a plunger inside it or something. Anyway as long as LinuxCNC can be configured to use it, that's all I need. :cheer:
Last edit: 11 Sep 2013 14:58 by Einar.

Please Log in or Create an account to join the conversation.

More
16 Sep 2013 05:25 - 16 Sep 2013 05:29 #38869 by Einar
Replied by Einar on topic Converting a Boxford 250B
More measurements made things a bit clearer.

As far as I can find the machine have an inductive limit switch for Z that will sense both ends of travel. Homing is done as a combination of this and another inductive switch sensing a disc on the end of the ballscrew.

The X axis is the same, except the sensor on the slide is only activating on the home end of travel. This could be made to activate on both ends by milling a pocket in the X slide, but I don't want to do that. At least not until I find that necessary.

Then there is an additional limit switch that is activated by an adjustable rod on the Z slide.

The spindle sensors are one index sensing a slot in the disc on the spindle. The other senses 50 holes in the disc.

Then there are switches sensing any open cover or doors to the electric innards. Which I have disabled now.

The VFD is controlled by a direction signal, a run signal and a 0-9V signal controlling the speed up to 175Hz = 3000RPM on the spindle.have a

The step signals start at 500Hz and increase to 6100Hz in 200ms. There is a bit of jitter at the higher frequency, so with a signal that have less jitter it may even go higher.

The Z axis have a 2.5mm pitch ballscrew and 20:25 belt drive from the stepper. On X it uses a 5mm screw, also with 20:25 ratio.

Next up is to remove the PC that run the Boxford control program and install the terminal block.
LinuxCNC:
Installed on a PC that was a server in our office before it followed me home.
Installing from the CD went without problems.
I hooked up the USC board and ran the diagnostic program. No board found!
Then checked the BIOS settings and enabled EPP/ECP. Reboot, and this time the board was found.
Running the diagnostic again with diocontinuous parameter made the LEDs blink and no errors reported. :cheer: :laugh: :silly:

So this makes it possible to continue without waiting for a new USC. But will it be possible to do screwcutting?

It is a bit unclear to me if the new EEPROM is for using it with some new PC's, or if it also is necessary to do screwcutting.

Anyway, when the lathe is done, I will use this board for my mill that is currently running Mach3 through parport only.
I will go to LinuxCNC on this too. It did run EMC many years ago. At that time I found it too difficult to use. Now it seems LinuxCNC is much better.

The control panel is on the right side of the machine. Not very convenient if one wants to see what's going on.
I have a nice jog wheel rescued from the Okuma we scrapped. So I want to make a pendant with this and some other useful functions.
At work I design a handheld device with Modbus RS485, a small graphics display and some buttons. All I have to do on that end is to add more buttons, the handwheel and some lines of code. So I'm looking for how to connect into LinuxCNC using Modbus.
_______

The spindle reader:


X-axis inductive sensor:


Z-axis inductive sensor:
Attachments:
Last edit: 16 Sep 2013 05:29 by Einar.

Please Log in or Create an account to join the conversation.

More
16 Sep 2013 18:46 #38880 by andypugh
Replied by andypugh on topic Converting a Boxford 250B

As far as I can find the machine have an inductive limit switch for Z that will sense both ends of travel. Homing is done as a combination of this and another inductive switch sensing a disc on the end of the ballscrew.

That should allow for very accurate homing. Depending on homing speeds, axis acelleration and how long the sensing disc is active for this might be easy to configure. If the axis can stop from homing speed inside the distance that both the bed-sensor and the the disc-sensor are active then you can just and2 the two switches together. If there is a danger of the axis passing out of the other side of the active region then you might need to get a bit clever, maybe with HAL components to make a latch (disc-sensor signal can only go low when bed-sensor is low, homing is a search, back-off find slowly).

Then there is an additional limit switch that is activated by an adjustable rod on the Z slide.

Possibly to protect against hitting the tail-stock?

The spindle sensors are one index sensing a slot in the disc on the spindle. The other senses 50 holes in the disc.

That will work nicely for threading, but as it can't sense direction it won't support rigid-tapping. This may not be anything you ever miss, though. You will need to configure the encoder counter for single-channel + index, which I don't have enough experience with the USC hardware to know how to do. You could consider adding a 3rd opto-sensor or possibly swapping the existing hole-detecting sensor for a two-channel style. uk.rs-online.com/web/p/slotted-optical-switches/1711830/

The VFD is controlled by a direction signal, a run signal and a 0-9V signal controlling the speed up to 175Hz = 3000RPM on the spindle.

There are a number of ways to handle this. The simplest might involve reconfiguring the VFD to give full speed for 5V input, as this is often configurable in the VFD.

The step signals start at 500Hz and increase to 6100Hz in 200ms. There is a bit of jitter at the higher frequency, so with a signal that have less jitter it may even go higher.

The USC should be able to do a lot better there.

Please Log in or Create an account to join the conversation.

More
16 Sep 2013 22:15 - 16 Sep 2013 22:16 #38884 by Einar
Replied by Einar on topic Converting a Boxford 250B
Thanks for your comments Andy.

Yes, I think it will be a pass both sensors, back up past the disc sensor, then slowly approach it again.
The slide sensor has a bit hysteresis, but on the disc sensor it is negligible.

The adjustable rod is on the other end. It can be adjusted to avoid tools running into the chuck. Well sort of, it does not take into account that boring tools will stick out further than surfacing tools. This is a machine mainly targeted at schools, so that may be the reason it's there. Hitting the tailstock is not avoided in any way. Except that the step motors will stall before any damage is done.

I can easily add another sensor to the spindle to get A & B signals. But rigid tapping is probably not possible. The spindle motor is an AC Async motor. The slowest speed is about 300RPM at the spindle. Nominal speed (at 50Hz) is appx. 850 RPM at the spindle. So torque at tapping speed will be very low. Even if this is a 2.2KW motor.

I have no way of reconfiguring the IMO Jaguar VFD, as I don't have the manual for it. I will try to contact IMO to see if I can have a copy.
Or wire up an amplifier to get the voltage I need. ..... Just checked, and the DAC for USC can output 0-10V, so that will be fine I guess. I put it on the order I',m preparing for Pico Systems.

The jitter number I got for my PC seemed huge! It got better after I put in a video card instead of the built in video. But as I understand it this is less important with the USC board? I need to find another video board, as this one is somewhat jumbled on the top and bottom. If it does not work out, I can use another PC. Although it seems overkill to use a really fast modern PC with a gaming video card for this I think.
Last edit: 16 Sep 2013 22:16 by Einar.

Please Log in or Create an account to join the conversation.

More
16 Sep 2013 22:35 #38887 by andypugh
Replied by andypugh on topic Converting a Boxford 250B

Yes, I think it will be a pass both sensors, back up past the disc sensor, then slowly approach it again.
The slide sensor has a bit hysteresis, but on the disc sensor it is negligible.

This may require a bit of HAL logic, but is otherwise straightforward.

I have no way of reconfiguring the IMO Jaguar VFD, as I don't have the manual for it. I will try to contact IMO to see if I can have a copy.

It is probably online. www.drivesdirect.co.uk/Downloads/IMO%20CUB%20Manual.pdf for example.

Although it seems overkill to use a really fast modern PC with a gaming video card for this I think.

Big fast modern PCs often give very poor jitter numbers.
The favourite for a long time (though now hard to find) was the Intel D525MW. It has low jitter, a parallel port, and is very cheap.

Please Log in or Create an account to join the conversation.

More
17 Sep 2013 15:36 #38928 by Einar
Replied by Einar on topic Converting a Boxford 250B
Thanks to you and Cradek (Cris?) for helping me yesterday. I really appreciate it! Even if we did not get it sorted it feels good not just fumbling in the dark. And I did learn a lot, although I felt like I was asked to type commands in chinese. :lol:

My Jaguar is older than those and other IMO drives I found. It also seems mine is made by IMO, while the newer are relabelled Fuji drives or some other brand.
Which makes me think the parameter set will not be the same. And not having the manual I don't want to poke around and mess up a working VFD. A >5KVA drive is not cheap, even today.

Interesting what you write about the PC. What I have is not top modern. It was 3-4 years ago. I will install the CD on another PC, probably today. It just sits there gathering dust. And I can make errors much faster using 2 PC's than one. :silly: No, it's more just in case I did something silly when installing. Not knowing what I do I might be more lucky next time around.

Please Log in or Create an account to join the conversation.

Moderators: piasdom
Time to create page: 0.147 seconds
Powered by Kunena Forum