Light Machine Corp. Benchman XTr (retrofit)

There was an issue with the upper endstop in this machine, so I had to pull the head off. You may be able to see what you need without removing the head. Home the Z- axis. Remove the accordion dust cover, there are 4 allen head bolts on each side. That should expose what you need to see. If you cannot see what you need, then remove the head by removing the 4 - 3/8" allen heads from the headstock. I need to tram the head still, perhaps I will use duel touch probes.

As for the head falling, you may need some PID tuning or increase your allowable error. At this point, I have the zbrake turn off with the axes are enabled.

Steve, which spindle is on your machine?

andypugh wrote: Mesa offer a board that works _very_ nicely with resolvers.

Mesa 7i49

I like it so much that on my recent Lathe retrofit I removed an encoder and fitted a resolver

The 7i49 produces a very high quality velocity output, and it would be relatively easy to route that back to one of the analogue outputs on the board to get a velocity-proportional voltage.

Wow! Underestimated Mesa there. And 6 channels too. I stand (sit) corrected. My previous experience with Tamagawa was a PITA and this sounds like a slick solution.

This week has been pretty hectic, but today I am trying to set a plan to get the tool changer working. I will layout a description of the system, hoping it will aid any discussion on the programming aspect. Then, I will layout my concept of the process needed to make a complete tool change process. When something looks incorrect, please let me know! After conceptualizing the process, I will try to layout the files and components to make it all work!

Here is a quick video of the tool change process from utoob.
Benchman XT tool change video

The tool changer is a 20 position carousel with four pneumatic actuators and a unidirectional 24v DC motor. The four actuators are controlled by SMW valves that are controlled by the 7i77 outputs. The actuators control: (1) carriage up/down, (2) carriage left/right, (3) tool holder up/down, and (4) power drawbar open/close. Feedback for the first three actuators are endstops on each limit of the actuator. I think there is feedback on the power drawbar, but its has been so long since I mapped out the connections that it will need to be verified. The DC motor is connected to a relay, which is also controlled by the 7i77 output. Feedback on the rotation of the carousel is produced with a shaft encoder. Additional feedback in the system is provided by a hall effect sensor on the spindle shaft for the tool change orientation and a PSI sensor for the compressed air.

Here is my concept for the toolchange routine. Let me know
1 Call for tool change - PIN?
2 Stop spindle - M05
3 Index Spindle Orientation - M19
4 Move tool head to change position - Z = position to be determined
5 PSI Check = TRUE/FALSE
6 Rotate to current tool pocket position if needed - carousel component (need direction)
7 Lower Carriage - C U/D output = TRUE, CD Endstop input = TRUE
8 Tool Holder up - TH U/D = TRUE, TH U Endstop input = TRUE
9 Carriage Slide Right - C L/R output = TRUE, C R Endstop input = TRUE
10 open draw bar - DB O/C output = TRUE, feedback?
11 Tool holder down - TH U/D output = FALSE, TH U Endstop input = TRUE
12 rotate to new tool pocket position (again help)
13 tool holder up - TH U/D output = TRUE, TH U endstop = TRUE
14 close draw bar - DB O/C output = FALSE, feedback?
15 carriage slide left - C L/R output = FALSE, C L Endstop input = TRUE
16 tool holder down - TH U/D output = FALSE, TH D Endstop = TRUE

From what I have seen in JT's turret sim, I understand that classic ladder will only deal with pins, not encoders directly. This is why the carousel component must be used, to let CL know when the carousel is in the right tool position. After reading through the description of the carousel component, I am trying to understand it better. Andy, can you provide an example of code where the component is used? The encoder is a quadrature encoder, but not sure which encoder type to specify (Grey, binary, index or single). Can anyone help me out with the definition of each? Also, I think the num_sense corresponds other feedback on the carousel shaft, but not sure. I dont have any other feedback than the encoder.

Any flawed thinking, or a point in the right direction?
Thanks,
Aaron

steve_a wrote: Before I tear into the Z axis, does anyone know how to adjust the Z axis on a Benchman? My unit is so loose that when the brake is deactivated the head just travels down on its own. This is a problem because the servo spikes on a downward move at about .006" error for a very short time before it corrects. I'm thinking I need to either counterbalance the Z axis or tighten the gibs to reduce the amount of "shock" from gravity. Any input would be appreciate.

I am no expert by any means, but I think this can be resolved by some PID tuning....

It is normal for the head to fall with the brake deactivated and no servo control, maybe best to check backlash in the ball screw/nut assembly or just backlash in general in the machine head

I have the 5K RPM spindle. My machine is also a 20 tool changer unit. I think I am in the ball park for what I need to do with this unit for the moment. I still need to look at the gibs but that can wait until I have the spindle working. I THINK I have found the adjustment allens for this unit but I'll look at that later. As a point of interest, I DID find a manual for the preceding model of this mill and they do have a procedure for adjusting the Z axis. It doesn't help here except to indicate that the Z axis was considered the only axis that might need an adjustment and so I am thinking this unit is probably the same.
Pretty excited for the moment. In my previous posts I mentioned that I was going to change out the original 72V supply because the control and smoothing capacitors are on the main control board which I had to replace. I believe Macgalempsy managed to figure out a way to keep his. More power to MAC! Still with a Dremel tool and a tapping set I managed to install a new drive motor power supply. I've ordered stand offs for my controller board and some connectors for the spindle encoder. Hoping to have the spindle running in a couple of days. That would give me a working mill. Thanks again to Macgalempsy for his diagram of the J2 connector. I was able to verify it and that information will make the wiring so much easier. Parts should be in shortly and the rest "should" be straight forward. I hope!
Thanks to all for the great info and feedback!

I believe you should work on the PID tuning because I had the same problem and it got cleared up. The tuning is everything, it took a while for me to get it figured out. Now my motors are probably a little over-tuned because the hum just a hair, once in a while. If I recall correctly, the mass slides on long linear bearings and do not recall seeing a tensioner.

This week has been a huge milestone in the BenchmanXTr project. The back was put back on the controller box!!!

https://www.flickr.com/photos/58181938@N03/28774564032/in/dateposted-public/

Since the last time I checked in a lot has happened. After having the 3rd computer die last week, it was time to buy a new rig. On the advice from JT, I opted for a Gigabite GA-J1900-D3V motherboard, 8gb ram, a mini itx tower, a small laptop HDD and a DVD player. The HDD and DVD were used from my endless pile of jonk. A hickup occurred when it became apparent that the PSU in the case only had one SATA power connection. Fortunately, the local radioshack had an adapter from ide to sata power jacks. A solid test rig for under $250, not bad. Within the first few hours of running LinuxCNC on the new computer, it started getting watchdog timeouts... I thought the power savings settings had been disabled, but one setting was not so straight forward. If you buy this board, make sure you turn off the C state of the PPM settings in the bios. The latency of the machine dropped by 1/3 from 33k to 11k, and bye bye biting...OUCH!

Also this week, several processes have been completed.

The Renishaw TP2 touchprobe is now doing active touchoffs on the X, Y, Z through tabs on the Axis panel. I had tried using a glade version that Learning LinuxCNC had used on his youtube channel, but for some reason I could not get it to work. Now I am using the one from mesa.net and if I could get the tip size to permently change, that would be exciting. The whole assembly is pretty long, it is working. A tip for first time probers. Make sure your probe signal is connected to all pins before you attempt a test. If it is not, you will break a tip...ask me how fast $20 shatters!

The spindle is finally working the way it should. Folks doing a retrofit should realize what it took to get this thing working. After the first time I really dove into the spindle, I was almost convinced it was going to be impossible. The waybackmachine helped a lot finding the manual for the VFD and at somepoint, I was able to get the WINDOWS XP software to plug in my wifes ancient laptop and a crossover serial cable to access the drive. If anyone needs the software for the allen bradley vfd, I will gladly hook it up. There are some advanced hidden modes, so if you get into it, take screen shots of everything and save all the files. Why do you ask? Because my vfd had a bad fault and needed to be replaced. The only way I got the replacement was from an ebay seller who I was able to work out a core exchange. He wanted $700, so I offered him the core and $300. Once the new one arrived, the values from the old vfd were entered, and it all worked out fine. Here is an interesting fact for those going to program their spindle: 5000 rpm on the VFD equals 7500 at the spindle. After getting a better understanding from the HAL spindle example, I was unable to get the thing to turn. Ahhh... dont forget to enable the spindle on the MESA card! And it was up and spinning. Feedback worked out, but for the tach on the screen to work, it took a little extra than the example. Use the mult2 component to change the speed from RPS to RPM. The final part of the this piece is the spindle speed override potentiometer. The trickiest part to me was the using the scale and conv-float-s32 components, which was required to convert the scale.0.out (float) to halui.spindle-override-value (s32).

Any thoughts on why m19 does not index the spindle?

Only a few more things to do! ATC... I guess its time to read up a little more. I did take a look at a plc tutorial website, which helped explain a lot about the components of classic ladder. Until next time!
MAC

Before you get too far into tuning it I would first adjust all the linear bearings so that you're tuning with the amount of friction that you'll have when milling. There should be 4 setscrews on each axis that press against the 4 linear bearings. The manual calls for 4-8 in/lb then back it off by 1/8 a turn. Personally I would also go through and adjust these based on how much play you can deal with. Mount an indicator (preferably one with 0.0001" resolution so the tool looks as rigid as a wet noodle) to said axis carriage and measure against a fixed machined surface. Then manually push and pull on the carriage so you can see how much play you have. You'll want to check across the full range of travel as you'll probably have more wear in the middle. Eventually you'll get to a adjustment that wont be too tight to move and not have too much play. Remember the total play in each axis is a sum of how much backlash is in the ballscrew, how tight the belts are to the servos, how closely the servo drive regulates the position and the play that the other two axises have in their linear bearings.

Also the manual calls for "SAE 10W engine oil" to lubricate these linear bearings. This probably isn't the best choice as engine oil really isn't designed for sliding surfaces on machine tools. What most ways call for is something similar to Mobil Vactra No 2, which is designed to have a low stiction and has tackifiers that help keep it from sliding off the non-horizontal surfaces. You can find Vactra No. 2 pretty easily but the trick is it has the viscosity of SAE 20. Vactra No. 1 has the lower viscosity which you probably want since this is such a small tool, but I spent an afternoon searching the internet and couldn't find it smaller then a 55 gallon drum. What I ended up doing was buying a quart of No 2. and also a quart of a spindle oil and mixing them together to get the desired viscosity. It's not ideal but at least it has the tackifiers in it then.

This was helpful to read: www.practicalmachinist.com/vb/general/sy...-227553/#post1596217

MacGalempsy wrote: Any thoughts on why m19 does not index the spindle?

All that M19 does is set a couple of HAL pins. It sets a pin saying "orient the spindle" and another pin with the requested angle.

You then need a PID component to be activated to move the spindle to the setpoint, just as with any other axis.

Have a look at this Wiki page for one of the many ways that this can be approached:
wiki.linuxcnc.org/cgi-bin/wiki.pl?SpindleOrient

If you found the software to communicate with the VFD I would really like to have that! I called and they told me it was unavailable. I'm sure my unit is probably configured correctly but for the long term operation, I would like to be able to work with it in case of changes!