Advanced Search

Mesa 7i85 and 7i85S can do 4 encoders, and can usually be found.
A Mesa 7i92TM or 7i96S with 7i85 and 7i83 and 7i84 is all you would need. 7i96S has some IO so might not need 7i84 at all.

The machine is a Fadal with a 21 tool Geneva.

Hi all - I'm just trying to plan out what is possible before I get too far into Linuxcnc . Its an old Colchester TriumhCNC lathe. +/-10V velocity control as usual with a 3TRM2 amp. The amp works fine but the old Crusader11 controller is now dust. On the lathe there are two slide encoders plus a spindle encoder. The encoders are fairly modest at 250ppr and the spindle is 1000ppr. Im looking to provide some HW assist with the encoders - the obvious route would be with a Mesa card feeding into a DB25 port but a) you cant get them and b) I would like to have a go myself.  Decoding the encoders into count and direction or whatever is a pretty simple HW task. Im OK with VHDL and can make my own PCB's but the question is - does this sound like a possible route to take. Would it be possible to generate a hal layer to receive such an interface. I'm pretty flexible about what the interface might be at the mo. Just trying to sus out what may be possible, Thanks in advance for any assist.

Those are 'joint.' pins and are available for extrajoints. Have a look in the 'Halshow' window that pops up when you run one of the extrajoints sim config:

 

i apologize for the doc files. i will have to post the config files when i am in front of that machine tomorrow. i ssh into that machine from my office to play with the halcmd and i have to copy/paste the files to some kind of editor. and the files i tried to upload they didn't accept so i converted it to doc so it would take it. i hate anything doc anyway. not a big fan of anything that bill guy makes. as far as the machine goes, i will verify the wiring and try again. one question i have is; is the spindle light on tb2, on the mesa 7i96s supposed to come on when i turn on the spindle? is that normal? i mean there is nothing physically connected to it. so why would that light come on? i know there is a way i can change that. but i am at the mercy of you fellows for now until i can get more familiar with this awesome program. and i am still learning hal to be able to spot if something is connected or not. this speed control is not that big of a deal for me. if i get it working through the mesa board fine but if not i am not going to sweat it. i was just going through the manual for that drive (for the tenth time at least) and i can control the speed with a pot if thats the way i have to go for now. its a mountain of information to go through and i only started a couple months ago with linuxcnc. and its not in laymen terms. and i am a layperst.

What version of linuxcnc and how did you get it?

Is your machine imperial based? There was a fix for (what sounds like) this, that has not been released yet.
It looks like the position is out by about 25.4 times.

It's already there, I suggest you read the documentation to see what's available and what Linuxcnc actually does.
linuxcnc.org/docs/stable/html/
Linuxcnc does everything. An FPGA that interprets gcode is not what Linuxcnc works with, all the planning and interpretation is performed by Linuxcnc. Is is unlike the GRBL model where the PC is just used for sending gcode via serial (usb to serial)
What an FPGA or any external board is useful for is step\direction generation based on timing sent by Linuxcnc.

linuxcnc.org/docs/stable/html/

I had some time to try out some new work holding, and it does open up ways to go about secondary operations. But what i noticed is working holding factors into these little machines quite a bit. when you start getting into thicker pieces. radial cutting becomes a pain when you want a good finish. one other thing i don't care for with the new clamps is the fact that you have to count on aluminum threads. over time they will fail due to wear and since it is part of the clamp that will become a problem at some point. So i think i will try a new approach with a 3 part clamp that uses the steel T nuts for the clamping force. once I thought of it i was like stupid simple is a Plus sometimes

one other thing i was working with was the uses of work offsets. and came into the issue of table variation on the Z axis. i wanted to try not skimming the table top but i think if i do it will make things much more simple in the long run.

i actually am waiting for some solid carbide High-feed mills to try too. I think they might work well with this machine they are not cheap but if they work well it is what it is and you just suck it up

I'm building an FPGA based motion controller that will perform very simple actions (linear moves and arcs). I plan to have it implement a small set of G-code, likely something like G00, G01, G02, G03, G04, G20/G21 depending on the native units of the machine, G38.*, G90, M3, M4, M5, M7, M8, M9, F and S. I'm looking for some other application to do all the other work like cutter compensation, keeping track of tools, handling relative moves, etc. Is there a good layer to introduce this into LinuxCNC and take advantage of everything LinuxCNC has to offer? Does this functionality already exist, if not do I want to be making something at the driver level? I've looked around a bit, but don't know LinuxCNC that much so I could easily have missed something.

spumco wrote:

If you have an ATC that doesn't follow the basic mechanical config intended by the PB devs, you'll need to do some heavy editing. 

To clarify... the macros are designed to work with a geneva drive and a 'dumb' on/off motor.

COlger81 wrote:

I have a working Gmoccapy config using carousel component. Is it possible to use Probe Basic with the carousel component instead of the macros?
 

Yes, but it requires a fair amount of hand-fiddling.

If you have an ATC that doesn't follow the basic mechanical config intended by the PB devs, you'll need to do some heavy editing.

Search the qtpyvcp forum... I think I posted my config a (few times) that is for PB using carousel.  I have a stepper motor and 5:1 reducer with one index sensor, and one per-pocket sensor.

You should post actual . hal and .ini files here, not .doc files

 

Does it change at all when you set z-accel-fraction way low?  If not, you may have a typo in your hal file.

You can check if it's active using halshow.  Find the z-axis jog-accel-fraction pin/parameter and you can change it directly in halshow.

If the response changes a little, but not much... then do you have any mechanical issues? Backlash or sloppy ballscrew end bearings?

here is my .hal and .ini files.  

so, that helped on the X and Y axis. but the Z is still very choppy, almost regardless of what value I use there.

other thoughts?