Emco PC Turn Turret

Hey,

I haven't gotten that far with my development yet, and the hardware on the PC turn 50 may be different I'm not sure (my unit is a PC Turn 55 with two optical sensors one for tool position plus an opto on the 'home' location)

But, there may be some case for using additional logic here: wiki.linuxcnc.org/cgi-bin/wiki.pl?HAL_Components

Also, HAL timedelays may be useful?: linuxcnc.org/docs/2.4/html/man/man9/timedelay.9.html

In one of the examples linked by tommylight the toolchange_index.hal uses timedelays on the servo thread.

I'm new to the more advanced usage of HAL and don't fully comprehend your situation but those might be worth exploring. Also, some of the M Codes used in the mapped gcode file can include delays that may be useful (e.g. M19)

I'll let someone that knows HAL better than I answer you further though.

There are of course some differences between the 55-series and the 50 but that is mainly the shift to 3-phase steppers and a Lenze VFD. The tool turret is 8-position as well compared to 6 but the basic construction of the tool turret is the same with the same two optical sensors.

I think that the problem can be traced to the behavior of the index and pocket signals, as can be seen in the attached picture. Since the turret backs off the pocket sensor during the locking of the tool a change to the next tool isn't complete until the pocket sensor activates the second time. If it stops during the first activation it will remain within the same pocket as before.

I've tried to implement some sub routines in my M6 remap where I first run the turret motor until passing the first pocket trigger, thereby leaving the detection of the second/next pocket to the carousel component. Unfortunately nothing happens when I try a "manual" tool change but I don't get any errors either.

Got you. It's probably best to start a separate thread and share your HAL and INI files as I'm sure someone would be able to chip in. There may be some need to explore O codes sub routines and conditionals in the mapped GCode files, I haven't looked at it in any depth yet. If I do I'll certainly share it here (or develop an Arduino based solution to interface with Carousel vs serial)

Hope you get it working, best, John.

I was worried for a minute that there were several Emco users struggling with the carousel comp.

I am hopeful that we will find a fix for the fact that the changer backs off the sensor after completing the change soon.

Thanks Andy. I haven't really got that far just yet but hope to soon. Im just getting the mesa 7i96 prepped outside of the lathe for the other functionality before digging into the carousel. I've been looking at it but am getting a little lost with the conceptual layers in HAL so I've a lot of learning to do with HAL and comp and am busy with work at the minute. I don't doubt I might encounter it when I get there but there's no way to say either way yet. I have been looking at other emco comps too to see if modifying those might be easier. Although I'd prefer to stick with built in functionality if possible.

My initial thoughts were to modify it with an additional conditional so that pockets are only counted when DC motor is active and ignored when it is reversing. I'm not sure if that would work in the context of having a comp that suits many purposes though, I understand

As an update, I managed to get this turret working correctly via classic ladder. There's a gent on Youtube that goes by the name of feral engineer that kindly shared his classic ladder files, with a few changes to nets I made it work with the MESA 7196 via SSR output > 5V Relay > 24V control board relay. (His unit was a newer 8 position turret whilst mine is 6 position which required one small change in the comparison logic.

Essentially his solution places a 100ms delay on the position/pocket sensor which feeds a counter, such that the sensor must be high for 100ms in order to be counted, whilst the index/home sensor resets the counter to 1 each pass. Combined with comparison of the requested tool via TX this solves the latching issue and an option of such a delay might be something to consider in a comp (if it isn't blocking code). I'd go about it differently if I were coding it in C personally, but, it's working so I'm not messing with it at the minute. It was interesting to learn a little about classic ladder regardless.

I will eventually share my files (provided he is okay with that) once I get the remainder buttoned up.