#linuxcnc-devel | Logs for 2015-03-06

[07:11:38] <skunkworks> oh well.
[07:11:39] <skunkworks> http://www.practicalmachinist.com/vb/deckel-maho-aciera-abene-mills/deckel-nc-upgrade-cons-pros-299846/index8.html#post2503889
[07:12:04] <skunkworks> cradek inserts xkcd comic about arguing on the internet.
[07:19:39] <skunkworks> wasn't there newer block diagrams of linuxcnc
[08:39:43] <cradek> https://xkcd.com/386/
[08:40:59] <skunkworks> Thank you :)
[08:46:54] <seb_kuzminsky> skunkworks: that's the diagram i know of
[08:47:46] <seb_kuzminsky> i have a branch where i started to modernize it but never finished, it's called 2.5-architecture-docs
[08:48:00] <seb_kuzminsky> and it makes me embarrassed every time i see it
[08:50:51] * skunkworks laughs and points
[08:53:31] <seb_kuzminsky> heh
[08:53:39] <seb_kuzminsky> too many projects, too little see-through
[08:55:16] <skunkworks> https://groups.yahoo.com/neo/groups/mach1mach2cnc/conversations/messages/146900
[09:23:06] <cradek> heh where he says yaskawa I bet he means emc
[09:23:31] <cradek> I wonder what the actual bug was (with small moves) and whether we still have something like it
[09:24:45] <skunkworks> Ours says it will gouge.. I think mach just does it.
[09:25:09] <cradek> ah an undetected error condition
[09:26:22] <skunkworks> I think this was one of the questions..
[09:26:23] <skunkworks> http://electronicsam.com/images/KandT/testing/Screenshot%20from%202014-09-30%2014:51:39.png
[09:26:45] <skunkworks> if the cutter gets too big - the vertical line gets consumed.
[09:29:28] <cradek> skunkworks: yeah then you need a much fancier algorithm OR an error message
[09:29:39] <cradek> skunkworks: if they have neither, :-(
[09:33:10] <skunkworks> right
[09:45:50] <skunkworks> http://www.practicalmachinist.com/vb/deckel-maho-aciera-abene-mills/deckel-nc-upgrade-cons-pros-299846/index8.html#post2504005
[09:54:17] <cradek> skunkworks: sounds like you have to be a programmer to use it
[09:54:29] <skunkworks> I know - right..
[09:55:23] <skunkworks> Kflop seems to be neat solution - but the people that complain that you have to be a programmer to use linuxcnc seem to think kflops programming isn't programming...
[09:57:14] <ssi> fact of the matter is you have to be a machine integrator to use any reasonably complete machine integration tools
[09:57:21] <archivist> it is very hard to convince someone that their preconceived ideas are not right
[09:57:27] <ssi> and being a machine integrator means being a programmer, an EE, a control theorist, etc
[09:57:42] <archivist> an all rounder :)
[09:58:00] <ssi> yes... that's why I enjoy it so much
[09:58:35] <archivist> reminds me of Number 5 and mooore information....
[10:02:53] <archivist> just getting some machine manuals this week meant I could see some more spindle and head drawings
[17:00:09] <mozmck> PCW: you around?
[17:00:37] <PCW> Yeah
[17:01:57] <mozmck> If I remove a stepgen from some pins in PinDesc, do I need to change the number of stepgens in the ModuleID above?
[17:08:55] <PCW> yes
[17:09:25] <mozmck> ok. I figured out I needed the C-step pin for general IO
[17:10:41] <PCW> Ahh yeah if you need a higher numbered one for a chargepump you cant disable intermediated ones
[17:10:54] <PCW> intermediate
[17:10:59] <mozmck> ohh... ok then I'll adjust the numbers.
[17:11:30] <mozmck> I have a platform USB cable now - in case I really mess it up :)
[17:12:07] <PCW> all special functions can be used as GPIO but the special functiona can only be enabled on a 0 through N basis
[17:12:16] <mozmck> ok, makes sense.
[17:12:33] <PCW> (thats a driver limitation for command line simplicity)
[17:13:10] <PCW> the hardware can do it on a pin by pin basis but rather awkward to specify
[17:13:44] <PCW> normally you cant mess things up with pin files
[17:13:54] <mozmck> heh!
[17:14:18] <mozmck> So I see the chargepump just has a step pin and no dir - I guess that's fine
[17:14:39] <PCW> usual first time fatalities are not having the right ucf file (or no ucf file) associated with the top level file
[17:15:36] <mozmck> I have the 7i92.ucf listed under TopEthernetHostMot2, so I think I'm fine there.
[17:15:40] <PCW> yeah you can just use the pins you need from a module
[17:16:05] <PCW> (unused inputs are logically terminated at 0)
[17:17:44] <PCW> (so for example if you make a QA only quadrature counter, you can still use it in the up/down mode but QB will be tied to 0 internally so it will only count down)
[17:19:12] <mozmck> ise spits out plenty of warnings...
[17:21:29] <mozmck> If I need or want to program with JTAG do I use the same bit file?
[17:22:55] <PCW> Yes
[17:23:08] <mozmck> ok
[17:24:04] <PCW> if you want the fallback to work you have to set the bitfile options mentioned in the manual
[17:24:34] <mozmck> manual...
[17:25:07] <mozmck> looking now
[17:27:30] <PCW> For programming the onboard flash via JTAG you need a hex file and its terribly slow so I never do that
[17:28:11] <mozmck> Oh, I see - I think.
[17:28:30] <PCW> instead I program the FPGA directly via JTAG and the program the flash via Ethernet
[17:28:44] <PCW> then program
[17:28:59] <PCW> much faster and only bitfiles needed
[17:30:16] <PCW> if the bitgen options are not in the manual the writer should be shot
[17:30:20] <mozmck> So if I haven't used the -g options then fallback won't work?
[17:30:27] <mozmck> they are :)
[17:31:03] <mozmck> Now I have to figure out how to set those options in ISE
[17:31:38] <PCW> Yeah fallback when CRC errors are detected needs those options
[17:32:17] <PCW> (so if you pull the plug when programming for example, you can recover)
[17:32:52] <mozmck> That makes sense. I do something similar in my bootloader on my microcontroller projects.
[17:34:08] <PCW> unrecoverable are wrong ucf, wrong bitfile (right chip but wrong hardware) etc
[17:34:36] <PCW> (well unrecoverable on cheapo 1 EEPROM cards like the 7I92)
[17:37:04] <mozmck> So how do I set those options?
[17:38:02] <PCW> They should be set if you used our project file let me check with 13.3
[17:39:37] <PCW> on 13.3 its in the properties of Generate Programming File
[17:40:52] <PCW> 14.x seems the same
[17:41:29] <mozmck> I see them under the Process Options for Generate Programming File
[17:41:56] <mozmck> Looks like they are set. I'm using 14.7
[17:50:09] <mozmck> PCW: Looks like it worked!
[17:58:30] <PCW> Jolly Good!
[18:00:53] <PCW> The $39 Ebay DC7800 runs hm2_eth at 4 KHz (at least for a few hours so far)