Back
[06:48:35] <cncbasher> are their any differences in compiling a bit file for a 6i25 than a 5i25 are they the same as far a the firmware goes ?
[07:59:05] <Tom_itx> cncbasher, pretty sure they're the same
[08:32:40] <jepler> The W7/W8 jumpers set the "PCI BUS Clock". You need down/up (the default) if you're going to use a 5i25 bitifle
[08:34:07] <jepler> I have a 6i25 but have only used it in 5i25 mode. It is actually the card I tested when seeing whether the linuxcnc hostmot2-firmware.git could be made to build 5i25 firmwares.
[08:38:39] <cncbasher> thanks both , iv'e just made a file but i get a flashing red , so trying to figure whats changed
[08:38:52] <jepler> that exhausts my knowledge or list of guesses about 5i25s
[08:38:58] <cncbasher> haha
[08:39:06] <jepler> .. and I won't have any time to play this weekend, I fear
[10:21:09] <pcw_home> flashing red either indicates you have a corrupted bitfile (so the fallback config was loaded)
[10:21:11] <pcw_home> or you left the fallback flag in the top level VHDL file set when you built the bitfile
[10:22:25] <pcw_home> ( "sudo mesaflash --device 5i25 --readhmid" should tell you what happened )
[10:29:29] <cncbasher> thanks
[10:32:59] <pcw_home> a flashing red init light is how fallback configs are flagged ( the fallback config also is GPIO only )
[10:34:32] <cncbasher> thanks , i'll look at my build structure , i'm having no luck at the moment LOL
[10:35:58] <pcw_home> well you didnt brick the card at least
[10:36:22] <cncbasher> ha , still have not solved bengits board yet , so theirs time
[10:36:55] <cncbasher> btw merry christmas
[10:38:22] <pcw_home> Merry xmas to all!
[10:38:23] <pcw_home> a hint if playing with home made bitfiles the first time:
[10:38:25] <pcw_home> sudo mesaflash --device 5i25 --write mybitfile.bit --fallback
[10:38:26] <pcw_home> sudo mesaflash --device 5i25 --reload --fallback
[10:40:08] <pcw_home> this loads the untested bitfile in the fallback area and the load the FPGA with the fallback code
[10:40:10] <pcw_home> (so if it bricks the card, a power cycle will reload the user config )
[10:40:11] <pcw_home> obviously dont do this if you have corrupted the user config
[10:41:01] <cncbasher> ok iv'e made notes !, thanks for that
[10:42:09] <jepler> http://sourceforge.net/p/emc/bugs/451/ I guess somebody should look at this, too bad there's not much information
[10:42:40] <jepler> I think I have the hardware I need (7i90 as sserial remote?) but not sure how to introduce a single communication error..
[10:43:35] <jepler> not a very thorough bug report, I am guessing it's filed based on some forum report
[10:44:05] <jepler> honestly I'm not sure that's a good idea because it means we don't have a point of contact to ask for more information or have the original reporter test a fix
[10:48:29] <pcw_home> Yeah the error description is not very good, I'm pretty sure I know what the error is though
[10:50:29] <pcw_home> ( better error reporting was added but it does not distinguish between sticky and non sticky error bits )
[10:50:34] <jepler> between 2.6 and 2.7, micges_ made a number of changes regarding error handling
[10:50:51] <jepler> and I made one change that looks like it's in the path of read error handling
[10:52:01] <jepler> but yes on the surface seems much more likely to be one of micges's changes
[10:52:16] <jepler> I wonder whether he'll have time and a test setup to look at it anytime soon
[10:52:53] <pcw_home> Yeah also My fault for not having a better description of the error bits in the sserial manual
[10:55:01] <pcw_home> We have hardware for injecting noise in the outgoing or incoming data stream. too bad i didn't test the 2.7 changes with that :-(
[10:58:30] <pcw_home> I had to look at the SSLBP source code ( very painful ) to see what was going on
[10:58:34] <mozmck> pcw_home: fyi, turning off irq coalescing on some NIC causes realtime problems.
[10:58:54] <pcw_home> maybe a driver bug
[10:59:25] <mozmck> I think my computer has a marvel chipset which showed that problem.
[10:59:56] <pcw_home> does the chip work ok without poking?
[11:00:14] <mozmck> Could be, but it certainly caused problems! Realtime error on linuxcnc startup, and following error if I tried to move any axis.
[11:00:20] <mozmck> Yes, works great
[11:00:57] <mozmck> computer jitter is only around 15us max
[11:01:00] <pcw_home> that suggests that it doest implement IRQ coalescing at all
[11:01:29] <mozmck> Probably not, but trying to turn it off did bad stuff!
[11:01:32] <pcw_home> or the defaults have it off and the driver is buggy
[11:02:03] <mozmck> I tried it with 3.18.xx and 4.1 kernels
[11:02:21] <pcw_home> on realtek (which doesn't have the option) you can't set it
[11:02:37] <mozmck> It just does nothing there right?
[11:03:04] <pcw_home> I think ethtool bellyaches
[11:03:06] <mozmck> I have a realtek card in that computer I think - I should put the 7i92 on that one and try it to make sure.
[11:04:06] <mozmck> I'm using the line hardware-irq-coalesce-rx-usecs 0 in /etc/network/interfaces to set it
[11:05:05] <pcw_home> try ethtool from the command line, I think is says "no such option" or so such
[11:05:27] <mozmck> I think I saw that. I'll try it again.
[11:05:43] <pcw_home> maybe the Marvell driver should say the same thing :-)
[11:06:18] <mozmck> :-) I may have to check the chipset in my script before blindly setting that option
[11:06:32] <pcw_home> Yeah
[11:07:12] <jepler> hm I thought perl6 was supposed to be released on christmas
[11:07:28] <mozmck> april 1st
[11:07:49] <pcw_home> I could never get the Atheros chip on on of my MBs to work it acted like it had IRQ coalescing on but
[11:07:50] <pcw_home> no way to turn it off
[11:10:02] <mozmck> I may wind up compiling a list of bad chipsets and warn the user his won't work if it's one of those.
[11:10:51] <pcw_home> luckily Intel and Realtek are by far the most common on newer MBs
[12:07:29] <jepler> linuxcnc-features got a writeup on hackaday
http://hackaday.com/2015/12/25/linuxcnc-features-is-the-garage-fabs-missing-cam-tool/
[12:09:31] <jepler> the mentioned gcmc looks interesting too
http://www.vagrearg.org/content/gcmc (gplv3)
[15:12:14] <cncbasher> pcw_home: ive made a new bit file for ozcnc on the forum , can we add a chargepump easily , may as well do it as i have got this far, as he does not wish to change the wiring , making the bit file was the only option
[16:40:07] <pcw_home> The dmmbob pinout should be correct (though you might want to move the PWM output to pin 17 to get access to relay #1)
[16:40:08] <pcw_home> the thing that missing is a chargepump signal, this is normally done with a extra step generator
[16:42:02] <cncbasher_> yea iv'e done that as per the geko
[16:42:24] <cncbasher_> the only changes really was the spindle pwm
[16:43:05] <cncbasher_> but i'll look in the morning and see if i can find why he gets the fallback ,
[16:43:56] <cncbasher_> i'm building under linux , i'll try a windows pc
[16:44:56] <cncbasher_> jepler updated the linux build to take 5i25 , hence why i asked if their was any differences to a 6i25
[16:45:59] <cncbasher_> thats basicly what i did in moving the pwm to 17 and making pin 1 gpio
[16:47:05] <cncbasher_> i'll just have to pull the 5i25 out the mill
[17:55:08] <pcw_home> Check that you dont have the generic "fallback" set true in the top level file
[18:33:14] <cncbasher_> set u email jt