#linuxcnc-devel Logs

Sep 06 2018

06:42 AM jthornton: when using the software stepgen component with the parallel port is it possible to have different step and direction timings for each stepgen?
06:44 AM jthornton: for example if I had 3 gecko 203v's and 1 gecko 201 the step and direction timings are different
06:45 AM rmu|w: jthornton: settings for G201 should work for G203
06:46 AM jthornton: that's not my question :)
06:48 AM jthornton: and no the timing for a g203v will not work with a g201 because the 201 direction setup is 20000 and the 203v is 2000
06:49 AM rmu|w: but the timing for a g201 will work with a g203
06:50 AM rmu|w: the man page says you can specify timings per stepgen
06:53 AM jthornton: I missed that
06:53 AM rmu|w: with software-stepgen, signals are always at least one thread period long anyways, except if you use those "auto-reset" feature of the parport module
06:54 AM rmu|w: http://linuxcnc.org/docs/html/man/man9/stepgen.9.html says parameters are specified as stepgen./N/.parameter, /N/ being the instance number of stepgen
06:58 AM jthornton: yea I've been staring at the man page for the last hour lol
07:00 AM rmu|w: direction setup means hold direction signal AT LEAST AS LONG as direction setup time before doing a step
07:03 AM jthornton: I'm programming a parallel port configuration tool in pyqt5 for 2.8 so I'll have drive timing on each joint
07:05 AM rmu|w: i suggest having editable driver presets and choosing the preset per joint
07:06 AM jthornton: https://imagebin.ca/v/4ElVeevqbFsb
07:06 AM jthornton: that is what I have now
07:12 AM jthornton: looks like to get the minimum base period I need to find the highest step time and step space
07:20 AM rmu|w: is that a question? minimum base period is determined by latency
07:20 AM rmu|w: if you need step frequency that is higher than 1/base period then tough luck
07:20 AM jthornton: https://paste.ubuntu.com/p/mcR5cGtzTj/
07:21 AM jthornton: that's pretty much the code from stepconf
07:22 AM jthornton: time to hit the shower and start my real work
09:56 AM pcw_home: There's no real reason not to use 20000 or even 50000 for the setup/hold times for _any_ drive (at least with the software stepgen)
10:00 AM pcw_home: similar thing with step timings in general the should greater than the drives minimums but people tend to set them at the minimums (often causing flaky behaviour, temperature sensitivity etc)
10:09 AM Jin|away is now known as Jin^eLD
10:09 AM Jin^eLD: hi
10:23 AM JT-Shop: pcw_home: thanks
10:23 AM JT-Shop: should you still try and use doublestep?
10:28 AM pcw_home: doublestep?
10:36 AM JT-Shop: yea if you set the step space to 0 you can get two steps per period
10:37 AM JT-Shop: http://linuxcnc.org/docs/2.7/html/hal/parallel-port.html#_using_doublestep
10:43 AM pcw_home: You can get _one_ step per base period using reset
10:43 AM pcw_home: otherwise you get a step every other base thread
10:43 AM rene_dev_: pcw_home why not always have 50% duty cycle on the step pin, thats how other controllers handle that. requires no timing shit at all
10:44 AM pcw_home: Bad for OPTO life mainly
10:46 AM pcw_home: IF you want 50%, quadrature is better in every way
10:46 AM rene_dev_: quadrature is always better in every way ;D
10:46 AM rene_dev_: how does that affect opto life?
10:47 AM pcw_home: LED on-time is a life limiting thing on OPTOs
10:48 AM pcw_home: unfortunately most step drives use really crappy OPTOS
10:48 AM rene_dev_: yeah, but pepole that are using softstepping are mostly using $10 china breakouts
10:48 AM pcw_home: Yeah
10:50 AM rene_dev_: and cheap drives :D
10:54 AM rene_dev_: but the dir pin also has a 50% duty cycle
10:54 AM rene_dev_: so no point in having the other opto last longer
10:56 AM pcw_home: yeah but may be lower current / higher CTR degradation resistant
10:59 AM pcw_home: thats is the step OPTO pin needs to have a stiff pullup on the transistor side for speed so may not tolerate CTR degradation as well
11:00 AM pcw_home: on the other had most cheap drives have no reverse input protection on the OPTOS and even tiny ESD will reduce the CTR markedly
11:04 AM pcw_home: One minor issue for software stepping is that you can't do 50% with the double step option
11:05 AM pcw_home: or you would not want to anyway (spin too long)
04:13 PM jepler: https://www.renesas.com/us/en/products/optoelectronics/technology/ctr.html shows CTR declining by <50% in 2e5 hours ~= 23 years, not sure I believe it. The next table, which shows that merely by keeping your optos at -25C you get a 1140 year operating lifetime seems more suspect still. (Figures 4 and 5)
04:17 PM jepler: and all without discussing duty cycle hm
04:29 PM pcw_mesa: Yeah, I dont trust the OPTOS in Chinese drives much OTOH the electrolytic caps probably only last a couple years on a hot drive so maybe its not an issue
05:12 PM Jin^eLD is now known as Jin|away