On Thursday 22 March 2018 09:45:49 Maxime Lemonnier wrote: > rt-preempt made no difference (well my HAL Latency Test are even > better) > > Unplugging the second ppmc card cut the time roughly in half > (expected) > > 22 s32 OUT 277802 ppmc.0.read.time > 22 s32 OUT 194461 ppmc.0.write.time > 16 s32 OUT 486932 servo-thread.time > > I conclude that the NetMos chips are just slow. sad. I can't even do > 2KHz safely with these timings.
I've come to theconclusion that amd cpu's are generally slower than intel. And that generally speaking, Dell's are usually pretty decent. Here are the relevant figures for the Dell I originally was going to use to run the sheldon, until I got this wild hare and decided to use a r-pi3b, that decision driven by the lack of sufficient i/o of a 5i25 card even with bobs on both ports, whereas the $65 7i90HD has 72 gpio's. from its dmesg: CPU0: Intel(R) Pentium(R) 4 CPU 3.20GHz stepping 03 2GB of dram. latency-test servo-thread jitter 28072 period 994XXX base-thread jitter 20649 period 246XXX bear in mind the base-thread isn't used by the mesa 5i25, so thats pretty good, not the fastest, but plumb usable, and a similar machine is running my G0704 mill. Finding a good parport, or even using the mobo's version, driving a 7i90HD would give 72 gpio's, but comm time might be excessive enough to cause ghost following errors. Not tried by me. The 7i90HD comes rigged to talk to a parport, but the actual comm bandwidth by using the pi's spi port, is I believe, faster. The spi firmware is installed with mesaflash of course. The 5i25 seems to be able to stand more noise, but I will not say I've not blown one as I have, poor frame grounding blew the pin I was using as a probe input. The 7i90HD (both are 3.3 volt logic) seems to be easier damaged by stepper motor noise pickup, so will need a 7i42TA for each one of the 3, 50 pin ports as a noise and spike suppressor, which adds to the cost but makes it bulletproof at the same time it also gives you a much much better way to get things connected as it gives you the usual green terminal strips for connections to the machine as opposed to trying to keep jumper wires connected to the correct pins of a 50pin male socket. Look at the cost of the 7i42TA's as the BoB's it doesn't need. Both the 5i25, and the 7i90HD have more than enough pwm's for machine control. I don't know enough about the Pico board to comment. One other thing of note, anytime a pwm signal goes thru an opto-isolator, anything resembling control linearity is lost. SainSmart is the only bob currently for sale without the useless opto's in the output paths. Inputs are still fed thru opto's on the SainSmart bob's, so I had to remove the 2 used for encoder A/B inputs as they were waaaaayyyyyyy too slow if the A/B signals are coming from the motor shaft, not the spindle. > > On Thu, Mar 22, 2018 at 9:22 AM, Maxime Lemonnier < > > [email protected]> wrote: > > Interesting, more in line with what I would expect. What lspci -vv > > tells you about the chip on the PCI card? > > > > As I wrote you this morning I did try another PCI card I found: > > > > NetMos Technology PCI 9865 Multi-I/O Controller (prog-if 03 > > [IEEE1284]) > > > > The PCI card is even worse: > > > > 10 s32 OUT 534652 ppmc.0.read.time > > 10 s32 OUT 366536 ppmc.0.write.time > > 8 s32 OUT 907936 servo-thread.time > > > > > > Maybe I could try to boot in rt-preempt (using rtai right now) > > > > On Wed, Mar 21, 2018 at 9:46 PM, Jon Elson <[email protected]> wrote: > >> On 03/21/2018 03:15 PM, Maxime Lemonnier wrote: > >> > >> OK, I checked my system with a mediocre Dell Pentium 4 CPU and a > >> PCI card, and ONE universal PWM controller. ppmc.0.write was 60 > >> us, max was 95 us. > >> ppmc.0.read was 128 us, max 188 us. > >> The total thread was running 386 us. > >> This is a Debian Wheezy system with the rt-preempt kernel. > >> > >> Jon > >> > >> ------------------------------------------------------------ > >> ------------------ > >> Check out the vibrant tech community on one of the world's most > >> engaging tech sites, Slashdot.org! http://sdm.link/slashdot > >> _______________________________________________ > >> Emc-users mailing list > >> [email protected] > >> https://lists.sourceforge.net/lists/listinfo/emc-users > > ---------------------------------------------------------------------- >-------- Check out the vibrant tech community on one of the world's > most engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Emc-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-users -- Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene> ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
