On May 7 2013 2:55 AM, Michael Haberler wrote: > Am 07.05.2013 um 10:26 schrieb EBo <e...@sandien.com>: >>> >>> I think it would be a very useful study topic (and a paper worth >>> publishing) to start with this work, and go a step further >>> >>> a sketch for the work based on the above I would find interesting >>> in >>> general and very relevant to the discussions here: >>> >>> - the question I would care about is 'what is the incidence to >>> loose >>> steps, given a certain load, and a certain noise distribution in >>> the >>> stepping signal' >>> - start by measuring actual noise (latency profiles) of currently >>> used RT OSes, including a vanilla kernel >>> - create a signal generator can load and regenerate noise profiles >>> (better not based on linuxcnc but hardware, say a microcontroller) >>> - create a setup of stepper motor(s), a hires encoder, and a DC >>> motor >>> with a controllable load (eg switchable shunt resistors or >>> somesuch) >>> - come up with a way to detect lost steps based on input signal and >>> encoder signal >>> - automatically run various speed, noise and load profiles and >>> qualify them by 'lost steps' incidence >>> >>> I think the result of such a study could provide fact-based answers >>> to what latency in a soft-stepper context actually means, and that >>> be >>> very valuable >>> >>> Note this does _not_ address the (IMO more interesting) question >>> how >>> latency impacts 'path tracking quality' of a real and complete >>> motion/pid etc servo setup; that would be worth a separate attempt, >>> probably more based on control theory than measurements plus some >>> verification >>> >>> no junior researchers out here itching to publish? >>> >>> - Michael >>> >>> ps: on the question of shunt resistors - Amit's company might have >>> some scalable answers here, but Amit better explain their business >>> himself ;) >> >> publishing? What is the target journal or outlet. Maybe one of the >> academics can convince a student to take it on. It would take a lot >> to >> really make this publishable in a peer reviewed journal/conference. > > whatever, whoever - provided somebody actually does at last some > meaningful work instead of contributing yet another round of > conjectures > > as for publishable - that's the sales part of the job which might > require senior attention > > given the amount of crap published as 'academic articles' and which > even carries some lowlifers to tenure, the content shouldnt be the > issue ;)
speaking strictly for myself... writing such a paper would take me more time than running such an experiment and would not help me in the least. I do not mind helping solve the conundrum though. Do you have access to suitable hardware that has uber low latency that we could drive such a beast? EBo -- ------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
