Peter C. Wallace wrote: > >In other words the host filling the FIFOs knows exactly how many >(whole+fractional) steps are output each frame. > > > OK, I guess I'm not clear on what is between the CPU and the DDS. If the computer was just placing a new value in the addend register of the DDS, then it would have to know the frequency of the DDS' master oscillator to extreme precision. Now, maybe your FIFO is being emptied at a rate created from the same clock, and that's the secret to how it works. The CPU just has to fill FIFO entries, knowing they will be pulled from the FIFO at precise intervals. And, if the FIFO ever runs empty, the system stops. My understanding of Window's "real-time" performance is that there is NO upper bound on latency to a user-mode process. My experience using Windows systems certainly confirms this does happen. Just pop in a CD, and there can be ** 15 second ** delays where the mouse cursor doesn't move, etc.
>This was mainly done for people using Windows so that real time reponse was >not required to get a perfectly timed output stream. > >I'm working on an alternate step gen for EMC that will be free running and >unbuffered. Its a little more involved because you cant avoid the direction >setup time problem as easily as the buffered approach and therefore requires a >dir change deadtime step accumulater-deccumulator. I should have something >tommorow or Monday. > > I had to go through all the same on my stepper controller. It isn't all that hard. Especially because direction reversals are always going to happen at low speeds, so a 20 us delay of the first step in the new direction is not likely to make a motor lose sync. Jon ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
