Hello Reg, Thanks for the reply. Our I/O subsystem was designed by our company. It consists of semi-intelligent modular I/O racks that are combined in the right numbers for each job. We have interfaced to these via an ISA bus card, but a PCI version is soon to follow. The motion base is an older style analog system. All of our newer Millennium sims use DOS based digital systems, the host communicates with these via fiber (most of these machines are running on Harris NightHawk UNIX hosts). We also design and build our visual systems. This machine will probably be re-fitted with a new Vital 8 daylight system. It uses dozens of Power PCs in parallel to generate real-time raster/calligraphic scene presentations. The ground scene content is based on satellite imagery, and is swapped in from a dedicated visual host system as the pilots eye view moves. I am working with Linux 2.0.36-0.7 with RTL 1.1 patched in. Are the FPU problems indeed solved on this version? I have read that 100usec jitter was common in previous versions (while using the FPU). Some of the later e-mail indicates the cause for the delays have been isolated, but I haven't seen any indication that they have been implemented yet? What iteration rates are you running? I believe I will try to sample the I/O at 30hz. These older machines all ran their "fast band" at 20hz, with lower priority modules running at 10, 5, 2, and 1 hz. Were you using a DOS extender on your DOS based 6 degree sims? Sincerely Todd > ---------- > From: > [EMAIL PROTECTED][SMTP:[EMAIL PROTECTED]] > Reply To: [EMAIL PROTECTED] > Sent: Sunday, March 28, 1999 3:48 PM > To: Gearheart, Todd > Subject: Flight simulators > > Dear Todd, > > I've been going down a similar track and have chosen to convert our > 6 axis controllers from DOS to RTLinux. We developed our own > hardware interfaces based on an ISA card containing 2 Hitachi UPP > chips - each has 10 A/Ds (10bit) and 24 16bit counters (Up-Down, > PWM etc), and are currently making a dual Xlinix chip that is > programmable so that we can get instaneous capture and output > independent from the computer. Thus latency is nolonger a problem. > > At the moment we have a 6 axis hydraulic robot that runs on PWM > signals, and a 6 axis electrically driven Stewart platform that used to > run on PWMs but now uses D/As for constant current (force) > drives. So far we have a single hydraulic ram running very well > under RTL and are still playing about with the various tasks and > displays (using Matlab on another Linux box). > > We had some initial problems with latencies and saving the floating > point registers but I gather those have been solved in the latest RTL > releases. So far RTL looks like a very good replacement for DOS > and we anticipate using it for all our newest projects. > > Best regards, > > Reg > --- [rtl] --- To unsubscribe: echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED] ---- For more information on Real-Time Linux see: http://www.rtlinux.org/~rtlinux/
