Timeline for Writing to parallel port with minimum latency
Current License: CC BY-SA 3.0
20 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 16, 2012 at 9:03 | comment | added | Nick Johnson | There are alternatives to Arduinos for embedded electronics - and doing the individual steps from a microcontroller is a much better idea than trying to bitbash using the parallel port of 5 year old computers. You could build a simple board based around an ATTiny and a stepper driver quite cheaply, for instance. How do you plan to control 1000 stepper motors from 3 parallel ports or 3 Arduinos, though? | |
| Oct 15, 2012 at 21:40 | answer | added | Chris Stratton | timeline score: 3 | |
| Oct 15, 2012 at 18:16 | answer | added | gbarry | timeline score: 0 | |
| Oct 15, 2012 at 15:54 | answer | added | davidcary | timeline score: 0 | |
| Oct 15, 2012 at 13:54 | answer | added | Blup1980 | timeline score: 0 | |
| Oct 15, 2012 at 13:48 | answer | added | Blup1980 | timeline score: 4 | |
| Oct 15, 2012 at 12:39 | comment | added | EM0 | It will be running additional code, of course, but all on the same thread. Basically, all the instructions for the motors will be read into memory up-front (the PC has more than enough RAM for that) and then it's just a matter of sending them to the motors. Nothing else should be happening at that point. I will test it out on RTLinux as well, regular Linux was just easier for the initial test. | |
| Oct 15, 2012 at 10:37 | comment | added | Wouter van Ooijen | Sorry, I meant I would not trust dividing the total time by the number of cycles to yield the single-cycle time. Too much chance of other things going on. On average the figure will be right, but that is not OK for smooth operation of a Chinese army of steppers. | |
| Oct 15, 2012 at 10:35 | comment | added | Dave Tweed | Right now, you're just running a single simple test loop on the PC. If that's all this box will ever be doing, then the Linux-based PC is way overkill for this task. But we're presuming (because you haven't told us) that this PC will eventually be running additional code to define the overall behavior of the 1000 motors, and this other code will interfere with the timing of the parallel port. | |
| Oct 15, 2012 at 8:24 | comment | added | EM0 | Why do you say measurements under Linux cannot be trusted? The machine was not doing anything else at the time and the time was consistently 3.60 or 3.59 seconds. If the OS was interfering I would expect much more variation from test to test than 0.3%. Could you post any more details or links on CAN or RS485 network? I'm new to this stuff. | |
| Oct 15, 2012 at 7:49 | comment | added | Wouter van Ooijen | I would say that any measurement on Linux should be distrusted. I am not sure about RTLinux. I also agree with the others that this is probably not a good architecture for what you want. I would consider a CAN or DIY RS485 network with microcontrollers, each controlling a bunch of steppers, commanded by high-level commands (not step-level commands ). | |
| Oct 15, 2012 at 6:51 | comment | added | EM0 | Apparently I do realise that, since I mentioned RTLinux. The initial test was not done under RTLinux, but are you suggesting that that was the limiting factor in this particular case? | |
| Oct 15, 2012 at 6:23 | comment | added | Wouter van Ooijen | Do you realize that Windows, Linux and other desktop systems can take control at unpredictable moments and stall your application for let's say half a second? There is a good reason some people use what is called a RealTimeOperatingSystems, which those desktop OSes definitely are NOT! | |
| Oct 15, 2012 at 3:46 | review | Close votes | |||
| Oct 22, 2012 at 3:00 | |||||
| Oct 15, 2012 at 3:31 | history | edited | EM0 | CC BY-SA 3.0 |
added 1163 characters in body
|
| Oct 15, 2012 at 2:53 | answer | added | user3624 | timeline score: 2 | |
| Oct 15, 2012 at 2:26 | answer | added | D.A.S. | timeline score: 1 | |
| Oct 15, 2012 at 1:39 | answer | added | Dave Tweed | timeline score: 2 | |
| Oct 15, 2012 at 1:31 | review | First posts | |||
| Oct 15, 2012 at 19:39 | |||||
| Oct 15, 2012 at 1:29 | history | asked | EM0 | CC BY-SA 3.0 |