Re: usb latency

At 09:28 AM 4/23/2007, Gene Heskett wrote:
>Hi Greg & company;
>
>Is there available, for a reasonable price, a usb dongle that would 
>do nothing
>but echo the packet back to the src so that some sort of a handle on the
>latency could be measured/obtained?

I doubt one is available. You could reprogram any cypress fx2 high 
speed device to do it, I don't know what commercially available 
devices exist though.

>Over on the emc list, we have a developer that wants to use emc, which
>requires hardware response to such things as limit switch closures in at
>least sub-millisecond response times else the machine, particularly if moving
>at a high speed, might move far enough to destroy the switch (& maybe more
>than that) before it gets stopped by the driving computer.

1) are you sure about the required speed. sub-millisecond is very 
fast. When I did a pinball machine years ago, a 5ms debounce was 
sufficient for the very fastest ball hitting a target. In the pinball 
machine, the requirement was more not missing a switch closure than 
detecting it quickly and doing something. However the debounce was 
done to catch the first closure and try to ignore later ones for a 
known amount of time (the target switch would vibrate open/closed rapidly).

2) With .5ms to 5ms what OS are you going to run on the host? It must 
be a rtos to guarantee response to an event.


>This developer is already conversant with building usb stuff from the chipset
>up, so the actual implementation doesn't seem to be a huge problem for him,
>if the answer, in the 400 megabaud speeds of today, is a usable figure.

High speed isoc IN and OUT could be fast enough over the bus, but 
refer to the issue in 2) above.

>This question and issue is primarily driven by the gradual disappearance of
>the std db25 parport hardware from the newer machines, the portables in
>particular.
>
>We would like to be able to give a definitive yes or no answer to 
>the question
>of "will it work?"

In my opinion, USB under linux is not a good solution for very low 
latency requirements. I think a hardware redesign with intelligence 
in the gadget to prevent disasters would be much more sane.

Regards, Steve 


-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel