On 10/31/2010 1:22 PM, Vanessa Ezekowitz wrote: > On Sun, 31 Oct 2010 20:47:16 +0100 > Groepaz <firstname.lastname@example.org> wrote: > >> haha, i dont know but.... you and the other 2 people ..... >> >> personally i dont know anyone who - today - uses more than two drives (most >> infact are moving away from even that to 1541u or similar). and especially >> for a device like this, i think 4 would be more than enough. most probably >> wont even use more than one =P > > For arguments like this, I think of basic UNIX philosophy: either > allow exactly one and only one parameter or instance of something, or > allow as many parameters/instances as the user and the system's > resources can handle > > Since allowing only one drive doesn't make any logical sense in and > of itself, and doing so could get in some peoples' way, it makes > sense to choose the latter of the two options. Then it just becomes > a matter of letting the user know what those resource limits are > (i.e. max number of drives). It isn't that simple. We can't just allow any number of drives. First of all, the s2 and nibbler protocols in OpenCBM both use the ATN line. So those only will never work with more than one drive on the bus. The s2 protocol is the default warp mode in OpenCBM. If you stick to the s1 or IEC routines, you can access any number of devices. However, this doesn't come for free. The more devices you allow, the slower the transition from high to low and vice versa. Everyone will experience this delay, not just the people with more devices. Luckily, I think the increase in delay is relatively small compared to the lack of speed in these protocols. I have measured this with my oscilloscope. With one drive, we can get away with 0.5 us delay between transitions. With two drives, you have to wait 2 us to be reasonably certain the line is stable after switching it. I don't have or want more devices than that, so I want to come up with an artificial "drive" that is a set of pullup resistors and some capacitors. This would simulate the max bus load we want to support. To do this, I need more concrete data. What is the RC load of 4 drives on say 4 meters total bus length? Is there any spec for a max in old Commodore docs? If not, is there a spec for a max RC load per IEC device? Note that this max will only be per IEC bus. You can always use separate USB ports to access more drive chains. -- Nate Message was sent through the cbm-hackers mailing listReceived on 2010-10-31 22:00:16
Archive generated by hypermail 2.2.0.