Re: Alternative IO cards



Hi Nic

I would expect less work in using a standard USB stack 
augmented with an external single chip microcontroller to do 
detail timing. Should be less than a $10 cost adder.
Compared to custom stacks both ends that chase the kernel.
and are perhaps incompatible with any other usb peripheral. 
(say like the mouse/keyboard)


PCI is always possible. remember however there are a 
few PCI standards (5v, 3,3v, 32 bit, 64 bit, 33MHz, 66mhz, PCI-X
etc  that sometimes are mutually incompatible, and at othertimes
will slow the whole system to the slowest card. Its evolving fast.

I'd suspect that as you say, adapting to a pci card is less work in the
short term, but will die faster, compared to USB. 
 
I suspect USB will be with us for a while. 

maybe I need to polish my crystal ball some more though..

john





On Sun, May 26, 2002 at 07:39:43PM -0400, Nic van der Walt wrote:
> 
> 
> >Also - thought strikes me that modules like freqmod may suffer.
> >Due to the delays in software layers between the rt module and the port
> pin.
> 
> USB could do it, if the driver code is tight enough. Would need to be an
> extension
> of freqmod that talks directly to the USB hardware on the motherboard.
> 
> The slave device would have to be custom USB, generic USB to PP cables
> tend to be slow.
> If you control both sides of the USB link completely it should be more
> than good enough.
> 
> >One solution would be to offload much of the real time behaviour to
> >the external interface.
> 
> That is going towards an expensive processor based system. Dumb PCI card
> might
> be better, or simply use a PCI based parallel port card and keep the
> status quo.
> 
> Parallel ports are good for another year or two.
> 
> Nic.
> 
> 



Date Index | Thread Index | Back to archive index | Back to Mailing List Page

Problems or questions? Contact