schreite@helena.physik.uni-stuttgart.de
Mon, 28 Sep 98 12:50:39 +0100
Hi,
I was to quick within my last e-mail, it is indeed more complicate, than
I thought in my first, immediate reaction.
Let me speak for ECP only now (for I don't know very much about EPP):
The data/command transfer should not be a problem:
Connect:
HostClk -> PeriphClk
HostAck -> PeriphAck
and vice versa
Data8 -> Data8
This works together, since both Clk signals have negative logic, both
Ack signals have positive logic. In handshaking Clk each is expected before
Ack. So no confusing should occour with this symmetric cable wiring.
But you are right, there is a problem with the direction control.
nReverseRequest could be read by nAckReverse and vice versa,
but the logic is wrong (and maybe the timing). So you need indeed at least
two active logic components in your "cable": negation of both
nReverseRequest out-signals.
Further a ECP port could be confused, if it receives an
nAckReverse with no prior sent nReverseRequest and first of all no
nPeriphRequest received - but you should try it - perhaps it will
work. So just try, if it works,
if nReverseRequest is just negated and connected to nAckReverse and vice versa.
If not, try if it is only a problem of the lack of the nPeriphRequest
reveive. So connect the negated nReverseRequest to both: nPeriphRequest and
nAckReverse (and vice versa).
I don't think, that the order of which signal has to be low first
for the reverse channel, nAckReverse or nReverseRequest, is important.
But if so, you need two more logic devices in your cable:
An and gatter with negated in- and output, which takes care, that
nAckReverse is not set low at that ECP card input pin before that
ECP card put its nReverseRequest low.
The rest, i.e. which default directions, who can when change the direction, ...
is a software problem.
No warranty for what I wrote! Please read it, think about it and tell me,
if I did make another bug!
Roger.
-- To unsubscribe, send mail to: linux-parport-request@torque.net --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2.0b3 on Wed 30 Dec 1998 - 10:18:25 EST