AM79C90 Advanced Micro Devices, AM79C90 Datasheet - Page 60
![no-image](/images/manufacturer_photos/0/0/14/advanced_micro_devices_sml.jpg)
AM79C90
Manufacturer Part Number
AM79C90
Description
CMOS Local Area Network Controller for Ethernet (C-LANCE)
Manufacturer
Advanced Micro Devices
Datasheet
1.AM79C90.pdf
(62 pages)
Available stocks
Company
Part Number
Manufacturer
Quantity
Price
Company:
Part Number:
AM79C900AJC
Manufacturer:
AMD
Quantity:
5 510
Company:
Part Number:
AM79C900AJC
Manufacturer:
SIG
Quantity:
5 510
Company:
Part Number:
AM79C900JAJC
Manufacturer:
AMD
Quantity:
359
Company:
Part Number:
AM79C900JCDV
Manufacturer:
AMD
Quantity:
1 831
Part Number:
AM79C901AJC
Manufacturer:
AMD
Quantity:
20 000
Part Number:
AM79C901AVC
Manufacturer:
AMD
Quantity:
20 000
Part Number:
AM79C90JC
Manufacturer:
AMD
Quantity:
20 000
C-LANCE device incorporates redesigned ALE logic to
prevent this glitch from occurring.
However, in the C-LANCE, when ACON = 0 (active
high ALE), ALE is driven high before it is tri-stated at
the end of every bus mastership period. In the LANCE,
when ACON = 0 (active high ALE), ALE is not driven
high before it is tri-stated at the end of every bus mas-
tership period.
This difference will not cause any problems in designs
that set ACON = 1 (AS; active low ALE). It could cause
problems in designs in which ACON = 0. The ALE sig-
nal is intended to provide a strobe signal for an external
10. ALE Behavior
The LANCE device may drive the ALE pin LOW at the
end of each bus mastership period when ACON = 1
(ALE/AS active low—AS mode). When the bus master-
ship period ends, the ALE pin is tri-stated; hence, if
ALE is pulled HIGH by external logic, a glitch on ALE
results. The glitch occurs about when the LANCE de-
vice is releasing the bus by bringing HOLD high. The
address latch. The rising edge, coupled with a subse-
quent falling edge that will occur if the pin is externally
pulled down, will cause an invalid address to be
strobed into the external address latch. However, since
this occurs at the end of the bus mastership period, and
further master cycles are not performed by the
C-LANCE subsequent to the invalid address being
strobed (until the next bus mastership period), the in-
valid address generally has no effect. A design could
have problems with this if external logic is continuously
decoding the latched address and taking some action
on it even though the C-LANCE is not executing any
master cycles.
11. External Loopback on a Live Network
The LANCE device has an erratum that causes loop-
back failures when external loopback is run on a live
network. This erratum is fixed in the C-LANCE device.
12. Software Reset (STOP Bit) Handling
a. Latching of the STOP bit: In the LANCE device, writ-
b. Preservation of CSR1 and CSR2: The LANCE de-
60
ing the STOP bit in CSR0 causes all bus signals to
immediately float. With READY pulled up externally
(READY is open drain), this causes READY to deas-
sert prematurely during the Slave cycle. If DAS and
CS remain active, the LANCE device can errone-
ously start another Slave cycle. The C-LANCE de-
vice latches the STOP bit and, when it is set, allows
the Slave cycle in progress to complete before re-
setting the part.
vice does not preserve the contents of CSR1 and
CSR2 during the initialization process. Hence,
when the STOP bit is set, the contents of CSR1 and
CSR2 are not the same as they were before initial-
i z a t i o n a n d t h ey mu s t b e r ew r i t t e n b e fo r e
Am79C90
13. CSR0 Slave Read Data Stability
In the LANCE device, the status bit latches in CSR0
may change at any time, as governed by the occur-
rence of the external events they monitor. Hence, the
ERR, BABL, CERR, MISS, IDON, and INTR bits in
CSR0 may change during a Slave read cycle in which
they are being accessed. This can cause timing viola-
tions on the DAL lines. In the C-LANCE device, CSR0
is latched in a shadow register during a read so that
timing on the DAL lines is guaranteed.
14. INEA Bit Behavior
With the C-LANCE device, an INEA bit can be set in
CSR0 at any time, regardless of the state of the STOP
bit. This actually removes a restriction that was present
in the LANCE device, in which the INEA bit in CSR0
could be not be set while the STOP bit was set.
This difference between the two devices does not affect
normal device operation, but could disrupt diagnostic
code written for the LANCE device.
15. Effect of Setting the STOP Bit on CSR0 Bits
In the LANCE device, CSR0 is reset when the STOP bit
in CSR0 is set. This reset happens even if the STOP
bit was already set. When the reset occurs, all of the
other bits in CSR0 are cleared. In the C-LANCE, CSR0
is reset when the STOP bit is set in CSR0 only if the
STOP bit was not already set.
This difference between the two devices does not affect
normal device operation, but could disrupt diagnostic
code written for the LANCE device.
16. AC Specification Changes
The following differences in AC specification exist
between the C-LANCE and the LANCE.
#06 (t
#08 (t
#18 (t
#30 (t
#45 (t
re-initializing. This is not really a problem in the
LANCE device, but it can add extra instructions to
software. The C-LANCE device removes this soft-
ware burden by preserving the contents of CSR1
and CSR2 during initialization so that when the
STOP bit is set, they do not have to be reloaded be-
fore re-initializing. Note, however, that if the default
values of CSR3 (defaults for BCON, ACON, and
BSWP are 0, 0, and 0, respectively) are not used,
CSR3 must still be reloaded after setting the STOP
bit in the C-LANCE device, since CSR3 is cleared
when the STOP bit is set.
TEP
TDP
RDS
RDAS
RDYS
) maximum
) maximum
) minimum
) minimum
) minimum
C-LANCE
60 ns
60 ns
35 ns
40 ns
65 ns
LANCE
70 ns
70 ns
40 ns
50 ns
75 ns