|
|
|
 |
|
 | |
 |
We are having problems with EEPROM communication. What is the likely cause?
The combination of the 82559/82559ER and a Fairchild/National* or some newer Catalyst* EEPROMS does not produce reliable results. Try to use Atmel* or ST* parts.
The datasheet in Chapter 3 on Signal Descriptions indicates the FLA[1]/AUXPWR should be pulled down if no auxiliary power is used but Chapter 4, section 4.3.1.4.1 (82559) and section 4.2.4.6 (82559ER) indicates this pin should be left floating if no auxiliary power is used. Which is correct?
This pin must be left unconnected if no auxiliary power is used. It may cause problems if it is pulled down externally. This was a datasheet errata which has been corrected in datasheet revision 2.2 for the 82559 and revision 1.3 for the 82559ER.
Section 6.1.1 of the 82559 data sheet specifies the crystal and the oscillator requirements for the PHY as a 25 MHz crystal or a 25 MHz oscillator used to drive the PHY unit's X1 and X2 pins. The PHY unit derives its internal transmit digital clocks from this crystal or oscillator input. The internal Transmit Clock signal is a derivative of the 25 MHz internal clock. It states the accuracy of the external crystal or oscillator must be "± 0.0005% (50 PPM)." However, 0.0005% is 5 PPM, not 50 PPM. Which is correct?
50 PPM is correct. This is a datasheet errata as of yet undocumented.
82559 Specification Change #6 - internal pull-up removed from ISOLATE# pin is missing from the 82559ER stepping document yet is the same stepping.
Therefore, it requires external pull-up as documented in the 82599 datasheet in Chapter 3.
What are the purposes of the Rbias10 and Rbias100 Resistors?
The bias10 and bias100 resistors determine the differential output voltage. This is also affected by the placement of the magnetics and RJ-45 connector and the distances between them. It is recommended that the distance between the 559 and the magnetics module, and the magnetics and the connector, be one inch or less but not longer than 4 inches. By increasing these distances you lose total transmit distance on the wire. In other words you may not be able to transmit the full 100 meters. Conversely if the output voltage is too high, it is less likely that the design will pass FCC emissions testing and more likely receiving units might ignore any signal with an amplitude greater than the TP-PMD specified maximum.
The values chosen in the reference drawing reflect the minimal distance of the component placement to get the required output voltage. The requirement for the output voltage is 950mV to 1050mV. A way to test for this is by performing the 100Base-TX Physical Layer Conformance test. A test kit can be requested through your field representative.
Note: This kit works in "pure" DOS environment only, i.e., not in a DOS window.
Magnetics transmit side center-tap cap is clearly indicated on AP-392 LOM to not be installed unless needed to pass FCC testing and clearly states it should be in the pico farad range.
However, numerous customers have installed a 0.1uF in the transmit side, as is shown for the receive side center tap.
This causes unpredictable behavior such as failure to auto-negotiate but works in forced modes, works on long cables but not short cables, will only work in full duplex mode, boots up in power down mode and etc.
How should the Flash Address pin FLA[6-2] on the 82559/82559ER be terminated when there is no Flash device?
Documentation doesn't specifically address termination when a Flash device is not used although AP-392 LOM design guide shows them un-terminated. Some customers have tied them high because they don't like floating pins. This can cause communication problems such as only forced 100Mbit operation will function and, also, can cause very erratic behavior. Leave these pins un-terminated (floating).
|
|
|
|
Intel® 82559 and 82559ER Controllers