AR# 72372

UltraScale GTH reads different adaptation loop codes between DMON output and IBERT

Description

When reading the adaptation loop codes from IBERT, I see different values to the expected values from DMONITOROUT.

Solution

For UltraScale GTH transceivers, some RX equalization adaptation loops are signed.

The Digital Monitor (DMON) section of the User Guide show the user how to read out the adaptation loop current state. 

For non-signed adaptation loops, the values read from the DMONITOROUT port are magnitude values. 

For signed adaptation loops, the values read from DMONITOROUT port are double-zero offset binary (Excess-K) values. 

See the attached spreadsheet for double-zero offset binary conversion tables. 


When the user reads adaptation loop codes using IBERT, the signed adaptation loops codes are post-processed to be shown as a two's complement value.

In IBERT, this is done to make debug easier.

For example, the OS adaptation loop is signed with a value range of -63 to +63. This range is represented by a 7-bit double-zero excess-64 (offset binary) value. 


If the user manually overrides the OS value to code 7b1000001 and then proceeds to read the DMON adaptation value through port DMONITOROUT[6:0], the port output would read 7b1000001.

However, if the user reads the OS adaptation value through IBERT using DFE_OS_LAST, the value would be 7b0000001.


The table below lists all signed adaptation values and their proper encoding for UltraScale GTH transceivers.

Note: UT in UltraScale GTH is an unsigned adaptation loop.

Important: this table applies for IBERT designs from Vivado 2018.3 and later.

Adaptation LoopRange
(Decimal)
Output
Bit Width
DMONITOROUT
Encoding
IBERT
Encoding
RXDFETAP3{-31,+31}6-bitDouble Zero, Excess-32Two's Complement
RXDFETAP4 {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAP5 {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAP6 {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAP7 {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAP8 {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAP9 {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAPA {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAPB {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAPC {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAPD {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAPE {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFETAPF {-15,+15}5-bitDouble Zero, Excess-16Two's Complement
RXDFEOS{-63,+63}7-bitDouble Zero, Excess-64Two's Complement
RXLPMOS{-63,+63}7-bitDouble Zero, Excess-64Two's Complement
 

Attachments

Associated Attachments

Name File Size File Type
US_GTH_RXEQ_DMON_vs_IBERT_Conversion_Tables.xlsx 45 KB XLSX

Linked Answer Records

Associated Answer Records

AR# 72372
Date 07/24/2019
Status Active
Type General Article
Devices