US-12627535-B2 - Link equalization training method for mobile device

US12627535B2US 12627535 B2US12627535 B2US 12627535B2US-12627535-B2

Abstract

The disclosure provides a link equalization training method for a mobile device, comprising: setting to start-up link; performing a first power mode change to change a link rate to a first data rate; performing link equalization training on a receiver of a second node in response to a plurality of training sequences from a first node; when link equalization training in the receiver of the second node is not completed and there is no any error event, making a decision mechanism to perform link equalization training in the receiver of the second node; when link equalization training in the receiver of the second node is not completed and there is any error event, making the decision mechanism to reset for start-up link; and when link equalization training in the receiver of the second node is completed, setting link-up in the first data rate for data transfer.

Inventors

Kuan-Chou LEE
Mau-Lin Wu
Tai-Lai Tung
Wei-Cheng Tang
Li-Hung Chiueh
Min-Chau Jan
Ko-Yin Lai
Chih-Kang Hsu
Chang-Chen CHU
Ting-Wei Lai
JYUN-HAO HUANG
Hsuan-Jung Hsu
Yung-Chih Lin

Assignees

MEDIATEK INC.

Dates

Publication Date: 20260512
Application Date: 20240618

Claims (17)

1 . A link equalization training method for a mobile device, comprising: setting to start-up link; performing a first power mode change to change a link rate to a first data rate; performing link equalization training on a receiver of a second node in response to a plurality of training sequences from a first node; when link equalization training in the receiver of the second node is not completed and there is no error event, making a decision mechanism for controlling continuation or reset of link equalization training in the receiver of the second node; when link equalization training in the receiver of the second node is not completed and there is any error event, making the decision mechanism to reset for start-up link; and when link equalization training in the receiver of the second node is completed, setting link-up in the first data rate for data transfer.
2 . The method according to claim 1 , further comprising: performing a second power mode change for changing the link rate to a second data rate lower than the first data rate, the second power mode change is performed before the first power mode change.
3 . The method according to claim 1 , wherein link equalization training includes a first phase, a second phase and a third phase; in the first phase, the first node transmits a plurality of first training sequences to the second node by using a preset gear setting of a transmitter of the first node, the first training sequences including a plurality of 2-level Pulse-amplitude modulation (PAM)-N signals, the second node performing link equalization training based on the plurality of first training sequences; in the second phase, the first node transmits a plurality of second training sequences to the second node by using the preset gear setting of the transmitter of the first node, the second training sequences including a plurality of all-level PAM-N signals, the second node performing link equalization training based on the plurality of second training sequences; between the second phase and the third phase, the transmitter of the first node enters into a stall mode to change the preset gear setting of the transmitter of the first node; in the third phase, a transmitter of the second node sends receiver request preset information in an information exchange frame to the first node, the receiver request preset information is used to inform the first node about an optimized gear setting selected by the receiver of the second node after link equalization training, the information exchange frame in the third phase including a plurality of 2-level PAM-N signals; and after the third phase, the transmitter of the first node changes the preset gear setting based on the optimized gear setting selected by the receiver of the second node.
4 . The method according to claim 3 , wherein in link equalization training, in the first phase, the transmitter of the first node sends the plurality of first training sequences including the plurality of 2-level Pulse-amplitude modulation (PAM)-N signals to the receiver of the second node to train link equalization of the receiver of the second node; in the second phase, the transmitter of the first node sends the plurality of second training sequences including the plurality of all-level PAM-N signals to the receiver of the second node to train link equalization of the receiver of the second node; in a first round of link equalization training, a default preset gear setting is used, in a second or later rounds of link equalization training, the preset gear setting is changed by the receiver of the second node; the transmitter of the first node changes a preset Feed-Forward Equalizer (FFE) gear setting of the transmitter of the first node in the stall mode; in the third phase, the first node and the second node exchange information about optimized gear setting selected by the receiver of the second node after link equalization training, wherein in the third phase, the default preset gear setting is set in the second phase, and in the third phase, in the first or later rounds of link equalization training, the default preset gear setting is used, and the preset gear setting in the first phase and the second phase of the next round is set in the third phase; when link equalization training for all active lanes are not done, the transmitter of the first node changes the preset gear setting based on the optimized gear setting selected by the receiver of the second node in the stall mode.
5 . The method according to claim 4 , wherein in link equalization training, in the first phase, the receiver of the second node receives the first training sequences from the transmitter of the first node to train link equalization of the receiver of the second node; in the second phase, the receiver of the second node receives the second training sequences from the transmitter of the first node to train link equalization of the receiver of the second node; the receiver of the second node changes equalization setting in the stall mode; in the third phase, the first node and the second node exchange information about the optimized gear setting selected by the receiver of the second node after link equalization training; when link equalization training for all active lanes are not done, the receiver of the second node changes equalization setting in the stall mode.
6 . The method according to claim 5 , wherein an information exchange frame includes a first indicator for detecting equalization information frame, a second indicator for indicating receiver link equalization training interruption request if any error event occurs, and a third indicator for indicating whether receiver link equalization training is done or not.
7 . The method according to claim 6 , wherein before both the transmitter of the first node and the receiver of the second node enter burst state in the third phase, in a first part of the third phase, whether both the transmitter of the first node and the receiver of the second node enter into burst state is determined; when both the transmitter of the first node and the receiver of the second node enter into burst state, the method returns to the step of performing the first power mode change; when not both the transmitter of the first node and the receiver of the second node enter into burst state, the transmitter of the first node sends the information exchange frame to the receiver of the second node, and the receiver of the second node ignores any cyclic redundancy check (CRC) errors.
8 . The method according to claim 7 , wherein after both the transmitter of the first node and the receiver of the second node enter burst state in the third phase, in a second part of the third phase, the transmitter of the first node sends the information exchange frame to the receiver of the second node; the receiver of the first node determines that whether an error counter is higher than an error counter threshold or not, wherein when an error is found by the receiver of the second node, the error counter is updated, and when the error counter reaches the error counter threshold, an error event occurs and the error counter is freeze; when the error counter is higher than the error counter threshold, the transmitter of the first node sends the information exchange frame to the receiver of the second node to indicate that the transmitter of the first node acknowledges a receiver link equalization training interruption request; when the transmitter of the first node sends the information exchange frame to the receiver of the second node in k, k being a natural number, times or more than k times, the transmitter of the first node exits the third phase; when the error counter is not higher than the error counter threshold and the receiver of the second node receives the information exchange frame m times, m being a natural number, the transmitter of the first node sends the information exchange frame to the receiver of the second node to indicate detection of equalization training; when the transmitter of the first node sends the information exchange frame k times, the receiver of the first node determines that whether the error counter is higher than the error counter threshold or not; when the error counter is higher than the error counter threshold, the transmitter of the first node sends the information exchange frame to the receiver of the second node to indicate that the transmitter of the first node acknowledges receiver link equalization interruption request, and when the transmitter of the first node sends the information exchange frame to the receiver of the second node in k times or more than k times, the transmitter of the first node exits the third phase; when the error counter is not higher than the error counter threshold and when the receiver of the first node receives the information exchange frame in m times, the transmitter of the first node exits the third phase.
9 . The method according to claim 8 , wherein in a normal case after both the transmitter of the first node and the receiver of the second node enter into burst state in the third phase, the transmitter of the first node and the transmitter of the second node send the information exchange frame to each other until the receivers of the first and the second nodes trigger; the receivers of the first and the second nodes receive once the information exchange frame without any CRC error; the first indicator of the information exchange frame is changed from a first value to a second value and the receivers of the first and the second nodes trigger the transmitters of the first and the second nodes to send the information exchange frame to the receivers of the respective other nodes; the transmitters of the first and the second nodes send the information exchange frame to each other at least k times; when the receivers of the first and the second nodes receive the information exchange frame one time, and do not experience at least k consecutive CRC errors, the link equalization training is not interrupted and the transmitters of the first and the second nodes exit the third phase; after the transmitters of the first and the second nodes send the information exchange frame at least k times, the link equalization training is interrupted, and the transmitters of the first and the second nodes exit the third phase; and after a link equalization training timer is timeout, the link equalization training is interrupted, and the transmitters of the first and the second nodes exit the third phase.
10 . The method according to claim 9 , wherein when the receiver of the first node receives once the information exchange frame, the receiver of the first node determines that the receiver of the second node experiences at least k consecutive CRC errors, the link equalization training is interrupted; and when the receiver of the first node experiences at least k consecutive CRC errors, the transmitter of the first node sends the information exchange frame to the second node and the link equalization training is interrupted.
11 . The method according to claim 10 , wherein when the receiver of the first node receives end of burst (EOB), the receiver of the first node exits the third phase.
12 . The method according to claim 11 , wherein when all active lanes on the same node have link equalization training done, the transmitter of the first node sends the information exchange frame for indicating link equalization training completed; and when the transmitter of the first node only sends the information exchange frame and the receiver of the first node only receives the information exchange frame, link equalization training on all active lanes on both nodes is done.
13 . The method according to claim 12 , wherein the decision mechanism includes: a decision flow for timeout; a decision flow for link equalization training interruption; a decision flow for optimal transmitter equalization update; and a decision flow for link equalization training.
14 . The method according to claim 13 , wherein in the decision flow for timeout, reading information from the information exchange frame and writing into a plurality of registers respectively for all active lanes; when a link equalization training timer is not timeout, proceeding to the decision flow for link equalization training interruption; when the link equalization training timer is timeout, determining whether timeout of the link equalization training timer is first time; when timeout of the link equalization training timer is first time, an initial transmitter equalization gear is changed, the transmitter equalization gear in the first phase and the second phase is unchanged, the link equalization training is retried, and device management entity (DME) is reset; when timeout of the link equalization training timer is not first time, in response to a timeout retry counter is equal to or higher than a maximum timeout retry counter, the decision mechanism decides that the link rate is downgraded; the link rate does not ever go to the first data rate or above again; and DME is reset; and when timeout of the link equalization training timer is not first time, in response to the timeout retry counter is lower than the maximum timeout retry counter, the decision mechanism determines that the initial transmitter equalization gear is unchanged, the transmitter equalization gear in the first phase, the second phase and the third phase is not changed; and the link equalization training is retried.
15 . The method according to claim 14 , wherein in the decision flow for link equalization training interruption, a link equalization training timeout retry counter is reset; when link equalization training interruption occurs, the flow proceeds to the decision flow for optimal transmitter equalization update; when link equalization training interruption does not occur and when a retry upper limit is not reached, the transmitter equalization gear is not changed in the first phase to the third phase, the first phase to the third phase are retried; and when link equalization training interruption does not occur and when the retry upper limit is reached, the link rate is downgraded, the link rate does not ever go to the first data rate or above again, and DME is reset.
16 . The method according to claim 15 , wherein in the decision flow for optimal transmitter equalization update, a software interrupt retry counter is reset, and information about an optimal transmitter equalization gear is read; when a current performance is better than a previous performance, recording a current transmitter equalization gear, and setting a local optimal receiver equalization gear; and when the current performance is not better than the previous performance, setting the local optimal receiver equalization gear.
17 . The method according to claim 16 , wherein in the decision flow for link equalization training, when a bit error rate of transmitter equalization gear for all local lanes is lower than a bit error rate threshold, determining a link equalization training mode among a first mode and a second mode, in the first mode, once a first transmitter equalization gear for a lane is found, the lane reports link equalization training done and does not change the required transmitter equalization gear until all of the active lanes report link equalization training done, in the second mode, all available transmitter equalization gears are tried to find an optimal one for transmission; when the first mode is selected, a local device reports link equalization training done and in the third phase of the next round, the local device sends the information exchange frame; in case the second mode is selected, when not all transmitter equalization gears are trained, the local device requests a peer device to set a new transmitter equalization gear, and in the third phase of the next round, the local device sends the information exchange frame, and the flow returns to link equalization train, and when all transmitter equalization gears are trained, the local device reports link equalization training done and in the third phase of the next round, the local device sends the information exchange frame; in case that not the bit error rate of transmitter equalization gear for all local lanes is lower than the bit error rate threshold, when all transmitter equalization gears are trained, resetting DME and the flow returns to start-up link setting, and when not all transmitter equalization gears are trained, the local device requests the peer device to set the new transmitter equalization gear, and in the third phase of the next round, the local device sends the information exchange frame, and the flow returns to link equalization train.

Description

CROSS-REFERENCE TO RELATED ART This application claims the benefit of U.S. provisional application Ser. No. 63/555,919, filed 2024 Feb. 21, the disclosure of which is incorporated by reference herein in its entirety. TECHNICAL FIELD The disclosure relates in general to a link equalization training method for a mobile device. BACKGROUND Mobile devices have become indispensable tools in people's lives, providing convenience, connectivity, and access to a wide range of services and information. Mobile devices have become essential in people's lives for several reasons. (1) Communication: Mobile devices enable instant communication through calls, texts, emails, and social media apps, keeping people connected with friends, family, and colleagues regardless of location. (2) Information Access: With mobile devices, people have access to vast amounts of information at their fingertips. They can browse the internet, access news, research topics of interest, and find answers to questions quickly and conveniently. (3) Productivity: Mobile devices allow people to stay productive on the go. They can manage tasks, appointments, and emails, as well as access documents and files through various productivity apps and cloud services. (4) Entertainment: Mobile devices serve as entertainment hubs, providing access to music, videos, games, e-books, and streaming services. They offer a wide range of entertainment options to help people relax and unwind. (5) Navigation: Mobile devices with GPS capabilities provide navigation assistance, helping people find directions, locate businesses, and explore new places without getting lost. (6) Social Connection: Mobile devices facilitate social interactions through social media platforms, messaging apps, and video calls, allowing people to connect with others, share experiences, and maintain relationships. Mobile Industry Processor Interface (MIPI) specifications are widely used across the Mobile and IoT industries, mainly for applications like cameras, sensors, modems, storage, audio, displays and other peripherals. In the last few years, the automotive industry has also started to adopt many of these protocols. Growing adoptability of these protocols to these different markets is driving quicker enhancements to these MIPI specifications. MIPI UniPro (Unified Protocol) is one such protocol which is seeing growing adoptability. UniPro is a high-speed interface. UniPro is aimed at providing high-speed data transmission (Gigabits) with minimal interface pins and low power consumption. The purpose of UniPro development is to support various Applications, Interfaces, Devices, Bandwidth by using a universal protocol. UniPro exhibits excellent scalability, with a maximum support of 128 devices. The UniPro protocol stack encompasses layers L1 to L4 of the Open System Interconnection Reference Model (OSI). Additionally, a PHY adapter layer (PA layer) (L1.5) is introduced between layers M-PHY layer (L1) and Data link layer (L2) to connect to PHY Interface. In the context of use with UniPro, M-PHY (Physical Layer; L1) is used in a dual-simplex, high bandwidth serial link that employs a low-swing, differential signaling technique for high-speed communication. Unidirectional PHY Links are not supported because various higher protocol layers require return information, e.g., for flow control. The PHY Adapter Layer (Physical Adapter Layer, PA Layer; L1.5) is responsible for abstracting the details of the PHY technology, thus providing a PHY-independent interface to higher protocol layers. PHY Adapter Layer supports multi-lane, controls power modes for M-PHY, and provides symbol encoding, and automatic capability discovery for M-PHY. The main responsibilities of the Data Link Layer (DL Layer; L2) are to provide reliable links between a transmitter and a directly attached receiver and to multiplex and arbitrate multiple types of data traffic, e.g., priorities. Data Link Layer assembles data and control frames from higher layers, provides retransmission on errors, controls the priorities and support flow control. Network Layers (L3) has a purpose to allow data to be routed to the proper destination in a networked environment. Transport Layer (L4) is the highest UniPro protocol layer involved in the transportation of data; and thus provides the data service interface that is used by hardware or software using UniPro. Unlike lower protocol layers, the Transport Layer tends to concentrate on relatively abstract mechanisms. These mechanisms allow a single physical Packet stream between two Devices to support multiple, independent, logical Packet streams or “Connections”. Device Management Entity (DME) controls the layers in the UniPro stack; and provides access to various control parameters in all layers, manages the power mode transitions of the Link, and handles boot-up, hibernate and reset of the entire UniPro stack. M-PHY is a high speed data communications physical layer protocol standard targeted at the