A DFT test redirection architecture and method for the entire process of chiplet binding
By introducing a core detector, TAP controller, and SIB test network into the core system, test vector redirection for 2.5D and 3D core systems was achieved, solving the problem of low test efficiency during core bonding and realizing efficient and low-cost fault detection and test path redirection.
Patent Information
- Application Number
- CN202410840693.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-27
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-06-27
AI Technical Summary
In 2.5D and 3D core systems, existing technologies struggle to effectively redirect test vectors, resulting in low testing efficiency and high costs, especially as faults and damage cannot be accurately detected during core bonding.
A DFT test redirection architecture is adopted, including a chip detector, a TAP controller, a pad port, and a SIB test network. Automatic redirection and detection of test data paths are achieved through multiplexing logic and daisy chaining. Combined with IEEE 1149.1 and IEEE 1687 gateway registers, flexible testing of chip integrated systems is realized.
It enables real-time detection of damage and faults during the core bonding process, improving testing efficiency, reducing production costs, and supporting flexible testing of multi-layer core systems and efficient access to core/embedded instruments.
Smart Images

Figure CN118412294B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of DFT testing technology, and in particular to a DFT testing redirection architecture and method for a whole process of core particle binding. Background Art
[0002] DFT (Design For Test) is to add circuits other than functional logic to the chip design, so that the chip can be quickly tested and distinguished after production. In the more demanding chip integration system, further DFT design is needed to accurately locate faults and discover defects, and to easily input test vectors for chip system testing to reduce the difficulty of product testing, improve testing efficiency, and reduce testing costs.
[0003] As digital systems become more complex, the scale of testing also increases. Automatic Test Pattern Generation (ATPG) has become an indispensable part of the DFT implementation process. ATPG uses programs and algorithms to automatically generate test vectors for certain forms of circuits and fault models. ATPG can correctly generate test vectors with high fault coverage, thereby ensuring chip yield and improving chip quality.
[0004] In today's mainstream hierarchical chiplet integrated system architecture, the design to be tested is often divided into different levels such as core, subsystem and soc, and a step-by-step method is used to gradually complete the test of the entire chip. ATPG will generate test vectors corresponding to different levels and sub-modules. When generating test vectors for sub-modules, the entry of vector data is the input end of the scan chain inside the core. For 2.5D and 3D chiplet stacking systems that are well packaged by connecting the interfaces of heterogeneous chiplets through TSV binding, during actual testing, vector data is input from the pad port of the bottom-level chiplet. The logic from the input pad port to the internal ports of other chiplet modules / cores is different. Handwritten vectors or test vectors generated by ATPG cannot be used directly, and test vector redirection is required. Therefore, how to successfully reuse the test vectors generated by 2.5D and 3D chiplet systems before binding during binding and after binding has become an urgent problem to be solved. Summary of the invention
[0005] The purpose of the present invention is to provide a DFT test redirection architecture and method for the entire process of chiplet binding, so as to provide an effective, low-cost, low-test-resource-occupied, low-area-overhead vector redirection test architecture and method for the chiplet integration system before, during and after binding.
[0006] In order to solve the above technical problems, the present invention provides a DFT test redirection architecture for the whole process of chiplet binding, including a chiplet integration system composed of a 2.5D chiplet system or a 3D chiplet system, wherein each single chiplet in the chiplet integration system has an architecture including:
[0007] A core particle detector, comprising an upper core particle detector and a lower core particle detector; the upper core particle detector and the lower core particle detector of each single core particle receive and output a first logic signal driven by the lower core particle detector and the upper core particle detector of the adjacent single core particle, and output a second logic signal if the first logic signal is not received; the multiplexing logic is controlled according to the output signal value of the core particle detector, that is, the multiplexing logic selects a test data path of an input signal from a pad port or TSV according to the output signal value of the core particle detector; for a single core particle located at the bottom layer, the test data path of the input signal from the pad port is selected; for a single core particle located at the top layer during binding and located at the middle layer or the top layer after binding, the test data path of the input signal from the TSV is selected;
[0008] A TAP controller is used to generate a test signal, determine the position of a single chip in a chip integrated system according to the output signal value of a chip detector, wherein the position includes a top layer, a middle layer and a bottom layer; and control a scan test link; wherein the scan test link includes a serial chain of single chips before binding, a daisy chain composed of single chips during binding and a daisy chain composed of single chips after binding; wherein the single chip before binding is located at the bottom layer;
[0009] The pad port is connected to the TAP controller as the test access port of each single chip before binding. The signal lines of the test access port include: a test data input port TDI, a test data output port TDO, a test mode selection port TMS, a test clock port TCK and an optional test reset port TRST;
[0010] The SIB test network is connected and controlled by a TAP controller, and is used to test the core and / or embedded instrument in each single chip; wherein the SIB test network includes a plurality of SIB circuit modules; the state of each SIB circuit module includes standby, on and off; when the SIB circuit module is not used, it is set to the standby state; when the SIB circuit module is used to test the mounted TDR and the core and / or embedded instrument in the single chip, it is set to the on state; when the TDR mounted by the SIB circuit module and the core and / or embedded instrument in the single chip do not need to be tested and must pass through the SIB circuit module, it is set to the off state, and at this time the SIB circuit module plays a bypass function.
[0011] Preferably, after the test is completed, the test response is output from the pad port of the single chip located at the bottom layer.
[0012] Preferably, the TAP controller includes: an FSM circuit, a core locator, an instruction register, an instruction decoding circuit and a data register; the FSM circuit is a synchronous 16-bit finite state machine, which receives a first signal from a pad port, converts it into a required test mode and outputs it, maintains synchronization between all components on the scan test link, and allows the movement, capture and update of test data; the first signal from the pad port includes a TMS signal and a TCK signal; the instruction register and the data register receive the output of the FSM circuit and the TDI signal from the pad port, the instruction register outputs the instruction to the instruction decoding circuit, the instruction decoding circuit translates the instruction into a corresponding test mode enable signal and outputs it to the data register, and the data register is used to perform different test functions; the core locator is used to optimize the data path when the test data is bypassed, so as to receive the output signal of the core detector to autonomously determine the core position and control the output multiplexer.
[0013] Preferably, the data registers include a bypass register, a boundary scan register, an IDCODE register and a P1687 gateway register.
[0014] Preferably, each of the SIB circuit modules comprises: AND gate 1, AND gate 2, multiplexer 1, multiplexer 2, multiplexer 3, UpSIB register and ShSIB register; wherein the two input ends of the AND gate 1 are connected to the Update_DR port and the Select_DR port, and the two input ends of the AND gate 2 are connected to the Select_DR port, the Shift_DR port and the optional Capture_DR port; the output end of the AND gate 1 is connected to the input end 1 of the multiplexer 1, the input end 2 of the multiplexer 1 is connected to the output end of the ShSIB register and the input end 3 of the multiplexer 2, the input end 3 of the multiplexer 1 is connected to the output end of the UpSIB register and the input end 3 of the multiplexer 3, and the output end of the UpSIB register is connected to the input end 1 of the multiplexer 1. The output end is connected to TDR through the Seli signal, and the output end of multiplexer one is connected to input end one of the UpSIB register; the output end of AND gate two is connected to input end one of multiplexer two, input end two of multiplexer two is connected to TDI, the output end of multiplexer two is connected to input end one of the ShSIB register, input end two of the ShSIB register is connected to input two of the UpSIB register and the Reset signal, input end three of the ShSIB register is connected to the TCK signal and input three of the UpSIB register, the output end of the ShSIB register is connected to input end two of multiplexer three, and is connected to TDR through the TSI signal, TDR is connected to input end one of multiplexer three through the FSO signal, and the output end of multiplexer three is connected to TDO.
[0015] Preferably, the SIB test network is a single-layer SIB test network, and a plurality of the SIB circuit modules are linked into one link.
[0016] Preferably, the SIB test network is a multi-layer SIB test network, and the back end of at least one SIB circuit module of the first layer continues to be connected to at least one SIB circuit module, and the SIB circuit module of the first layer is connected to the TAP controller; wherein the SIB circuit module of the first layer, also called the Gateway register, serves as the interaction area between the TAP controller and the SIB test network, and controls the on and off of the entire SIB test network; the Gateway register is added between TDI and TDO, and is selected when its related instructions are loaded into the instruction register.
[0017] Preferably, the first logic signal is 1, and the second logic signal is 0; each upper core particle detector drives a logic 1 to the lower core particle detector in the upper core particle, and each lower core particle detector drives a logic 1 to the upper core particle detector in the lower core particle.
[0018] Preferably, judging the position of a single core particle in the core particle integration system according to the output signal value of the core particle detector specifically includes:
[0019] When the output of the lower core particle detector of a single core particle is 0 and the output of the upper core particle detector is 0, the single core particle is located at the bottom layer of the core particle integration system before binding;
[0020] When the output of the lower core particle detector of a single core particle is 1 and the output of the upper core particle detector is 0, the single core particle is located at the top layer of the core particle integration system during or after binding;
[0021] When the output of the lower core particle detector of a single core particle is 0 and the output of the upper core particle detector is 1, the single core particle is located at the bottom layer of the bound core particle integration system;
[0022] When the output of the lower layer core particle detector of a single core particle is 1 and the output of the upper layer core particle detector is 1, the single core particle is located in the middle layer of the bound core particle integration system.
[0023] The present invention also provides a DFT test redirection method for the whole process of chiplet binding, which adopts the DFT test redirection architecture for the whole process of chiplet binding as described above, and includes the following steps:
[0024] Before the chiplet integration system is bound, the TAP controller determines that the single chiplet is located at the bottom layer according to the output signal value of the chiplet detector, generates a test signal, controls the scan test link, selects the test data path of the input signal from the pad port, and tests the single chiplet before binding;
[0025] In the chiplet integrated system binding, if the TAP controller determines that the single chiplet is located at the bottom layer according to the output signal value of the chiplet detector, the test data path of the input signal from the pad port is selected; if the single chiplet is determined to be located at the top layer, the test data path of the input signal from the TSV is selected, and the top layer chiplet and the bottom layer chiplet in the binding are tested according to the selected test data path;
[0026] After the chiplet integration system is bound, if the TAP controller determines that the single chiplet is located at the bottom layer based on the output signal value of the chiplet detector, the test data path of the input signal from the pad port is selected; if the single chiplet is determined to be located at the top layer or the middle layer, the test data path of the input signal from the TSV is selected, and the top layer chiplet, the middle layer chiplet and the bottom layer chiplet in the binding are tested according to the selected test data path.
[0027] Compared with the prior art, the present invention has the following beneficial effects:
[0028] (1) The present invention adds modules such as a TAP controller and a chip detector to each single chip. During binding, the test vector can be redirected to the corresponding middle-layer chip through the multiplexing logic at the pad port of the bottom-layer chip to test the KGD damaged during the 2.5D or 3D stacking process; after the chip integration system is fully bound, the top-layer chip can be further tested through a daisy chain; the DFT test architecture of the present invention can be used for all binding levels without the need for configuration steps, and can accurately detect chiplets that are damaged during the binding process. It can also detect chiplets that fail during the binding process in real time to reduce production costs and improve test efficiency.
[0029] (2) The present invention introduces upper and lower chip detector modules and multiplexing logic inside each single chip, allowing the position of the chip in the system to be automatically detected so as to select the test data path of the entire chip stack; during the test process, only the JTAG port signal is driven to the upper chip, and multiple chips share the control signal and are connected in series to form a daisy chain, which can significantly reduce the test time and improve the test efficiency while occupying a small area and a small number of test ports.
[0030] (3) The present invention introduces a TAP controller inside each single chip, which can output a corresponding enable signal under a corresponding state, perform corresponding operations on the data register and the instruction register, and realize the control of the instruction input scan test link and the test data input scan test link; users can also add test instructions to generate target test enable signals and realize a variety of different test functions.
[0031] (4) The present invention embeds a SIB test network behind the TAP controller. The SIB test network can be multi-layered and can freely select the test scan chain path and the access to the embedded instrument through only one instruction. It can test some cores or embedded instruments in a 2.5D or 3D stacked core system simultaneously or separately. It provides an efficient and flexible data reading and control mechanism while saving internal wiring resources and test resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Figure 1 FIG. 4 is a diagram of a DFT test redirection architecture of a 2.5D chip integration system according to an embodiment of the present invention.
[0033] Figure 2 FIG. 4 is a diagram of a DFT test redirection architecture of a 3D chip integration system according to an embodiment of the present invention.
[0034] Figure 3 Schematic diagram of the TAP controller structure according to an embodiment of the present invention.
[0035] Figure 4 Schematic diagram of the structure of a core particle detector according to an embodiment of the present invention.
[0036] Figure 5 Schematic diagram of a SIB circuit module according to an embodiment of the present invention.
[0037] Figure 6 For the embodiment of the present invention Figure 5 Schematic diagram of the TAP controller controlling the transition between three states; (a) standby state; (b) on state; (c) off state.
[0038] Figure 7 FIG. 4 is a schematic diagram of a multi-layer SIB circuit structure according to an embodiment of the present invention. DETAILED DESCRIPTION
[0039] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description. It should be noted that the accompanying drawings are in very simplified form and in non-precise proportions, and are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention.
[0040] like Figure 1 The figure shows the DFT test redirection architecture of the 2.5D chip integration system. Figure 2 FIG. 1 is a schematic diagram of a DFT test redirection architecture of a 3D chiplet integration system. The embodiment of the present invention takes these two chiplet integration systems as examples to illustrate the test redirection architecture of the present invention.
[0041] Each single chip in the 2.5D chip integration system or the 3D chip integration system is configured with a pad port, a TAP controller, a chip detector, and a SIB test network.
[0042] The pad port is used as the test access port for each single chip before bonding. The signal lines of the test access port include: test data input port TDI, test data output port TDO, test mode selection port TMS, test clock port TCK and optional test reset port TRST. Among them, TMS is the only key control line; TDI and TDO provide input and output of test data; TCK provides test clock; and TRST is optional and can be omitted when the pins are limited.
[0043] like Figure 3As shown, the TAP controller is a test control circuit, which is responsible for generating test signals such as instruction signals, enable signals, and test control signals, and is composed of an FSM circuit, a data register, a core locator, an instruction register, and an instruction decoding circuit. The TAP controller uses data registers to perform different functional tests, wherein the data registers include bypass registers, IDCODE registers, boundary scan registers, etc., and the instruction registers are used to control the data registers. More specifically, the FSM circuit is controlled by external port signals TMS, TCK, and TRST, and the FSM outputs SHIFTDR / SHIFTIR, CLOCKIR / CLOCKDR, UPDATEIR / UPDATEDR, and SELECT signals. Among them, the IR series signals control the shift and update of the instruction register, and the DR series signals control the shift and update of the data register. The instruction register selects and controls the data register through the instruction decoding circuit, and only one of the multiple registers in the data register is selected at the same time. The input port TDI is used for inputting instructions and test data. The chip locator controlled by DETECTOR selects the signal input control of TDI and FROM_HOST. When this part is not needed, TDI is directly output to TDO. At this time, there is no data input and output on the HOST and FROM_HOST ports.
[0044] The FSM (Finite-state Machine) circuit is a synchronous 16-bit finite state machine that responds to changes in the TMS and TCK signals from the TAP port, controls the behavior of the test logic, converts the required test mode by receiving signals such as TMS and TCK, and maintains synchronization between all components on the scan test link to allow the movement, capture, and update of test data.
[0045] The instruction register and instruction decoding circuit allow the instruction data to be continuously transferred to the test logic during the instruction register scanning cycle. The instruction register can define the number of register units according to the length of the instruction, and load the instruction into the update register during the finite state machine update mode and output it to the instruction decoding circuit. The instruction decoding circuit is composed of a large number of combinational logic and registers, and its function is to translate the user-defined instruction code into a specific test mode enable signal to control the test circuit. The instruction register can be pre-loaded with instructions, or the instruction encoding can be customized according to the user's target. In the embodiment of the present invention, each instruction is encoded, and the BYPASS instruction, HOSTIJTAG instruction and other instructions are mandatory. As shown in Table 1, the internal test instruction set and its description of the instruction register in the TAP controller.
[0046] Table 1: Internal test instruction set and description of instruction registers in TAP controller
[0047]
[0048] The core locator is used to optimize the data path when the test data is bypassed. It receives the output signal of the core detector to independently determine the core position and control the output multiplexer.
[0049] like Figure 4 As shown in the figure, the core detector is divided into an upper core detector and a lower core detector. The principle is to use combinatorial logic to detect the presence of adjacent cores (bottom or top). Each upper core detector drives a logic "1" to the lower core detector in the upper core, and each lower core detector drives a logic "1" to the upper core detector in the lower core. If the signal receiving end of each core detector does not receive the driving value of other core detectors, it will default to a high impedance state "Z". At this time, the core detector will drive a logic "0" at its output, indicating that there is no adjacent core. The two core detectors used can accurately determine the position of the core in the core stacking system: bottom, middle or top. In the case of many intermediate cores in the stack, that is, in an integrated system with more than 3 cores, their configuration is the same regardless of their position in the stack. After the cores are bound, the output signal value of the internal core detector controls the multiplexing logic to select the test data path from the pad port or TSV. Tables 2 and 3 show the truth table of the core particle detector detection mechanism and the control multiplexing logic configuration table respectively.
[0050] Table 2: Truth table of the core particle detector detection mechanism
[0051]
[0052] Table 3: Control multiplexing logic configuration table of the core particle detector detection mechanism
[0053]
[0054] The multiplexing logic realizes the scan path selection between the input and output pads ports and the TSVs between the chiplets through the control of the chiplet detector, which can improve the reconfigurability of the test scan chain.
[0055] The SIB test network is an embedded test network structure that can flexibly test the core and embedded instruments in the chip. By establishing a flexible and variable data transmission path between the internal core and the embedded instrument, the data transmission and testing of the internal embedded instrument and the logic core are completed, providing support for the debugging, monitoring and testing of devices such as on-chip embedded instruments.
[0056] SIB circuit modules such as Figure 5As shown, each of the SIB circuit modules includes: AND gate 1, AND gate 2, multiplexer 1, multiplexer 2, multiplexer 3, UpSIB register and ShSIB register; wherein the two input ends of the AND gate 1 are connected to the Update_DR port and the Select_DR port, and the two input ends of the AND gate 2 are connected to the Select_DR port, the Shift_DR port and the optional Capture_DR port; the output end of the AND gate 1 is connected to the input end 1 of the multiplexer 1, the input end 2 of the multiplexer 1 is connected to the output end of the ShSIB register and the input end 3 of the multiplexer 2, the input end 3 of the multiplexer 1 is connected to the output end of the UpSIB register and the input end 3 of the multiplexer 3, and the output end of the UpSIB register is connected to the input end 3 of the multiplexer 3. The output end is connected to TDR through the Seli signal, and the output end of multiplexer one is connected to input end one of the UpSIB register; the output end of AND gate two is connected to input end one of multiplexer two, input end two of multiplexer two is connected to TDI, the output end of multiplexer two is connected to input end one of the ShSIB register, input end two of the ShSIB register is connected to input two of the UpSIB register and the Reset signal, input end three of the ShSIB register is connected to the TCK signal and input three of the UpSIB register, the output end of the ShSIB register is connected to input end two of multiplexer three, and is connected to TDR through the TSI signal, TDR is connected to input end one of multiplexer three through the FSO signal, and the output end of multiplexer three is connected to TDO. That is: AND gate 1 controlled by external port update_DR and select_DR, AND gate 2 controlled by select_DR and shift_DR (capture_DR is optional), two multiplexers 1 and 2 controlled by two AND gates 1 and 2, multiplexer 1 and multiplexer 2 as inputs of two UpSIB registers and ShSIB registers, the ShSIB register transmits data to TDR through the TSI signal, TDR returns data to output multiplexer 3 through the FSO signal and then outputs it by TDO, the UpSIB register is used to maintain the switching state of the SIB circuit module and control the output multiplexer 3 and TDR, and the external signals Reset and TCK control the reset and drive of the two UpSIB registers and ShSIB registers.
[0057] The SIB circuit module is controlled by the TAP controller to realize the transition between three states: standby, on and off. When the SIB circuit module is not used, it is in the standby state, such as Figure 6 As shown in (a); if the TDR and the post-instrument mounted after the SIB circuit module need to be tested, they are in the open state, such as Figure 6As shown in (b), the selector of the output path of the SIB circuit module will select the TDR path; when there is no need to test the SIB post-instrument and it must pass through the SIB circuit module, it is in the closed state, such as Figure 6 As shown in (c), the SIB circuit module plays a bypass function at this time. The TAP controller outputs the Sel selection signal, Rst reset signal, ShD shift signal and UpD update signal to control the SIB circuit module. These signals are shared by all instruments and SIB circuit modules in the entire test network and are sent to each selected SIB circuit module register under the control of the selection signal. TDR is a data shift register, which exchanges data with the embedded instrument mounted after the SIB circuit module through TSI and FSO signals.
[0058] The essence of the SIB test network is to determine whether the back-end circuit is connected to the test link. The SIB circuit module itself can be connected in series in the form of a daisy chain to form a SIB link. All structures mounted by the SIB circuit module are connected to the link composed of the SIB circuit module in this way. According to its hardware structure, two network structures with SIB link as the core are proposed: one is to directly connect SIBs to form a link, which is called a SIB single-layer network structure. Figure 7 When only the first layer of SIB circuit modules exists, it is called a single-layer SIB network. The other is to continue to connect SIB links at the back end of each SIB circuit module in the SIB link, and so on to form a nested SIB link, which is called a SIB multi-layer network structure, such as Figure 7 As shown, the JTAG port is input into the TAP controller, and the TAP controller outputs TCK, ShD, UpD, Sel and Rst signals to each SIB circuit module unit. The test input ScanIn is input into the first layer SIB1, passes through multiple SIBs in the first layer, and is returned to the TAP controller by ScanOut in SIB3. Each SIB in the subsequent layers of the SIB structure uses the two signals TSI and FSO for data exchange, and the SIB uses TDR for data exchange with its post-instrument.
[0059] The first-layer SIB circuit module, also called the Gateway register, serves as the interaction area between the TAP controller and the test network, and controls the on / off of the entire SIB test network. The gateway register is added between TDI and TDO, and is selected when its related instruction is loaded into the instruction register.
[0060] The test redirection is performed using the architecture described in the embodiment of the present invention, including a single chip test phase and a chip integrated system test phase.
[0061] 1. Single chip test phase: insert a general TAP controller, multiplexing logic and chip detector into the single chip before the chip system is bound. In addition, a flexible and reconfigurable SIB test network needs to be established through the Gateway register.
[0062] During the single-chip test phase, each chip is tested before binding, and dedicated test pads are added to the JTAG port for each single chip. Since there are no upper and lower chips before binding, the upper and lower chip detectors fail to detect the logic drive from other chip detectors, which is a high-impedance state "Z", so the output drive of the upper and lower chip detectors is "0". At this time, the multiplexing logic selects the JTAG pads port as the data input port instead of the TSV data path. Through the logic control of the chip detector, the test stimulus is input from the TDI pad port, and the test of the device under test in the SIB test network is realized through the TAP controller. After the test is completed, the test response is transmitted from the TAP controller to the TDO multiplexer, and then further transmitted to the TDO pad port for observation.
[0063] When testing the core and embedded instruments in a single chip, it is necessary to use the gateway register associated with the TDR to establish a SIB test network. This embodiment uses a SIB multi-layer network structure to connect the TAP controller with the embedded instruments of the SIB test network and access and control the embedded instruments through the JTAG port. The overlapping areas of interaction are connected together through special gateway registers. By controlling the state of the Gateway register, the on and off of some connections between the TAP controller and the SIB test network can be controlled.
[0064] 2. Chip Integration System Testing Phase: After the single chip test is completed, multiple single chips are bound into a chip integration system through TSV. Figure 2 The 3D chiplet system shown uses three single chips for binding. After the binding is completed, the chiplet integration system is a structure in which three single chips are simply stacked up and down. With this DFT test architecture, only the JTAG signal of the bottom chiplet needs to be driven to the upper chiplet: the three chiplets share control signals (TRST, TCK, and TMS) and data transmission signals (TDI and TDO) to be serially linked and form a daisy chain through the 3D circuit. Each chiplet has complete multiplexing logic to control the JTAG port signal, so each chiplet is independent of each other.
[0065] The binding stage is the intermediate process of forming a core integrated system with multiple single cores. Since the KGD in the pre-binding test stage may be damaged during the single core stacking process (thinning, binding), the DFT architecture of the embodiment of the present invention takes this test level into consideration. With partial 3D stacking as the goal, multiple embedded instruments or IP cores in the core are tested here. This architecture considers an example of 3D partial stacking with only two cores, and two of the single cores are bound from bottom to top each time. At this time, there is a positional relationship between the upper and lower cores. At this time, the position of the bottom core remains unchanged, and the position of the bound middle core is now the top layer. The upper core detector in the bottom core and the lower core detector in the top core will receive the driving signal from each other, that is, logic "1", so the driving logic "1" controls the corresponding multiplexing logic. At this time, the TDO multiplexer of the bottom core will output the output data of the top core, and the 4 JTAG input ports of the top core will no longer use the original pad port but receive the input signal from the bottom core through TSV. For the test of the top chip, the test stimulus is input from the bottom TDI pad port, bypassed by the bottom chip TAP controller, and input to the TDI multiplexer of the top chip to replace the original TDI pad port. After passing through the test network, the test stimulus is input to the selected DUT, and the test response is driven to the bottom TDO pad port.
[0066] Since the die detector and data selection logic are fully combinatorial, this configuration can be used for any number of partially stacked 3D circuits. It is also applicable to stacking of the same die multiple times. Test-in-binding allows for known good stacking (KGS), which can increase the final 3D stacking yield.
[0067] The post-binding stage is targeted at post-binding testing to achieve a final 3D stack of multiple core particles in which all core particles are bound. Figure 2 As shown, a TDI-TDO serial chain is formed inside, and the post-binding test process is as follows: Since there are core particles on the upper and lower sides of the middle layer core particles, the upper and lower core particle detectors will receive logic "1", and they control the multiplexing logic of the middle layer core particles to select the TSV path as input or output. Since the middle layer core particles have been tested during the binding test, the top layer core particles will be tested directly through the TSV data path of the middle layer to achieve faster data interaction. After the top layer test is completed, the test data will be output to the bottom layer core particles through the TDO channel of the three core particles for observation.
[0068] for Figure 1The 2.5D core particle system shown contains only two single core particles, and there are only two core particles to be stacked, so only the pre-binding and post-binding stage tests are required. At this time, there are only bottom layer core particles and top layer core particles, and there are no middle layer core particles. The post-binding stage test process is the same as the mid-binding stage test process of the 3D core particle system.
[0069] All test levels (pre-binding, binding and post-binding) can be completed using the architecture described in the embodiments of the present invention. The pre-binding level is completed through the pad path, which is automatically selected by the chip detector. In addition, all JTAG instructions are allowed, including the "EXTEST" instruction that can be used to perform inter-chip TSV testing, as well as "INTEST" for intra-chip instrument testing and the Gateway register instruction that can dynamically configure many instruments to be tested simultaneously or independently.
[0070] The DFT test redirection method for the whole process of chiplet binding described in the embodiment of the present invention comprises the following steps:
[0071] Before the chiplet integration system is bound, the TAP controller determines that the single chiplet is located at the bottom layer according to the output signal value of the chiplet detector, generates a test signal, controls the scan test link, selects the test data path of the input signal from the pad port, and tests the single chiplet before binding;
[0072] In the chiplet integrated system binding, if the TAP controller determines that the single chiplet is located at the bottom layer according to the output signal value of the chiplet detector, the test data path of the input signal from the pad port is selected; if the single chiplet is determined to be located at the top layer, the test data path of the input signal from the TSV is selected, and the top layer chiplet and the bottom layer chiplet in the binding are tested according to the selected test data path;
[0073] After the chiplet integration system is bound, if the TAP controller determines that the single chiplet is located at the bottom layer according to the output signal value of the chiplet detector, the test data path of the input signal from the pad port is selected; if the single chiplet is determined to be located at the top layer or the middle layer, the test data path of the input signal from the TSV is selected, and the top layer chiplet, the middle layer chiplet and the bottom layer chiplet in the binding are tested according to the selected test data path;
[0074] In summary, the embodiment of the present invention proposes a DFT test architecture based on IEEE 1149.1 and IEEE 1687 gateway registers, which realizes the testing of 2.5D and 3D stacked integrated circuits at all 2.5D and 3D manufacturing levels. Before the core grain integration system is bound, a general TAP controller (Test Access Port) and upper and lower core grain detectors and other modules are added to each core grain; in the binding stage, the test vector can be redirected to the corresponding middle layer core grain through the multiplexing logic at the JTAG port (Joint Test Action Group) of the bottom layer core grain to test the known good core grain (Known Good Die, KGD) damaged during the 3D stacking process. Similarly, after the core grain integration system is fully bound, the top layer core grain can be further tested through daisy chain. This DFT test architecture can be used for all binding levels without configuration steps, can accurately detect core grains that are damaged during the binding process, and can detect core grains that fail during the binding process in real time to reduce production costs and improve test efficiency. At the same time, the Gateway register is used to configure a flexible and variable SIB (Segment Insertion Bit) network structure to achieve flexible access and dynamic testing of the core or embedded instruments.
[0075] The above description is only a description of the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention. Any changes or modifications made by a person skilled in the art in the field of the present invention based on the above disclosure shall fall within the scope of protection of the claims.
Claims
1. A DFT test redirection architecture for the entire process of chiplet binding, characterized in that: A core particle integration system including a 2.5D core particle system or a 3D core particle system, wherein each single core particle in the core particle integration system has an architecture including: A core particle detector, comprising an upper core particle detector and a lower core particle detector; the upper core particle detector and the lower core particle detector of each single core particle receive and output a first logic signal driven by the lower core particle detector and the upper core particle detector of the adjacent single core particle, and output a second logic signal if the first logic signal is not received; the multiplexing logic is controlled according to the output signal value of the core particle detector, that is, the multiplexing logic selects a test data path of an input signal from a pad port or TSV according to the output signal value of the core particle detector; for a single core particle located at the bottom layer, the test data path of the input signal from the pad port is selected; for a single core particle located at the top layer during binding and located at the middle layer or the top layer after binding, the test data path of the input signal from the TSV is selected; A TAP controller is used to generate a test signal, determine the position of a single chip in a chip integrated system according to the output signal value of a chip detector, wherein the position includes a top layer, a middle layer and a bottom layer; and control a scan test link; wherein the scan test link includes a serial chain of single chips before binding, a daisy chain composed of single chips during binding and a daisy chain composed of single chips after binding; wherein the single chip before binding is located at the bottom layer; The pad port is connected to the TAP controller as the test access port of each single chip before binding. The signal lines of the test access port include: a test data input port TDI, a test data output port TDO, a test mode selection port TMS, a test clock port TCK and an optional test reset port TRST; The SIB test network is connected and controlled by a TAP controller, and is used to test the core and / or embedded instrument in each single chip; wherein the SIB test network includes a plurality of SIB circuit modules; the state of each SIB circuit module includes standby, on and off; when the SIB circuit module is not used, it is set to the standby state; when the SIB circuit module is used to test the mounted TDR and the core and / or embedded instrument in the single chip, it is set to the on state; when the TDR mounted by the SIB circuit module and the core and / or embedded instrument in the single chip do not need to be tested and must pass through the SIB circuit module, it is set to the off state, and at this time the SIB circuit module plays a bypass function.
2. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 1, characterized in that: After the test is completed, the test response is output from the pad port of the single chip located at the bottom layer.
3. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 1, characterized in that: The TAP controller includes: an FSM circuit, a core locator, an instruction register, an instruction decoding circuit and a data register; the FSM circuit is a synchronous 16-bit finite state machine, which receives a first signal from a pad port, converts it into a required test mode and outputs it, maintains synchronization between all components on a scan test link, and allows the movement, capture and update of test data; the first signal from the pad port includes a TMS signal and a TCK signal; the instruction register and the data register receive the output of the FSM circuit and the TDI signal from the pad port, the instruction register outputs the instruction to the instruction decoding circuit, the instruction decoding circuit translates the instruction into a corresponding test mode enable signal and outputs it to the data register, and the data register is used to perform different test functions; the core locator is used to optimize the data path when the test data is bypassed, so as to receive the output signal of the core detector to autonomously determine the core position and control the output multiplexer.
4. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 3, characterized in that: The data registers include a bypass register, a boundary scan register, an IDCODE register and a P1687 gateway register.
5. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 1, characterized in that: Each of the SIB circuit modules comprises: AND gate 1, AND gate 2, multiplexer 1, multiplexer 2, multiplexer 3, UpSIB register and ShSIB register; wherein the two input ends of the AND gate 1 are connected to the Update_DR port and the Select_DR port, and the two input ends of the AND gate 2 are connected to the Select_DR port, the Shift_DR port and the optional Capture_DR port; the output end of the AND gate 1 is connected to the input end 1 of the multiplexer 1, the input end 2 of the multiplexer 1 is connected to the output end of the ShSIB register and the input end 3 of the multiplexer 2, the input end 3 of the multiplexer 1 is connected to the output end of the UpSIB register and the input end 3 of the multiplexer 3, and the output end of the UpSIB register is connected to the input end 1 of the multiplexer 1. It is connected to TDR through the Seli signal, and the output of multiplexer one is connected to input terminal one of the UpSIB register; the output of AND gate two is connected to input terminal one of multiplexer two, input terminal two of multiplexer two is connected to TDI, the output of multiplexer two is connected to input terminal one of the ShSIB register, input terminal two of the ShSIB register is connected to input terminal two of the UpSIB register and the Reset signal, input terminal three of the ShSIB register is connected to the TCK signal and input terminal three of the UpSIB register, the output of the ShSIB register is connected to input terminal two of multiplexer three, and is connected to TDR through the TSI signal, TDR is connected to input terminal one of multiplexer three through the FSO signal, and the output of multiplexer three is connected to TDO.
6. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 1, characterized in that: The SIB test network is a single-layer SIB test network, and a plurality of the SIB circuit modules are linked into a link.
7. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 3, characterized in that: The SIB test network is a multi-layer SIB test network, and the back end of at least one SIB circuit module of the first layer continues to be connected to at least one SIB circuit module, and the SIB circuit module of the first layer is connected to the TAP controller; wherein the SIB circuit module of the first layer, also called the Gateway register, serves as the interaction area between the TAP controller and the SIB test network, and controls the on and off of the entire SIB test network; the Gateway register is added between TDI and TDO, and is selected when its related instructions are loaded into the instruction register.
8. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 1, characterized in that: The first logic signal is 1, and the second logic signal is 0; each upper core particle detector drives a logic 1 to the lower core particle detector in the upper core particle, and each lower core particle detector drives a logic 1 to the upper core particle detector in the lower core particle.
9. A DFT test redirection architecture for the whole process of chiplet binding as claimed in claim 8, characterized in that: The position of the single core particle in the core particle integration system is determined according to the output signal value of the core particle detector, specifically including: When the output of the lower core particle detector of a single core particle is 0 and the output of the upper core particle detector is 0, the single core particle is located at the bottom layer of the core particle integration system before binding; When the output of the lower core particle detector of a single core particle is 1 and the output of the upper core particle detector is 0, the single core particle is located at the top layer of the core particle integration system during or after binding; When the output of the lower core particle detector of a single core particle is 0 and the output of the upper core particle detector is 1, the single core particle is located at the bottom layer of the bound core particle integration system; When the output of the lower layer core particle detector of a single core particle is 1 and the output of the upper layer core particle detector is 1, the single core particle is located in the middle layer of the bound core particle integration system.
10. A DFT test redirection method for the whole process of chiplet binding, using a DFT test redirection architecture for the whole process of chiplet binding as claimed in any one of claims 1 to 9, characterized in that: The steps include: Before the chiplet integration system is bound, the TAP controller determines that the single chiplet is located at the bottom layer according to the output signal value of the chiplet detector, generates a test signal, controls the scan test link, selects the test data path of the input signal from the pad port, and tests the single chiplet before binding; In the chiplet integrated system binding, if the TAP controller determines that the single chiplet is located at the bottom layer according to the output signal value of the chiplet detector, the test data path of the input signal from the pad port is selected; if the single chiplet is determined to be located at the top layer, the test data path of the input signal from the TSV is selected, and the top layer chiplet and the bottom layer chiplet in the binding are tested according to the selected test data path; After the chiplet integration system is bound, if the TAP controller determines that the single chiplet is located at the bottom layer based on the output signal value of the chiplet detector, the test data path of the input signal from the pad port is selected; if the single chiplet is determined to be located at the top layer or the middle layer, the test data path of the input signal from the TSV is selected, and the top layer chiplet, the middle layer chiplet and the bottom layer chiplet in the binding are tested according to the selected test data path.
