The technical field relates to application specific integrated circuit (ASIC) design, and, in particular, to ASIC integrated method for monitoring high impedance failures in chip interconnects.
High impedance connections in application specific integrated circuit (ASIC) attachments can cause system failures that are extremely difficult to debug. Many new connector systems (such as sockets for ball or land grid array packages) are especially susceptible to co-planarity problems, which result in new failure modes as compared to pin and socket connectors. These failure modes present significant challenges to the design and manufacture of high quality systems. Currently there is no way to predict and proactively deal with such failure modes, since such faults don't always manifest themselves as pure ‘opens,’ which, by nature, are much easier to detect.
The failure modes that result from high-impedance connections vary widely from easily detectable bus errors to completely unpredictable behavior. When these types of failure modes have been seen, the typical ‘solution’ has been to keep swapping boards until the system starts to work again. Debugging has been done by taking resistance measurements by hand (using an Ohm meter) to determine interconnect resistance. Unfortunately, this method is extremely time-consuming and, in many cases, results in an inaccurate reading. It is also something that occurs ‘after the fact.’ A more precise ‘in system’ approach is needed.
An advantage of the embodiments described herein is that they overcome the disadvantages of the prior art. Another advantage of certain embodiments is that they may detect if a pin failure occurs or pro-actively predict interconnect failures before the failures actually occur. Yet another advantage is that this prediction is accurate and performed automatically rather than manually or by hand. Still another advantage of certain embodiments is that they provide greater system uptime and a better customer experience. Another advantage of certain embodiments is that they provide means for monitoring the resistance of a representative sample of interconnect pins during operation and provide for logging the resistance information for use in debug or proactive fault management. A further advantage of certain embodiments is that printed circuit board area and cost are reduced since most of the hardware associated with monitoring resides on the chip.
These advantages and others are achieved by an apparatus for detecting high impedance failures in chip interconnects. The apparatus includes monitoring circuitry located on a chip. The monitoring circuitry includes a resistance continuity monitoring circuit (RCMC) that measures the resistance of a connection of a representative set of pins on the chip with a circuit board and outputs an analog measured resistance data. The monitoring circuitry further includes a signal path connecting the representative set of pins to the RCMC and a system interface for connecting the RCMC with other system components. The analog measured resistance data is converted to digital resistance data, which is displayed on an output device.
These advantages and others are also achieved by a method for detecting high impedance failures in chip interconnects. The method preferably measures the resistance of a connection of a representative set of pins on a chip to a circuit board using monitoring circuitry located on the chip. The measuring step produces an analog measured resistance data. The method further converts the analog measured resistance data to digital resistance data, displays the digital resistance data on an output device, and performs an algorithm on the digital resistance data.
These advantages and others are also achieved by a computer-readable medium comprising instructions for measuring the resistance of a connection of a representative set of pins on a chip to a circuit board using monitoring circuitry located on the chip. The representative set of pins connects the chip to the circuit board, and the measuring step produces an analog measured resistance data and is executed while the circuit board is operating. The computer-readable medium further includes instructions for converting the analog measured resistance data to digital resistance data, and displaying the digital resistance data on an output device, and performing an algorithm on the digital resistance data.
The detailed description will refer to the following drawings, wherein like numerals refer to like elements, and wherein:
A method and corresponding apparatus for monitoring high impedance failures in chip interconnects use monitoring circuitry on a chip to provide accurate and pro-active prediction of interconnect failures. The apparatus may include a resistance continuity monitoring circuit (RCMC), and a signal path connecting a representative set of pins to the RCMC. The RCMC measures the resistance of a connection of the representative set of pins with a circuit board during system operation and outputs a measured resistance data. The apparatus further includes additional analog-to-digital (A/D) hardware to perform an analog to digital conversion of the measured resistance data. The apparatus further includes a system interface for connecting the monitoring circuitry with other system management devices. The method then performs an algorithm on the measured resistance data, potentially warning of likely interconnect failures. The algorithm may include comparing the measured resistance data with a known threshold resistance value. Alternatively, the method displays and logs the measured resistance data for further study and analysis.
With reference again to
The RCMC 24 is an analog precision circuit that determines the current resistance value of the interconnect of the pins 16 with the circuit board 12. The RCMC 24 may be integrated with the A/D converter 28 as a combined circuit. In the embodiment shown in
With continued reference to
Each pin area 18 preferably has a representative monitored set of pins. In the present embodiment, these monitored set of pins are illustrated as the shaded pin pairs 20 seen in FIG. 1. In other embodiments, the representative monitored sets of pins may be different than a pin pair 20. For example, different patterns of pins may be selected as the representative monitored set of pins for all or some pin areas, as shown as the shaded pins 21 located at the four corners and the center of the ASIC 14. The different patterns of pins may include more than the two pins in a pin pair 20. The possible representative monitored sets of pins vary depending on the size and number of the partitioned pin areas.
The pin pairs 20 are preferably connected to the RCMC 24 via a signal path 22 so that the RCMC 24 can monitor and measure the resistance of each pin pair 20 interconnect, as shown in FIG. 1. Each pin pair 20 is preferably connected to the RCMC 24 via a separate signal path 22. Alternatively, the pin pairs 20 may be connected to the RCMC 24 via a single signal path 22 connecting all the pin pairs 20 or by multiple signal paths 22 at least some of which connect a plurality of pin pairs 20. However implemented, the signal path 22 is preferably accomplished through routing on the ASIC 14 and routing a trace on the circuit board 12 to connect the pin pairs 20 to form a closed loop of daisy-chained pins 16 (or pin pairs 20 if connecting more than one pin pair 20 in the signal path 22).
In using the apparatus 10 for detecting and rejecting high impedance failures, the assumption is that the interconnect resistance measurements of the representative monitored set of pins in a pin area 18, e.g., pin pair 20, apply to the other pins 16 in the pin area 18. In other words, the conditions of the pin pair 20, as determined from the resistance measurements, are likely also the conditions of the other pins 16 in the pin area 18. Therefore, if the resistance measurements of the pin pair 20 interconnect indicates an imminent failure, the other pins 16 in the pin area 18 are likely also imminently failing.
The RCMC 24 measures the resistance of the entire signal path 22, including the resistance of the pin pairs 20 interconnects and the trace on the ASIC 14. Preferably, the resistance of the trace is minimized so that measured resistance is the resistance of the pin pairs 20 interconnect. The resistance of the pin pair 20 interconnect is the resistance to a signal flowing from the circuit board 12 to the ASIC 14 through the pin pair 20. Alternatively, the resistance of the trace is known and normalized out of the measurement.
The measured resistance data of the pin pairs 20 interconnect is then transmitted to system management devices, as described above, for further processing. For example, other components on the circuit board 12, including, for example, a processor (not shown) executing instructions stored in memory (not shown), may receive the resistance data via the system interface 40 and perform various algorithms on the resistance data. For example, the measured resistance data may be compared to known values of pin pairs 20 interconnect resistance. These known values, for example, may specify a threshold resistance value for the pin pairs 20 interconnect. If the measured resistance data is greater than the threshold resistance value, it may be determined that an interconnect failure is imminent. In principle, as the resistance increases, the likelihood of failure increases.
If the measured resistance data indicates that a failure is imminent or has occurred, the system management devices take appropriate actions, such as signaling the user or system support personnel. The ASIC 14 or circuit board 12 may be replaced or the data may be further analyzed depending on the likelihood of interconnect failure as determined by the system management devices algorithms.
Different algorithms and threshold resistance values may be desired for different applications. For example, for a critical application with high speed, the threshold resistance value may be set low so that a small change in resistance will trigger a failure signal. On the other hand, for a less critical application, the threshold resistance value may be set higher so that the apparatus 10 may not show a failure signal until the measured resistance of the pin pairs 20 interconnect has changed significantly and a failure is imminent.
The algorithm of the above-described methods may be implemented as software. The software may be stored on a computer-readable medium as instructions for executing the above-described methods. Accordingly, with reference now to
While the method and apparatus for monitoring high impedance failures in chip interconnects have been described in connection with an exemplary embodiment, those skilled in the art will understand that many modifications in light of these teachings are possible, and this application is intended to cover any variations thereof.
Number | Name | Date | Kind |
---|---|---|---|
5383194 | Sloan et al. | Jan 1995 | A |
5444715 | Gruetzner et al. | Aug 1995 | A |
5912901 | Adams et al. | Jun 1999 | A |
6000051 | Nadeau-Dostie et al. | Dec 1999 | A |
6029263 | Gibson | Feb 2000 | A |
6087842 | Parker et al. | Jul 2000 | A |
6097203 | Parker et al. | Aug 2000 | A |
6286119 | Wu et al. | Sep 2001 | B1 |
6445188 | Lutz et al. | Sep 2002 | B1 |
6517369 | Butterbaugh et al. | Feb 2003 | B1 |
6614253 | Berkely | Sep 2003 | B2 |
6617841 | Thao | Sep 2003 | B2 |
6677744 | Long | Jan 2004 | B1 |
6724209 | Whitten et al. | Apr 2004 | B1 |
6806718 | Berkely | Oct 2004 | B2 |
20030025518 | Berkely | Feb 2003 | A1 |
20030213953 | Sohn et al. | Nov 2003 | A1 |
20040104740 | Burns et al. | Jun 2004 | A1 |
20040108866 | Burns et al. | Jun 2004 | A1 |
20040148122 | Whitten et al. | Jul 2004 | A1 |
Number | Date | Country | |
---|---|---|---|
20040249612 A1 | Dec 2004 | US |