Browse Prior Art Database

Method to Detect Hardware Failures in Water- Cooled Systems

IP.com Disclosure Number: IPCOM000228098D
Publication Date: 2013-Jun-06
Document File: 1 page(s) / 28K

Publishing Venue

The IP.com Prior Art Database

Abstract

Described is a method to highlight surface mount component failures in a water-cooled system.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 61% of the total text.

Page 01 of 1

Ȉ ˇ

In water-cooled, high-performance computing system, all surface mount components are facing the cold plate. That makes it impossible to inspect the card for damage to components without going through an extensive maintenance procedure to remove the card from the cold plate. Over the course of production, defective components had to go out on the board and one couldn't tell the whether the monitoring circuitry had gone bad or the component itself had failed.

    In this method, a circuit is worked that powers an LED into the Bus Converter Module (BCM) circuit of a card. When the BCM or other component in the circuit goes bad, an LED is lit on the back side of the card. This allows one to tell what failed on the card without having to take it up from the cold plate. This saves both manufacturing and maintenance time when a problem is reported on the card.

    These BCMs provide a redundant voltage DC-DC regulated voltage source. However, due to them all providing power to the main voltage plane, it is very difficult to identify which module fails. This can be remedied by the utilization of an LED circuit to detect hardware failures in these modules. The circuit would utilize a normally closed relay. The primary side of this relay would be connected to the output of the BCM, and the secondary side is connected to the power source of the LED circuit. In the case of a failure, where the output of the BCM is turned off, the relay would close and illuminate the LED mount...