Before the Dawn of IP Telephony - Part 28Transistor destroyed for unknown reason (summer ~ October 2002)
These contents translated a serialization article carried by ITPro IP telephony ONLINE published by Nikkei Business Publications, Inc. Jump to the original (Japanese).
eSound Venture Unit
Oki Electric Industry Co., Ltd
We were at the final stage of line unit (LU) development. It was then discovered that the LSI, which had already entered mass production, did not perform sufficiently under certain conditions during software assessment. The situation seemed hopeless with no measures to be found.
At the end of July just immediately before summer break, we were almost near completion of tests, with seven types of noise measures finished and battling towards the completion of the full indoor device for telecommunication carriers. It was just when we were getting over the difficult part and plans for shipment becoming clear. Suddenly, bad news came out of the blue.
Prototypes that are manufactured for verification of the plant production line prior to the start of mass production are referred to as "mass-production prototypes." And we were notified that the mass-production prototype did not pass the electrostatic test (a test that artificially reproduces static electricity that may occur in a real environment by instantaneously applying high voltage).
Failure of a test after reaching the stage of mass-production prototype verification is not common. Design prototypes are manufactured using a process equivalent to mass production, and the verification of mass-production prototypes is considered as a "precautionary check" among the parties concerned.
However, the bad news did not end there. A test using a different mass-production prototype also failed for the same reason. And about the same time, the same phenomenon was reported by one of our customers provided with a mass production sample. We were completely lost. We were prepared for all-out war until the cause was found and a proper measure was taken. And from that day, our battle in the summer heat had started.
Photo 1: Thermostatic chamber used for analysis
As a result of investigation, we discovered that the transistor was destroyed. But we did not know why. Through continuous investigation, we found that there was a slight change in output of the new device introduced in the previous episode at the time when the transistor became destroyed.
But we had no clue about why there was such change, and why such change led to the destruction of the transistor. From the outside, devices that operated normally and devices that led to destruction looked identical.
The design related to the characteristics of sound, which was realized by circuit design of hardware in the past, was built into the device and its settings were available using software. Part of the hardware design can be realized by setting parameters within the device. In other words, the design such as pattern design was simplified by conversion of hardware to software. This led to extreme difficulty in analyzing problems for those that did not understand the meaning and the relationship of registers within the device when a problem occurred. The temperature surrounding the thermostatic chamber after repeated verifications was hotter than the mid-summer heat (photo 1). And all the members working on this project were forced to spend days in the heat.
There was no mistake that we couldn't perform analysis without the cooperation of the device vendor. Our sweat-drenched efforts to find the cause continued while involving the agency and engineers of the device vendor in the U.S. Since the vendor was well aware that the device was selected by us being a fan of the product, they were very open about cooperating with our needs. Although I never regretted the use of this device, I definitely had mixed feelings when seeing the development members battling the heat in front of the thermostatic chamber.
Sharing the same ideal design concept
Let me go back a little again and explain why I liked this device so much and decided to use it.
In reality, more voltage than necessary must be handled in order to reliably ring the bell of a telephone. The telephone was invented more than a century ago, and it can be said that the technical fundamentals of that time were carried forward without many changes. In order to ring the bell, more voltage than what seems rational at the present is needed.
No matter how appealing the value born from innovation may be, there is always this aspect in the real world, and thus continuous changes are needed to prevent problems during actual use. Although I believe that there will be a communication tool that exceeds conventional telephones by the widespread deployment of broadband, it is crucial for the new system to accommodate telephones that are being used today. And a large voltage is not necessary in the case of just sounding an incoming call tone as voltage from a dry-cell battery is sufficient. However, several ten times the voltage of a dry-cell battery is needed to ring the bell of a conventional telephone. Since more voltage than necessary was applied, it was converted into excessive heat. It is critical that this energy be used to realize only the intended functions without producing heat.
This device was built with a revolutionary function of keeping to minimum the conversion to unnecessary heat. In order to suppress heat, it is critical to create only the voltage and current necessary for operating the needed function. Although this is not easy, I shared the same ideal design concept. However, the downside-and the only downside-was that impedance was susceptible to large noise.
Commercialization after implementing triple fail-safe function
As the result of continuous and bitter verifications, we discovered that the destruction of the transistor was caused by multiple factors of the register being rewritten by external noise such as static electricity, and the rewritten register being slightly off from the threshold of peripheral devices.
If the register is rewritten, the process of destroying the transistor would never occur even when the trigger is inputted if the value of the peripheral device is correct. However, if there is input of a trigger when the value of the peripheral device is slightly off from the threshold, the state within the device and value of the peripheral device changed causing destruction.
Photo 2: Product displayed as "Broadband & Mobile IP Communication" at CEATEC JAPAN 2002, immediately after start of shipment
Once the cause was found, implementing a measure was easy. A triple fail-safe function was implemented to prevent destruction no matter the circumstance after guaranteeing the accuracy of product parts, and the product was released into the market (photo 2).
Through the measure of this problem, we were able to eradicate the hidden demerit of this device and were able to recognize again its advantages and disadvantages. And my feeling as a "fan" toward the device vendor that cooperated continuously to break through the wall of the problem had turned into a feeling of "trust." Advantages can be maximized and disadvantages can be compensated. And when there is a problem, it can be solved quickly with reasoning. It was if an effective technical concept was integrated with our own to become one.
"Only provide products that we can control"-as obvious as it may sound, it is the basic stance for a product to be safely used by end users and spread widely in the world. In essence, it is the mission of vendors providing products and indispensable for carrying out responsibilities.
... To be continued