AKP (ASX) Ann: Audio Pixels Commences Integration Stage

Originally posted by BobF ↑

As I understand it, which is only based on first principles, the ASIC is an electronic device (made for AP by ICSense) that takes in a digital signal, "the music", and translates it into a series of timed voltage pulses that are sent to different addresses. These pulses exit the device at a set of metal terminals.

The MEMS (made for AP by TowerJazz) is the business end of the product. It houses a 2D array of "audio pixels" - microscopic pistons that can be independently moved into "on" and "off" positions. These are driven by the addressed electrical pulses from the ASIC. The entry points on the MEMS are a series of metal terminals around the edges of the square array that lead into the pixels of the device itself.

Integration is making the interconnection between these two sets of metal terminals such that the correct electrical signal is routed from the ASIC innards to the MEMS innards, so that the correct pulse travels from the ASIC to the correct audio pixel. If there is any loss or attenuation of the voltage pulse then one or more pixels will not switch on or off. As the connections address rows and columns of pixels, such that the pixel or a group of pixels at the intersection of rows and columns turn on or off, it will be vital that these connections are made perfectly, and that all the stipulated voltages get through all the circuits.

Achieving this requirement was mentioned as having recently presented problems, but which, according the announcement, has been resolved by a trivial circuit redesign, presumably on the ASIC. From previous announcements I had thought that this would have had to have been incorporated into the current 2nd generation ASIC.

Once this interconnection process can be performed correctly and reliably on a production basis this assembly can be potted up into a packaged device (by AEMtec) and will be ready to roll off the presses in huge numbers.

From previous announcements it appears that test lots of integrated devices are regularly produced at intermediate production points for testing.

I believe there is little doubt that this technology actually works as expected. Even the original "proof of concept" prototype chip from several years ago produced low-fidelity music (the "Ring of Fire" video clip). The current version has been shown and reported as producing sound of unsurpassed fidelity and power, but not yet as the product of a mass produced process.

It is unlikely that any of these have been set up to play music because from a testing point of view that is not necessary: the parameters that allow hi-fi music to be produced are better evaluated by a suite of electronic and audio parameter checks than by playing actual music. But if the testing checks out the device will undoubtedly be capable of playing music of exceptional quality.

But that will surely follow, and that will undoubtedly be the centrepiece of demos to the audio industry.

When a digital sound signal is sent to the device the ASIC will produce crystal clear music from the MEMS. Voila: a new audio revolution will be born! When this finished device is built into consumer electronic products all the promise of the AudioPixels technology will begin to be realised, and shares in the company will then have every reason to skyrocket.

We have now been told that we have approximately 6-8 weeks to wait for confirmation that the ASIC can be reliably interconnected to the MEMS on a mass-production volume basis. One could expect that packaging the resulting device would be a simple routine process. While many of us thought we had already got to that place some months back, we are now officially a lot closer to the completion of Phase IV.

Expand

Thanks Bobf for that most encouraging information.

On an otherwise idle Saturday, I have been thinking about the nature of the connections between the ASIC and the MEMS. It seems that the MEMS has 32 connections on one axis and another 32 on an axis at right angles to the first. This enables 32 X 32 = 1024 AudioPixels to be addressed individually or in groups by the ASIC. Thus there are 64 points of connection (32 + 32). If there is one bad connection amongst these 64, it would not just affect the operation of one AudioPixel but 32 of them. I can only imagine what that would do to the music or other sound being reproduced.

Therefore the assembly process by AEMtec needs to be absolutely accurate in a mass production situation to provide a reliable finished product from equally reliable ASICs and MEMS chips. This is obviously not a trivial challenge and I take my hat off to those who are bringing the project to success.

In addition to generating great wealth for themselves (and hopefully some for me) they are also looking at spin-off projects to help the blind to 'see'. This is surely the stuff that attracts Nobel attention.