serene’s posterous

One of the most interesting talks I attended this week was by Tim Westergren of Pandora.  He didn't have slides and just sat on the table (sometimes sitting on his hands, which is a habit that I have too) and started talking.  He is certainly the chief evangelist for the company.

When you think about it, it is really hard to scale a business which requires musicians to listen to music and analyze them in over 400(?) attributes.  There is a huge body of music out there, and there's always new music coming out.  But the amazing thing is that Tim Westergren is still passionate about his business despite the terrible lows he faced in the business, inspiring people, and not compromising on his beliefs to make more money.  There are no top 10 lists and no accepting of money to play or analyze music.

I was inspired to go play with Pandora again.  And it is a really great tool, and it does pick the music that I like with uncanny accuracy and reminds me about artists that I haven't heard in a while and perhaps introduce me to more music that I like.  The minus is just that the sound quality isn't all that great.  I would describe it as somewhat tinny.

There is an amazing amount of physics and technology advancements that
go into making logic and memory chips. I read this recent IEEE
Spectrum article about double patterning lithography.

Intel is already shipping mass production 45nm technology node chips,
and is planning to ship 32nm chips by end 2009. The definition of
technology node is the half pitch of metal lines of usually a memory
product, but it essentially gives an indication of the smallest
feature size in a chip. It is quite amazing to think that with the
various lithography tricks, manufacturers are able to obtain feature
sizes of under 45nm using 193nm wavelength light.

Lithography is probably not the most difficult problem faced by
integrated circuit manufacturers. Once they figure out EUV,
lithography probably won't be a problem until the end of scaling. At
these tiny dimensions, the transistors are extremely leaky. Electrons
simply tunnel through the gate dielectric. Quantum mechanical effects
are becoming dominant. Short channel effects. Parasitics are
degrading device performance, etc. It will be curious to see what
engineering tricks Intel, IBM etc. would come up with in the coming
years to ensure performance and density improvements in integrated
circuits.

This article raises the point that just producing energy from
alternative sources such as wind and solar farms is not the end of the
story. One needs to transmit the energy to the consumers too.

http://www.nytimes.com/2008/08/27/business/27grid.html?em=&pagewanted=all

Fossil fuels are so entrenched in our lives because they contain their
energy in a stable package that can be utilized where and when we need
the energy. For all the other alternative energy sources, we need to
figure out how to store and transmit the energy. The electricity
supply has to be stable.

For the small amount of energy generated by home solar panels
currently, it is not complicated to just hook it up to the electricity
grid. But beyond a small percentage of alternative energy, more
extensive work is probably needed to figure out how to maintain a
stable supply since there can be a long series of cloudy days or
windless days, or the time of highest demand is not when there is high
supply. Presumably, energy storage technology would have to be
improved. Lithium-ion batteries will not be suitable, sufficiently
energy dense or cheap.

Solar and wind farms are probably located a great distance from urban
centers where there is high demand for energy. There would therefore
be transmission issues. Is there sufficient capacity on transmission
lines? Transmission lines are expensive, and for great distances
there may be unacceptable energy losses especially if AC transmission
is used.

The path for alternative renewable energy sources isn't simple. A lot
of new peripheral technology needs to be developed for solar or wind
farms to be competitive to traditional power plants.

I was reading this article from IEEE Spectrum about First Solar, a
start-up that produces thin film cadmium telluride (CdTe) photovoltaic
(PV) cells.

http://www.spectrum.ieee.org/aug08/6464

The price that has been bandied around for PV cells to be competitive
with traditional power sources is US $1 per watt. Unfortunately that
does not give a complete picture. Some of the other concerns include:

1) Efficiency - if your cells are less efficient, even if the cost per
watt is the same, it requires larger areas and therefore higher
installation costs, which is not included in this number;

2) Expected life of the cells - should be included in cost calculations;

3) Availability of the raw materials - Te is rare on earth;

4) Toxicity of the materials - Cd is quite toxic; and

5) Alternatives and Competition - other possible options for PV cells
are CIGS (can be deposited as a thin film easily, but efficiencies
have been lower than expected); multi-junction high efficiency cells
(efficient but very expensive - probably best as niche applications or
need to use lenses and mirrors to focus light); amorphous silicon
cells (low efficiency but easier manufacturing process); and the
conventional silicon cells (mature technology, efficiency unlikely to
improve further but materials costs would decrease if current
polysilicon shortage disappears).

I guess the key takeaway point is that solar power is all about
economics, the winner would be the technology that has the lowest
overall costs.