Anablog

Technical Editor Paul Rako looks at analog technology in power supplies, interface, the signal path, and life in general.

Advertisement

Profile

About Me

RSS Feed

Add this blog to your RSS newsreader!

Analog Design Articles

Blog

Anablog

« Webinar on measuring headphone... | Main | Now NASA has an alternative mo... »

Monday, June 22, 2009

The next killer app? How about three of ‘em?

Jun 22 2009 8:52PM | Permalink |Comments (22) |

As I visited the IEEE IMS show about microwaves and RF last week I could see a very bright future for semiconductors and electronics. For years pundits have been bleating about the next “killer application” for semiconductors. Well I am seeing three of them coming down the pike. First is smart phones. The iPhone changed everything. It not only drove phone replacement sales, it is driving the entire cellular infrastructure. at&t was astonished at the amount of data that people consumed and sent with their iPhones. This means more cellular base station infrastructure, more data infrastructure, more touch screens and a massive amount of analog for cameras and flash drivers and all the other things an advanced phone needs. In countries like India they are passing over the laptop era and jumping straight to smart phones and their cousin, the netbook.

The second killer app I see is LED lighting. Sure, right now it only pays out for refrigerated coolers and architectural lighting that is 30 feet up and hard to replace. But the semiconductor companies never cease to amaze us in the progress they make. How about Cree? What they learn and spend on gallium nitride for their LEDs they can benefit on their RF and cellular base station division. That’s a pretty cool synergy. Pretty soon LEDs will be in general lighting and that is not big, not even huge, it is monstrous. If they can provide the energy savings of CFLs already, and don’t use mercury and don’t have 1-year lifetimes like the cheap CFL bulbs, well that is really good for the environment, rather than phony good.

Third, hard drives are going to get replaced by solid-state drives (SSD). Not all hard drives, but surely ones used for your operating system and anything that needs fast data or reliable data or low-power data. Or shockproof data, see the smart-phone paragraph above. This does not help my analog pals too much, but there will be a mind-boggling amount of silicon bought, processed, and sold for this. Check out this YouTube video of a computer built with 24 Samsung solid-state disks. Sure it may be $16 grand of disks today, but like LEDs and the power supplies for LEDs, they will come down in price dramatically in a few years.

So there you have it, three, actually four, killer apps. Smart phones, netbooks, LED lighting, and solid-state disk drives. Let the celebrations begin, in a couple years there will be a boom in Silicon Valley the likes we have never seen. Massive amounts of silicon need massive amounts of semiconductor machinery so Applied Materials will be doing just fine too. People say that solar power will also be a big push, but for that I think Sunpower is in a far better position than Nanosolar and all the companies making amorphous cheap printable panels using copper-indium-gallium selenide (CIGS) semiconductors. If you are going to pay $20 grand labor for someone to put panels on your roof, better they are 22% efficient Sunpower panels and not 12 or 14% CIGS panels. I know, they claim 19.5% for CIGS, but I will believe that when I can buy it and Consumer Reports runs a story on it. All those high efficiencies for CIGS are in laboratory settings. When I write Nanosolar and ask to confirm that they are in production, they don’t even answer my emails. My new philosophy is that until it is in the Digi-Key catalog it is a marketing fantasy.

As to the whole alternative energy thing, I predict it will fizzle for five years until gasoline goes back up over 4 dollars and carbon taxes drive electricity prices to 25 cents a kwh, about twice what they are in California and 3.5 times what they are in West Virginia. I do think solar-voltaic is the most sensible. Google claims to have some whiz-bang solar thermal setup but the eco-freaks won’t let us build power lines to the installation, so what good is that? The same goes for wind. Solar panels on your roof, that is the most practical thing we have going, and yet another reason for the boom in semiconductors and the machinery that makes them. Remember, LEDs need power supplies and solar power needs inverters. The capacitor and inductor makers are going to have a great future as well. Needless to say, there will be a huge demand for analog engineers, so keep reading EDN, the analog source.

Related entries in: Analog |

Reader Comments

at 6/23/2009 9:06:56 AM, Meredith Poor said:
The link below is a patent on making methanol from electricity (note the date: 1976). Since the electrolysis cell runs at 1.4 volts, one needs a 500 amp power supply to produce roughly 12Kwh per day worth of methanol. Seems like that would consume a lot of solid state components, regardless of what they are.

patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.htm&r=1&f=G&l=50&s1=3959094.PN.&OS=PN/3959094&RS=PN/3959094

at 6/23/2009 2:51:12 PM, Mr. Vaporware said:
"My new philosophy is that until it is in the Digi-Key catalog it is a marketing fantasy."

I couldn't agree with you more.......

at 6/23/2009 4:24:55 PM, Pete said:
Nanosolar is putting in utility scale production in Germany. They sell to panels to utility companies, this is not (yet) a consumer offering. You expect them to waste their time replying to a consumer email when they have 100% of production sold and no consumer products in the pipeline? Also, suggesting that Si will beat thin film long term is pretty ignorant. Remember this post in 5 years. The only metric that matters is cost per watt.

at 6/24/2009 10:18:57 AM, Paul Rako said:
What does a blond say when she wants to make love? "I can't believe I am so drunk!" What does a brunette say when she wants to make love? "Are there any blonds left?" Nanosolar is the brunette of the solar industry. And if Nanosolar was in full-scale production they would say: "Are there any Sunpower panels left?" Since the German government feels it is good policy to take money from taxpayers to subsidize solar installation that don't make any sense economically, well yeah, there is a pretty good market over in Germany. First off, I am not some consumer asking to buy a panel. I am an editor for a magazine with hundreds of thousands of technical readers. Companies with real stories always go crazy to reach and tell the story to the press. Second off, I have a pal that works for a direct competitor of Nanosolar, and he says that they learned to not do what Nanosolar does and brag about all their tech prowess before they can do high-volume manufacturing. Scientists and venture capitalists love to paint a picture of how successful things are, but real engineers know that going into high-volume manufacturing is the real challenge. Anybody can get something to work once in a laboratory. I see pictures of machinery that is supposed to make these panels and it looks like prototype stuff that could make a megawatt a year. Go look at Nanosolar website. The last real press release is December 18 2007. It announces the first production run. Great, but where is the press release saying exactly how many megawatts of panels are shipping every month? There is a title "1GW CIGS deposition throughput achieved" in the press release page but that just points to their blog entry. That talks about a machine that can coat a gigawatt worth of panels per year. Great, but why are they not shipping a gigawatt, a full year after that blog entry? Because the coating machine is a simple adaptation of a web machine (web in the sense of printing and film handling). That is like saying you have a can of paint and a sprayer and that you will be selling houses any day now. OK, look at Nanosolar blog again. The last entry is that they are hiring. Great, but look at the jobs-- They sure seem to be hiring a lot of R&D positions for a company that acts like it has figured everything out. And even more interesting is noting that all the manufacturing jobs are in Germany. Wait a minute, what about the press release two years ago that they bought an old Cisco facility here in Silicon Valley? I bet Germany gave Nanosolar a big bunch of cash to put the factory over there, another subsidy on top of the price subsidy. All I want to see is how many megawatts a month they are shipping. From what I hear over beer with my buddy is that the other companies in this CIGS tech think Nanosolar has done the whole industry a great disservice by talking smack about how great things are going, when really there is still a ton of work to do. I really wonder if the cost of the panels justifies the lower energy you get from them. As we use more smart phones that will push the use of III-V semiconductors like Gallium Arsenide. Those are direct-band-gap semiconductors that are used to make panels for spacecraft. Once we create the infrastructure to make 8-inch III-V wafers, there will be a tech even better than crystalline silicon that will put more cost pressure on the CIGS tech. Printing semiconductors has been a desirable dream for decades but they still make LCD TVs in a vacuum process. And don't think there are not smart people looking at adapting that machinery to make huge solar panels. I am not saying CIGS is worthless. Far from it. Everything is analog and there is a place for all creatures, great and small, in this planet. But there is a constantly changing cost dynamic and you have to look at the 25-year hit of poor efficiency versus the sunk cost of labor versus the cost of the panels themselves. And remember, the reliability of all these CIGS panels is surmised-- conventional solar has really been out there for 25 years so manufacturers understand the failure dynamics much better. No, like I say, when we can buy it in the Digi-Key catalog and measure in my backyard, I will believe all these CIGS claims. Until then my hopes remain with silicon PV. By the way, there is not enough known gallium in the world to make all the panels these CGIS folks brag about, but I am a Julian Simon kinda guy so I have faith we will find more.

at 6/24/2009 1:00:25 PM, Meredith Poor said:
I tend to watch the PV manufacturers and Nanosolar is certainly an outlier. Sunpower, Unisolar, Evergreen Solar, and GE Solar are 'conventional' in that they discuss their production, revenues, conversion efficiencies, and distributor arrangements. They all have investor conference calls, although GE doesn't spent much time on the solar part of their business.

One problem one runs into when one has private (and secretive) investors is that it's tempting to slack off. If the company doesn't have to make sales to survive, 'long lunches' cause no harm. One reason why they might be having problems hiring R&D people is that that candidates that show up might conclude the effort isn't serious.

at 6/24/2009 2:47:09 PM, Tony said:
Hey Paul, what about Solyndra? Their emphasis is on lowering the installation cost (so they only do solar panels for flat roofs, e.g. office buildings). I think there's a lot to be said for their approach -- look at the total system costs including installation.

at 6/24/2009 5:15:32 PM, Paul Rako said:
Good points Meredith and Tony. I also got wind of FirstSolar, who makes thin-film panels but uses cadmium tellurium (CdTe) chemistry. They have a revenue statement and sold a half a billion dollars of _something_ last year. I also found www.greentechmedia.com/articles/read/nrel-confirms-miasols-solar-panel-efficiency-1130/ that shows both CIGS and CdSe both get about 10% efficiency in the real world. That is half what Sunpower gets. That means you are doubling your labor cost for the same power installation.

at 6/25/2009 1:53:03 PM, Brent said:
If there isn't enough Gallium to make all the CIGS panels in discussion, is there enough to make the quantity of GaAs wafers that your response above alludes to?

at 6/25/2009 2:00:17 PM, JRB said:
The ROI is more than labor for installation plus capital cost and the efficiency quoted. It's also lifetime reliability (and degradation over time), sensitivity to ambient and operational temperature (year round, of course), maintenance (dust sensitivity) and of course system weight (especially on a roof). Take also the impact of how direct the light is vs indirect. How effective are these in cloud cover vs direct? How much snow and ice on the system does it take to impact the performance?

I would not make a general statement that a 20% efficient system is better than a 10% system. It's the total system and system life cycle that counts. Oh....life cycle: when it's time to discard, can it be recycled or is there a hazardous waste disposal issue (and cost)?

at 6/25/2009 2:31:10 PM, Larry P Kelley said:
I like all the comments above! Especially Paul Rako's.
Sand is everwhere.....Metallurgical Silicon is cheap. UMG (upgraded metallurgical-grade)silicon is fairly cheap, semi silicon is damned expensive. Sunpower has the right idea, but it could do a lot better if they could learn to use UMG silicon or especially Metallurgical Silicon. One of the secrets to using less pure silicon is to use thinner silicon. Everyone making crystal Si PV cells knows why. Maybe 0.002" instead of 0.012"-0.015". The problem is sawing silicon that thin with a 0.012" thick wire saw,(which translates to a 85% sawing loss) and the handling of a 0.002" thick wafer. My company has built many Si PV cells 0.005" thick with no sawing or grinDing needed. NO SILICON LOSS WHATSOVER!! LARRY P KELLEY lkelley@goruby.com

at 6/25/2009 2:35:47 PM, Larry P Kelley said:
JRB....do you work for First solar?

at 6/25/2009 2:38:46 PM, Meredith Poor said:
Re: JRBs questions.... Silicon PVs have 25 year warraties because manufacturers have 25 year operational life experience: many solar panels made in the 1980s are still producing power. The original Bell Labs cell is still producing power at about 50% of it's original conversion efficiency, and it's now 55 year years old.

Indirect sunlight actually improves PV output, since photons are coming from more directions. The experience with cloudy days is actually good, unless the clouds are thick enough to make the entire landscape dark.

Clearly snow will obscure a PV panel, but cold, by itself, is not material. PVs operate on satellites, and this is a far harsher environment than terrestial cells.

Recycling silicon is about like recycling glass or gravel. Much of what degrades solar panels does not affect the cell, but it's coating. Recoating a silicon cell might well restore it to it's initial capacity.

A lot of these issues have been brought up in Home Power magazine, which has documented answers to these questions over time.

at 6/25/2009 3:13:11 PM, jim said:
what the hell does any of your killer aps have to do with microwaves? did you even understand any of the IMS?

at 6/25/2009 3:22:29 PM, Larry P Kelley said:
Thank you Meridth Poor!

my experiences with Si cells I made in 1973 are possibly better than Bells. I actually left two cells hooked in series outside with absolutely NO protection at all for over 5 years in Michigan. Not even a roof or any cover glass at all. They were glued with silicone onto an acrylic prism and hooked to a buzzer, also glued on. The buzzer wore out. I found it hard to believe that the junction didn't get eroded off by the rain and snow. The prism held the cells at a 45° angle. The really strange thing was that during this time, the concrete sidewalk, which was the support for the solar cell prism, wore down by nearly 1/16" because the initials I put in the concrete nearly disappeared. Larry P Kelley lkelley@goruby.com

at 6/25/2009 4:25:24 PM, Inquisitive said:
One key I think to the discussion of the long term viability of PV solar is the expectation that silicon prices will eventually come down as volumes increase. I haven't heard much discussion around that however. Yes silicon prices have dropped in the last 5 years, but can we realistically expect prices to continue to fall to the extent that PV Solar competes with more conventional sources? In fact, one source described the energy payback (energy produced vs. energy consumed producing the solar panel) taking many years. Bottom line, when PV silicon is made from sand, why does it cost so much?

at 6/25/2009 4:36:43 PM, DM said:
FirstSolar ships cadmium telluride in large modules for utility-scale installations. They claim manufacturing costs (not sale price) of less than $1/Watt (peak). The availability of tellurium has been claimed to be an issue in some stories, but the USGS forecast up to 500 GW/year CdTe production levels can be achieved with known tellurium resources. Of all the non-silicon companies, I would bet on them.

Applied Materials is projecting a PV installed system electricity cost of about 15 cents/kWh in and 6%/year reduction (no subsidies, financing costs and system degradation included, etc). So that is lower than current peak rates, and is projected to be competitive with new power plants in several years.

As was discussed, both companies are public, and present lots of data to the press and investors.

Regarding power lines: don't paint the whole world as California. Boone Pickens pushed through a huge expansion of Texas power lines, to make it easier to delivery wind power around the state. The consumer cost for wind (it is hard to figure out the true cost due to mandates) is 35% higher than other sources in my town.

at 6/25/2009 11:16:40 PM, Meredith Poor said:
Note to Inquisitive: silicon is purified by a method not all that different from aluminum. Essentially, two graphite electrodes are placed in a molten bath of quartz (silicon dioxide) and a potential is put across the bath. One electrode slowly combines with the oxygen in the melt and boils away as carbon dioxide. Over time, all that remains is the silicon. Needless to say, precursor minerals are likely to have impurities, so this isn't as easy as it appears.
<br />
That being said, as wind turbines and PV become more efficient, the cost of making silicon from RE sources will go down. This is the opposite of 'peak oil' in that in the long term PV inherently gets cheaper and cheaper the more it is used.

at 6/26/2009 8:59:26 AM, Larry P Kelley said:
Once again Meredith Poor is 'right on the money'!

Most metallurgical grade Si is made slightly differently.

The thinner a multi crystalline Si PV is, the better metallurgical Si looks.

It works great when the Si thickness is about 0.003"

Lkelley@goruby.com

at 6/26/2009 10:41:41 AM, Bob Witkow said:
SSD technology is here, but the cost to build the fabs that will produce all that NAND is out of sight.

Assuming 50% of notebooks and netbooks sold in 2012 will have a 128 GB SSD, five new $4 billion fabs will be required. That's $20 billion!

It's great news for the fab and capital equipment guys, but it will be tough to convince any memory company's board of directors to approve with their intimate knowledge of margin erosion.

at 6/27/2009 2:33:06 AM, Paul Rako said:
jim: The IMS is not all about military microwaves anymore. There is that segment, but most of the show is about RF, and a cell phone as well as a netbook has 3 radios (main+GPS+Bluetooth) and the netbooks have Wi-Fi and maybe WiMax as well. No there is plenty of RF in phones. Those phones need massive RF infrastructure for base stations and back-haul. Meridith-- you mean silicon production boils off carbon dioxide? Don't tell anyone. It will get banned. I love Bob Witkow's comment because he blends that special talent for science and economics required of a great engineer. I sure see your point Bob. I guess we are down to what-- 35 nanometers-- these days. I only hope the cost of fabs justifies the money they can make. When I consulted to Applied Materials I gave a boss some guff about charging 1.4 million dollars for a machine that had 100 grand of parts in it. He pointed out that I forgot about the vacuum pumps, so it was more like 120 grand for that cluster-tool etcher, but then he pointed out: "Intel pays it off in 9 months." That's when he explained that Applied does not sell machinery. they sell a process and you have to buy their machine to hold the process. I hope the process pays out but your scepticism is well-placed.

at 6/29/2009 1:05:25 PM, Meredith Poor said:
The people legislating a 'carbon tax' on silicon refining will do so while swigging their (aluminum) cans of diet cola. Or polyethelyene water bottles. Or polycarbonate pitchers with the shiny 'chrome' (aluminized) 'plating'.

at 7/13/2009 1:11:40 PM, lookingon said:
the solution for two of killer apps

have you heard of ENER, they have patents for PCM which samsung and numonyx are putting into production

have you heard of ENER they own unisolar they have a vacuum deposition system for a-si thin film, real company,ral product, real profit. it rolls on flat to a buildings roof apparently you can put a bullet through it and it still works

Anablog

Profile

RSS Feed

Recent Posts

Recent Comments

Most Commented On

Archives

By Category

Analog Design Articles

Blog

Monday, June 22, 2009

The next killer app? How about three of ‘em?

Reader Comments

Post a comment

Please visit these other Reed Business sites

Display Name

Change Image Before submitting this form, please type the characters displayed above. Note the letters are NOT case sensitive.