Around the time the garage was finished enough, I looked into solar panels for the house, because

• the electricity bill was getting a bit steep (between $200 and $300 per month), although we found part of that bill was due to a broken high amperage fan that was running 24/7.
• electricity rates just went up, and are scheduled to go up even higher
• the federal tax credit for going solar became unlimited on how much it covers, making panels effectively half price between the CA and federal rebates (the federal credit used to be fairly curtailled).
• the tech, cool, and green factors of course did not hurt :)

Solar & PG&E 101

So, one would think that getting solar panels is not that hard: you call a company over, they measure your roof space, your design factor (i.e. which way the roof is pointing as well as its inclination vs the ground, something like 18 degrees in our case), and then you decide how many panels you can fit or how many you want to put to offset how much your power bill.
The first problem is a not so simple multi factor equation because you switch to a time of use electric meter where the time of the day changes which billing range you're in, and each range has 5 tiers, the more you use the more expensive your Kwh becomes.
Then, to make things fun, you can try to compute how much you end up paying by Kwh for your solar electricity by factoring in how much you paid for the system over 30 years, and currently that per Kwh price for baseline usage (i.e. the first tier) was around 13¢ for me, or more than what I pay from PG&E. However, all remaining tiers were more than solar. Then, you could say that PG&E prices go up and baseline price will go up beyond 13¢. This is entirely correct but you have to factor this against investing the money you'd have put in panels at 4% and see what's better off.
Oh, and if you decide to generate your baseline use too, you don't want to overproduce because if you do, PG&E will not give you a negative bill: over the course of a year you can only offset your bill, not make it negative.
The last kicker is that you don't have to produce as much as you use to have a $0 bill because when you produce in May-Sept between noon and 6 (or something like that), that electricity is worth several times a kWh you use at night or during the winter. I found that producing 8,000Kwh can offset a 12,000Kwh yearly use.

Fitting solar panels on limited usable roof space

Anyway, now that this is out of the way, our house made going solar a bit more challenging because our roof is very jagged and most of our roof is not getting good sun exposure: a lot of it is east, which is not that desirable, whereas our west facing roof (better because it gets sun in the afternoon when electricity is worth more), gets significant shading from the trees on the creek side of our property.
So, we're left with our garage roofs pointing south-south-east unshaded, and a few pieces of our roof pointing due south, and that made things interesting because we had to fit panels on limited roof space, part of which was triangular shaped, making rectangular panels harder to fit, especially the bigger ones.
After getting my quote from the first vendor, it became clear that we'd have to worry about price per Kwh as much as Kwh per square footage of roof used.

Picking Vendors

This turned out to be much harder than I thought because they had different panels with different efficiency and pricing, which just added more variables to the already multiple variable equation of what to get. Also, all vendors gave you production estimates for the year, and end pricing after rebates. However most did not give Kwh/year / dollar (i.e. how many dollars a Kwh every year cost you), and only one gave the square footage used by their panels, and the more important Khw/year / sqft (i.e a panel array that produce 23kWh/Y per sqft of roof space used). While I know that those are not perfect numbers, they were the most useful in my comparison.
I wondered why the different vendors don't provide that useful data, but then I also realized that the 4 tech sales reps I talked to all said "oh, I never had anyone ask me that, let me find out for you" on multiple occasions. So it looks like most people just look at their spreadsheets and pie charts, find them pretty and sign based on that, or maybe based on price without knowing if they're getting the proper technology, or an install on paper that will even fit in practise. Admittedly my install case was probably a bit harder than some, but I got quotes that just didn't produce enough to be interesting and where the salesperson didn't measure what was really possible, so didn't offer it, or a quote where quite frankly the panels didn't fit (I went on the roof and measured), or a quote with thin film where the numbers showing thin film degradation over time had a 3X difference between a spec sheet and the sales pamphlet (-0.5% per year vs -1.5% per year, which is a huge difference). I ended up picking between the four vendors that had been pre-screened by work, and I'll list them below in the least desirable to most desirable order:

• Rec Solar was last for two reasons: the salesman is the only one who never even climbed on my roof to measure exactly what surfaces were usable. He just eyed them and gave me a quote on paper without actually knowing if they would fit, what the design factor was, or if we could put more panels. Yet, I was supposed to sign, and then they'd come and see what they could really do for me. Yeah, right... Anyway, they did offer interesting and efficient smaller sanyo panels, but they were totally overpriced.
  Oh, the best part: one of the quotes was badly done and quoted a likely incorrect kWh/Y figure, giving a ridiculously high $4.3 per kWh/Y
• Akeena had a good, knowledgeable technical salesman, but their system was very overpriced and while it's nice for having panels that connect like legos, requiring fewer attach points and wires, my roof did not allow for that many panels to be strung together and put in a single place. Out of the two quotes I got, the prices were quite high, and the panel efficiency was fairly poor too for monocrystaline (supposedly better) by being beat by much cheaper polycrystaline panels (evergreen 210) from solarcity.
• Solarcity (First Solar reseller) came as a close second. The salesman really did his best to fit his system on our roof, address my concerns and my questions, and they threw in free monitoring of both production and home use (usually another $2k), as well as price matching. They gave me 3 fairly different options (at my request).
• Cobalt Power (Sunpower reseller) was the last option I got, based entirely on Sunpower panels. To their credit that was the only option they needed because their panels were quite good.

The major problem I had comparing the vendors is that they all give you predicted production numbers that they call conservative, but have nothing to back them up. In other words, if they promise you a system that could make 9000kWh, but due to a difficult or poor install, or an over optimistic design factor they end up producing much less, that's pretty much too bad. Most of them would not back up their production numbers with any guarantee whatsoever. In the end, I got Cobalt to back up 85% of the production they promised, but if you think about it, a production 15% lower than what was hoped for is pretty big (of course, it's hard to give a 100% number since it also depends on weather).

Cobalt Power ended up winning because they quite frankly had the best panels. Not only did they have the best efficiency (yearly power per sqft of roof space) which was a crucial factor for us, their panels were a bit smaller so they fit better in confined spaces, and they were also quite competitively priced and not much more expensive than Solarcity.
Solarcity could have worked, but I didn't trust thin film too much since no one else used it, its degradation over time was just unclear and I read conflicting information about it. As for their evergreen polycrystaline solution, it was actually decent, but was 30% less efficient per amount of space as sunpower panels, while not being that much cheaper, so it was hard to go with that, even though I really liked the Solarcity rep too.

So there we go, I signed up with Cobalt, knowing that it was the best bang for the buck for the most efficient solution that we could fit on the good portions of the roof.
At the end, I did agonize between 27 225W panels or 30 panels as I was slightly worried that 30 panels just might produce more than what we used (at which point they become a clear monetary loss). In the end, we agreed that as long as 27 panels brought me in baseline, erasing all of baseline just wasn't necessary since it's a best a wash, and more likely a small loss over investing the money you didn't spend.

Install starts in a few weeks, I'll report back then. In the meantime, below was a rather time consuming to gather and compute comparison table that should be useful to other folks. If you're looking at this yourself, I heartily recommend you consider Sunpower first (Cobalt Power distributor in the bay), you can Email Mark Byington markb(at)cobaltpower.com for more details on a home visit and quote.

. company panels qty panel size panel ft^2 total panel ft^2 W STC shading factor kWh/Y net cost ($) price/ kWh/Y kWh / Y / ft^2 comment . cobalt SPR 230 27 31.4" x 61.4" 13.38 361.26 6210 0.86 8463 27416 3.23951317499705 23.4263411393456 highest power density . cobalt SPR 215 30 31.4" x 61.4" 13.38 401.4 6450 0.84 8539 25882 3.03103407893196 21.2730443447932 . cobalt SPR 225 30 31.4" x 61.4" 13.38 401.4 6750 0.84 8909 27665 3.1052867886407 22.1948181365222 . cobalt SPR 225 27 31.4" x 61.4" 13.38 361.26 6075 0.85 8114 26216 3.23095883657875 22.4602779161822 . . solarcity sanyo 200 35 31.4" x 61.4" 13.38 468.3 7000 0.88 9006 33160 3.68198978458805 19.2312620115311 . solarcity evergreen 210 34 37.5" x 65" 16.92 575.28 7140 0.88 9047 28696 3.17188018127556 15.7262550410235 . solarcity FS -275 95 23.6" x 47.2" 7.75 736.25 7125 0.87 9196 27032 2.93953892996955 12.4903225806452 first solar thin film, note 95 panels, and lowest power density . . akeena andale ST175 24 32.2" x 62.6" 14 336 4200 0.86 5293 21948 4.14660872850935 15.7529761904762 . akeena andale ST175 32 32.2" x 62.6" 14 448 5600 0.81 6354 28535 4.49087189172175 14.1830357142857 . . REC solar sanyo 195 28 35.2" x 51.9" 12.69 355.32 5460 ? 5854 25246 4.3126067646054 16.4752898795452 quote bad, numbers don't even add up . REC solar sanyo 195 24 35.2" x 51.9" 12.69 304.56 4680 ? 5643 21589 3.82580187843346 18.5283687943262