[GTALUG] convert sun rays to cryptocurrencies.

Alvin Starr alvin at netvel.net
Wed Sep 6 23:49:57 EDT 2017

Lets start by changing the subject.

On 09/06/2017 08:12 PM, Stewart C. Russell via talk wrote:
> On 2017-09-06 09:06 AM, Alvin Starr via talk wrote:
>> A client came to me asking about helping him setup an Etherium mining
>> server pool.
> Hey, I know that there are some honest cryptocurrency types out there,
> but there are some definite shady ones around Toronto. One Etherium
> joker stole ~$500 of Raspberry Pis and accessories from my employer.
> Caveat consultor.
> I'm a (sometime) utility solar designer. The idea's interesting, but the
> implementation's tricky.
I have worked with this client for well over a decade and he is not shady.
But In general I agree that the cryptocurrency world is a bit dodgy at 
Ethereum looks to be a bit different in that is seems to be more 
intended to provide contracts and other services  over just being a 
trading medium like bitcoin, litecoin...

>> What if you glue a GPU with very little extra hardware to the back of a
>> solar panel.
> * Solar modules produce variable I-V output depending on sunlight and
> air temperature. You'll need to use some kind of inverter to stop your
> miners becoming puffs of expensive smoke in their first winter dawn.
There will need to be a  real quality DC-DC converter because there is 
not much point converting the DC output from the panel to AC only to 
convert it back to DC.

Possibly DC  to 12V to some batteries or ultra-capacitors to deal with 
transients or possibly to provide run time extension.
(more on this later)
>> Stick it out in the sun and let it calculate 8-10 hours a day.
>> Put a big heat sink on the GPU and it should be able to stay cool enough
>> just from air cooling
> * Solar modules run roughly 20°C hotter than ambient, often more. The
> huge project I worked on in Arizona quite frequently hit 45°C in the
> shade, so your GPU would be around 70°C under no load.
I kind of wondered about this.
To some extent this would not be much worse than the kind of environment 
that cell phone equipment needs to operate in.

> * The dead air behind modules is *very* still. Passive cooling would
> likely melt.
The question would be how much temperature differential is required 
between the air and the heat sink to dissipate the number of watts the 
system generates and what limits does that put on the operational 
> * You'd need to weatherproof your GPUs against water, dust and ice. Any
> crevices attract spiders and ants, and junction boxes in desert
> locations are a favoured haunt for scorpions and rattlesnakes.
I worked on passively cooled avionics projects in the past and its 
doable but somewhat complex when you have to start embedding heat 
transfer sheets to your PC boards.

The enclosures would need to be sealed and just have a set of wires 
going to the DC from the panel and some coax going to an antenna.

>> Take a few thousand of these and set them out in a sunny place and you
>> would have a coin generator.
> * You'd need a small amount of ride-through battery, as a heavy cloud or
> shading on the wrong bit of the module can reduce your power output to
> close to zero.
>> Once the hardware is paid for then the operational cost would be close
>> to 0 but for some glass cleaner and rags.
> * Almost everywhere on the planet requires some kind of permit or
> environmental impact assessment. Even glass cleaner and rags need an
> MSDS and hazard management plan.
Lowes is selling solar projects with what appears to be 0 environmental 
Lots of places are selling solar panels without any regulatory requirements.
> * Some module types (thin film) are in perpetual danger of being
> classified as hazardous waste due to their cadmium content.
I am sure Shell or Exxon... Would love to see that happen.
> * If you're far away from people, network is hard.
Only in North America. Just about every 3rd world country has ubiquitous 
mobile access.
There is also satellite internet.
The latency is huge and the upload bandwidth is small but it could be a 
workable solution.

> * If you're in an isolated area, theft of solar equipment is a huge problem.
> * There's also something about solar modules that attract people with
> shotguns. It usually ends badly for your installation, unless there's a
> big fence.
Well that is one I never thought of but in a strange way I can see it.
That and kids with rocks.

Your making a great argument that solar should never be used as a 
technology because its dangerous and a target for vandals and other 

>> They would need to be networked together but it does not have to be high
>> speed.
>> It could be a low power mesh to an edge that connects via something as
>> simple as cell phone data.
> * Solar modules, with all that wiring and semiconductor hanging about,
> make about the best radio signal attenuator I've ever seen. You only try
> to make a cell call or get a GPS fix under a solar array once.
Put antennas on the top corners of the panels.
Since your building the system you can route the RF to wherever you want.
> * The inverters you'll need can produce a fiendish amount of radio
> interference unless they're shielded in expensive enclosures.
true but not an insolvable problem.
>> Solar panels cost about $1/watt and there is no ongoing cost but power
>> here costs about $0.08-0.16/KWH.
> * Modules are way less than $1/W now. Racking will likely double the
> cost of your installation, though.
> * A utility-grade 330 Wp module is about 2 m² and weighs around 22 kg.
> At these latitudes, you'll get about 40 W average across the whole year.
> Since you're only proposing running through the day, you'd get more than
> that. Have a play on PVWatts <http://pvwatts.nrel.gov/> for hourly
> output. Choose a module like a Canadian Solar CS6U-330M for modelling:
> it's a solid Mono-Si module with decent efficiency.
I was not thinking so much about deploying in Canada.
We have way too much dark.
It feels like 16 hours of dark a day from October to June with the rest 
of the days being cloudy ;)

Initial my thought was pick a nice sunny island in some equatorial area 
and use the local ocean for cooling.
You would not have to pump much and so long as the GPU's are running 
less than 50C or so life would be good.

Another thought was. Why not float them on a lake or pool of water and 
let passive cooling take place.

Passive air cooling may not work in truly hot environments.

As a data point Ethereum mining with the latest AMD GPU can earn about 
$6K a year in ETH if you run 7x24.
That's with hardware costing less than $1000.
Then you have power, cooling, housing and network access.

Some of these are dumb ideas and some are just engineering challenges 
the hard part is to figure out which is which.

Remember not long ago there was no way that you could build a rocket to 
take off  and land after putting stuff in orbit.
That was just the dreams of SF authors.

I truly appreciate the feedback.
> cheers,
>   Stewart
> ---
> Talk Mailing List
> talk at gtalug.org
> https://gtalug.org/mailman/listinfo/talk

Alvin Starr                   ||   land:  (905)513-7688
Netvel Inc.                   ||   Cell:  (416)806-0133
alvin at netvel.net              ||

More information about the talk mailing list