Wolfnexus - Random Stuff
Solar powering a Raspberry Pi for Bitcoin mining - Part 3
Like this page? Donate with the QR code below!
Introduction - The Story Continues...:
After many months of waiting, I suddenly noticed my Butterfly Labs order for my two Jalapeno miners change from "Processing" to "Production". For those who haven't been involved with BFL as customers, this won't mean a great deal, but to those of us who have waited, it was the final light at the end of a rather long tunnel that was plagued with supply issues. It meant my units were finally on their way to me!
About a week after the order shifted from Production to Fulfilled (which took about 24 hours, I guess it included the burn-in and testing time too), I woke to find a USPS box jammed between my front door and security door (guess the local postal service here don't need signatures, irrespective of what the package says). Some quiet whooping and slightly confused stares from my girlfriend resulted as I unboxed the package.
Inside were two Jalapeno units, packaged carefully in their cardboard packaging, and two power supplies, each rated at 13V, 6A, made in China. Having seen some videos out there of teardowns of these units, they were definitely up to the job, but I seriously doubted they would survive long when supplying close to their theoretical maximum output. Fortunately, the two Jalapeno units I now had pulled nowhere near to this, as outlined in the main article below.
Some Quick Notes - The Power Supplies:
The power supplies shipped with my Jalapenos are clearly a third party manufactured unit, sourced from China. These do appear to have a CE logo (not to be confused with the China Export logo), and are billed as a "Replacement AC Adapter", presumably for laptops and similar devices. They are rated at an input voltage of anything between 100V and 240V, 50 or 60Hz, at approximately 2A maximum. They also bear the model number PC-C-090-130060CT.
These are supplied with an American NEMA plug to C7/C8 cable. While American power systems typically are 120V 60Hz systems, the plug itself is rated to 250V at 6A. The cable used appears to be H03VVH2-F domestic use flexible cable for appliances, and is rated to 300V at 6A. So in summary, the supplied cables and power supply should be suitable for typical power systems world-wide. I would recommend however that the mains to C7/C8 cable be replaced with one suitable for local power sockets, as many cheap adapters can be loose.
The Solar Power System:
This hasn't changed a great deal since I started the project, largely due to lack of funds (and ASICs). The Raspberry Pi itself has performed flawlessly on the supplied power for close on eight months now running continuously, and pulled about 5.7Ah in a 24 hour period. Unfortunately I did discover that the two 26Ah batteries that I had kept in service for about the last four years were severely degraded, and could no longer hold a charge even approaching 1Ah. I took them out of circuit, leaving the much newer 100Ah battery in-line for the time being, and began investigating what I could do about the batteries.
It did not look good. The two 26Ah batteries were both not far from their designated lifetime, and with their questionable background it was entirely possible that they would be irrecoverable. With limited funds available, I decided to see if they could be resurrected slightly before I took them to the recyclers.
First step, I got a small flathead screwdriver and gently levered up the "sealed" ports for the cells. These came in two banks of three, and consisted of a plastic cover, some sort of blotting paper underneath it and little rubber caps which made up the seals that prevented the electrolyte escaping. After gently levering the rubber caps out of their ports, I discovered why the batteries were holding almost no charge - the electrolyte had been mostly boiled off due to the years of being kept at close to full charge by the solar panels.
I got myself some distilled water and a simple syringe (you don't need the needled type usually), and started carefully refilling each cell, allowing the gel matting to soak it up. If you ever need to do this, be careful not to overfill any of the cells.
Once refilled, I put the caps back on the cells and gently tipped each battery back and forth to try and stir up the electrolyte a bit, then put them both on charge individually to test their capacity.
Still no good! :-(
Next step was to remove the caps again, and examine each cell for failure (not particularly easy, but a bright and narrow torch helps). No apparent signs of issues. I then got a tip from one of the users on the Butterfly Labs forums suggesting deep cycling the battery as the cells were probably unbalanced. This did make me a little nervous as I didn't want to muck up any of the cells by reversing their polarity, but with two apparently near-dead batteries there was not much to lose. I dug out an old 12V, 24W halogen lightbulb and set it up to act as a load while using some multimeters to measure battery voltage. The voltage on one battery dropped almost instantly when placed under load to about 10v, suggesting issues with one cell, while the other battery dropped to about 8-9 volts. Both batteries then held at that voltage for more than 12 hours, discharging through the halogen light.
I avoided discharging the batteries to dead flat, as most of the information I had read out on the net suggested this was one of the fastest ways to destroy a lead-acid battery. Instead, I discharged them both to about six volts, then placed them on a slow charge at about half an amp. Each battery then took about 18-24 hours to recharge, and the cells proceeded to outgas a bit, prompting me to check and refill them as necessary.
After repeating the process a few more times, I settled into a less violent discharge process, where I only pulled the batteries down to 10V before recharging them again. I repeated this process maybe a dozen times before the batteries finally stabilised, supplying on estimate about 18Ah, although this was rather difficult to measure. It is entirely possible given it was mid-winter at the time that I may have recovered considerably more capacity, but the way I view it, anything is an advantage!
With the new 90W panel bringing my generation capacity up to 170W, and the passing of the winter solstice, I'm hoping generation capacity will be enough to keep things running for some time yet, hopefully long enough to mine enough bitcoin to upgrade the system further by adding more panels and batteries.
The ASICs themselves - Power usage of a Jalapeno:
At long last, I was finally able to measure the true power usage of a Jalapeno unit. Almost all previous reviews of the units had taken power measurements from the wall, which include the inefficiencies introduced by the power supply itself. Measurements tended to vary by a bit too, so for my purposes getting an accurate measurement was important. I built my power supply out of one of the recovered batteries mentioned above, which had been charged to capacity. It had a resting voltage (open circuit) of approximately 13.3V
Results are below:
Jalapeno 1:
Idle state - V=12.2, I=1.1
Active state - V=12.0, I=2.5
Jalapeno 2:
Idle state - V=12.2, I=1.0
Active state - V=12.0, I=2.3
There were slight differences in the current draw of each unit, and this again varied based on the amount of work the unit was doing. The upside of this was that I now had a much better idea of what the units would require, and could revise my usage calculations from part 2, as below:
12 * 2.5 = 30
12 * 2.3 = 27.6
Total consumption (watts): 57.6
57.6 * 3 (taken from eight effective charge hours per 24 hour day) = 172.8
So I would need 172.8 watts of generation capacity to keep the two Jalapenos powered during the worst of winter. This pushes right hard up against my current generation capacity, and doesn't factor in the Raspberry Pi either.
Now for battery capacity. From the last entry, I estimated based on a 35W per Jalapeno basis that my 152Ah of battery capacity would be capable of powering the units for just over a day (25 and a bit hours). With the reduced capacity and load taken into account:
100/4.8 = 20.8333
So I should theoretically be able to run the units for almost 21 hours with no input. Not too bad, although it will be safer when/if I can introduce more battery capacity and maybe another solar panel.
Construction:
This has barely changed from the first page, apart from the addition of the new panel to the mix. The two Jalapenos are wired up by a 2.5mm barrel jack, which I have attached to a car socket plug. The socket I am using is fused for 10A, so it shouldn't melt down unless an electrical fault occurs, as the continuous current draw is about 5A.
From what I could gather from the design specs released by Butterfly Labs, the Jalapeno units could theoretically take an input voltage of more than 12V without suffering, however I suspect it would at minimum reduce the lifespan of the unit, and at maximum cause something to end up frying due to excess heat. Irrespective, it does mean they should be able to cope with higher voltages, as sometimes in particularly bright light when the batteries are fully charged the circuit voltage can spike to 18V. I doubt this will be a problem however, as it is unlikely the battery will ever be fully charged for long periods of time.
Software on the Raspberry Pi has not really been changed. I have updated BFGMiner to version 3.1.1, which works perfectly.
ITS ALIVE!:
I finally brought the whole thing online, and amazingly everything has worked perfectly. The solar charge controller registered the load correctly and the battery is holding firm at about 12.5V when cloud covers the sun. Even in these conditions, the panels are able to supply about 2 amps with no difficulty. Additional power requirements are met by the batteries, and I expect in bright sunlight the system will supply more than enough to recharge them again at least part of the way. The real test will be seeing how the system performs over the next few days, taking into account the weather conditions and charge of the batteries. I'll be keeping an eye on the controller to ensure there isn't a net loss of power occuring (which would mean my panels aren't up to the job). If this does become an issue I'll pull one of the Jalapenos out of circuit and put it back on my desktop PC (not as green, but no point losing the whole system for a while when it will perform happily with a single miner).
As of writing, the Raspberry Pi has been mining with the two Jalapeno units for an hour and a half. Pool statistics give me between 16GHash/s and 14GHash/s, which isn't unexpected, but I'll give it longer to sort out just how well the units are performing. Thanks to the cooler weather, neither unit has gotten warmer than 35-38c. I am planning to power down the units at night when it is coldest, let them cool, then power them up again, as this will ensure the self-test activates as many engines as possible and clocks them at a nice high speed.
Conclusion (or is it?):
It has been a long adventure, building and finally getting the system running. That said, there is still plenty of things left that can be done. One project I plan to embark on later is reflashing the units with the open source firmware available for the SC Single line of products. This should allow me to customise and scrape a bit more hashing power out of the units, as well as tweaking and playing with power consumption a little.
Another project I have in mind is trying to figure out a way to make the Raspberry Pi able to identify the voltage on the main battery. If I can figure this out I should be able to determine when the battery is low and in need of charging and get BFGMiner to either scale back or deactivate mining until it recovers. At the very least I could get it to email me to alert me that the system needs some attention.
If I can figure it out, I am pretty sure BFGMiner supports some level of RPC to derive miner statistics, which should allow me to put something on here showing how the system is performing. Keep your eyes open for future updates!
I almost certainly will be adding more panels and batteries to the system, if I can get my hands on them. Even once this project is dead and buried, the panels should still be useful for whatever next catches my fancy, and if well cared for the batteries should still be useable for years to come yet.
Comments, queries or thoughts? Please let me know using the comments section below. If you've found anything from this project helpful, interesting or informative and want to help me out, please drop a donation my way. My plan is to put donations and some of the mining revenue into further equipment and upgrades to the system where possible, with my next target being more batteries!
For Bitcoin donations: 1KPgXmNEmXmC1AtowjzmWgWfTS2QALp96Z
Or for Litecoin donations: LgXkAUTQLBKcU3VGLuEmroxNEFmXiw5TmV
Comments:
Maybe you could advise me if you dont mind. I have 300 watts of Solar panels. a 30 amp regulator/controller and 3 (4 soon) secondhand batteries ex 4x4 approx 100ah each (i think).Battery condition unknown. They charge and were running 12v lighting for 2 rooms and had no problem.
I am charging from a 240v 4 amp battery charger at the moment due to the fact i 'fried' the controller trying to fit an amp guage as i didnt know how to install due to my lack of knowledge. I had no probs fitting one in my car, but cant work out how to fit this volt/amp guage with shunt. i tried to place it between the controller and the batteries. Maybe you could advise me. Also, i run a Rasberry Pi and 1.6gh/s (5 miners) of Asic Miners through a reducer (12v -5v) directly from my batteries. how much mining equipment do you think I can run with my setup.
Should i go with jalapenos (i dont know anything about them) or do you think i should continue buying 333mh/s usb miners. Would I have to invert to run the jalapenos or can i feed them directly from batteries? I have a basic knowledge of electrical stuff, but i am self taught with no qualifications. So, if you reply to me, keep it simple please and I will understand. Again, thank you very much for your info as it is invalueable to me as it is hard to find info on solar Panels and bitcoin mining. have a good day - Hope to hear from you - Gerry
With that kind of setup, you can probably support a good 60w continuous draw of devices. As for the batteries, check what sort of batteries they are. AGM and other similar deep cycle batteries will cope with the regular charge and discharge a LOT better than starter batteries (which are designed to provide a heap of current very quickly to a starter motor).
As for measuring the current, that should be pretty simple. The thing to remember is that if you are measuring voltage, you measure across (in parallel with) the power supply. The resistance across a voltmeter will be very high. On the other hand, if you are measuring current, you measure in line with the circuit (in series). Ammeters should have almost no resistance.
Measuring volts - Parallel to the circuit
Measuring amps - In series with the circuit
As for the amount of mining equipment you can run, this really depends on what equipment you can source. The USB block eruptors are not particularly good value due to their low hashrate to cost ratio, but they are easier to source overall. Jalapenos will give a higher hashrate, but they do pull more power overall, as discussed on one of my other pages. This may well be offset by the fact they can be fed 12v directly without any external reducing circuitry (I have suspicions that they can actually handle quite a bit higher voltage than 12v, but I'm not really game to test)
When it comes to the usable capacity of your batteries, this can be a bit troublesome to measure. Your best bet is to charge them all to capacity, then connect a dummy load and measure both the voltage on the battery and current running through the load. Adding the time component will allow you to calculate the amp-hour capacity of your batteries. If you do this, try not to let your batteries discharge too far as excessive discharge will cause permanent damage. I carried out a similar process above in rejuvenating my older damaged 26Ah batteries, and it seemed to work quite well!