diy solar

diy solar

DIY Solar for a mining rig

hoya

New Member
Joined
Jul 4, 2022
Messages
12
Hi,
I am planning to build a DIY solar system for my small mining rig:
.
Requirment:
It is a rig with only 8 GPU cards with max 4,100W power requirement, I wish to have it running 24 hours a day, thus it would need 4,100w x 24h = 98.4kWh / day
.
I am do planning basing on and by referring to this video:
.
My place has sunshine hours ranging from 6 hours in Winter to 9.5 hours in Summer, I have some questions:
.
1. For planning, should I use 6 hours of sunshine or to use average say 7.8 hours or to use 9.6 hours for the calculation/estimation?
.
2. Within the maximum W allowed, how will the system (EG448VDC/120VAC Model: UL1741) handle the case when supply W (from solar panels) is greater than demand W (from the rig) at a time?
.
3. Since the batteries are expensive (the most expensive part), how to correctly calculate the number of batteries needed (to be specific, I plan to use EG4 48V 5kWh battery model: UL1973) so to have the rig running 24 hours a day for 365 days a year?
.
Can I use the following estimation?
Total power needed a day: 98.4kWh
Sunshine a day - 6 hours (directly power supply from solar): 4100w x 6 = 24.6kWh
Batteries requirement: 98.4kWh - 24.6kWh = 73.8kWh, thus 73.8 kWh / 5kWh = 15 pcs of server battery (rounding from 14.76)
.
My estimation:
EG4 48VDC/120VAC Model: UL1741 => 2 sets (max 16,000w input from solar panels)
EG4 48V 5kWh battery model: UL1973 => 15 sets
EG4 Server rack: 3 pcs (max 6 server batteries)
Other parts
.
Please help and advise and comment. Many thanks
 
Last edited:
Hi,
I am planning to build a DIY solar system for my small mining rig:
.
Requirment:
It is a rig with only 8 GPU cards with max 4,100W power requirement, I wish to have it running 24 hours a day, thus it would need 4,100w x 24h = 98.4kWh / day
.
I am do planning basing on and by referring to this video:
.
My place has sunshine hours ranging from 6 hours in Winter to 9.5 hours in Summer, I have some questions:
.
1. For planning, should I use 6 hours of sunshine or to use average say 7.8 hours or to use 9.6 hours for the calculation/estimation?
.
2. Within the maximum W allowed, how will the system (EG448VDC/120VAC Model: UL1741) handle the case when supply W (from solar panels) is greater than demand W (from the rig) at a time?
.
3. Since the batteries are expensive (the most expensive part), how to correctly calculate the number of batteries needed (to be specific, I plan to use EG4 48V 5kWh battery model: UL1973) so to have the rig running 24 hours a day for 365 days a year?
.
Can I use the following estimation?
Total power needed a day: 98.4kWh
Sunshine a day - 6 hours (directly power supply from solar): 4100w x 6 = 24.6kWh
Batteries requirement: 98.4kWh - 24.6kWh = 73.8kWh, thus 73.8 kWh / 5kWh = 15 pcs of server battery (rounding from 14.76)
.
My sstimation:
EG4 48VDC/120VAC Model: UL1741 => 2 sets (max 16,000w input from solar panels)
EG4 48V 5kWh battery model: UL1973 => 15 sets
EG4 Server rack: 3 pcs (max 6 server batteries)
Other parts
.
Please help and advise and comment. Many thanks
For one, your 8x rig will not draw 4100w. Thats over 500w per GPU which doesn't exist. Most 8x rigs will be in the 1500-2000w range.

To save you some time, when you get past the 2000w range, it is not feasible unless you are doing a commercial size solar system, and you do not use batteries.
 
Thanks!!

(Correction)
.
Requirment (try to use MAX):
It is a rig with 8 GPU cards, each consumes 340w (max) plus CPU/board loading at 240w (max):
Total W of rig: 340w x 8 + 240w = 2,960w (max)
24 hours a day: 2,960w x 24h /1000 = 71.04kWh / day
.
That is:
With max 2,960W a day assumed, I wish to have it running 24 hours a day, thus it would need 2,960w x 24h = 71.04 kWh / day
.
My place has sunshine hours ranging from 6 hours in Winter to 9.5 hours in Summer, I have some questions:
.
Question 1. For planning, should I use 6 hours of sunshine or to use average say 7.8 hours or to use 9.6 hours for the calculation/estimation?
.
Question 2. Within the maximum W allowed, how will the system (EG448VDC/120VAC Model: UL1741) handle the case when supply W (from solar panels) is greater than demand W (from the rig) at a time?
.
Question 3. Since the batteries are expensive (the most expensive part), how to correctly calculate the number of batteries needed (to be specific, I plan to use EG4 48V 5kWh battery model: UL1973) so to have the rig running 24 hours a day for 365 days a year?
.
Question 4: Will the following estimation be correct?
Total power needed a day: 71.04kWh (2,960w x 24h /1000)
Sunshine a day - 6 hours (directly power supply from solar): 2,960w x 6h / 1000 = 17.76kWh
Battery storage requirement: 71.04kWh - 17.76.6kWh = 53.28kWh, thus 53.28 kWh / 5kWh = about 11 pcs of server battery (rounding from 10.656)
.
My estimation:
1. EG4 48VDC/120VAC Model: UL1741 => 2 sets (max 16,000w input from solar panels)
2. EG4 48V 5kWh battery model: UL1973 => 11 sets
3. EG4 Server rack (max 6 server batteries) => 2 sets
4. Solar panels
Option 1: Solar panel (assuming 200W panel): 71.04kWh / 6h * 1000 / 200W => 60 sets (rounding from 59.2), or
Option 2: Solar panel (assuming 570W panel): 71.04kWh / 6h * 1000 / 570W => 21 sets (rounding from 20.77)
5. plus other parts
.
Please advise if above estimation correct? Many thanks
 
Assuming your 2960W is correct then you do have 71.04kWh/day. But that's through an inverter so assume an 85% efficiency. That means you will need 71.04kWh / 85% = 83.58kWh of usable battery capacity.

A 48V 100Ah LiFePO₄ battery is 5120Wh. Assume no more than 90% of battery usage means 4608Wh of usable capacity per battery.

Your calculations then go on to assume you will be able to run your system for 6 hours per day on solar in the winter. Very unlikely. That would require 83.58kWh / (6h/24h) = 20.9kWh of solar production over a 6 hour period in the winter. That's an average of 3490W over those 6 hours. Due to the low sun angle and short day you probably need 16kW-17kW of solar panels to produce 20.9kWh in the winter.

Your calculations also assume it will never be cloudy. Of course you can supplement your solar with grid power on days you come up a bit short.

So even if you somehow manage run off of solar 6 hours per day you still need 71.04kWh / 18h/24h = 53.28kWh of usable battery for the other 18 hours. 53.28kWh / 4608Wh usable per battery means 11.56 batteries. Call it 12 batteries.

$18k in batteries minimum. 16kW of solar panels is probably $8k - $12k. Plus the inverters and racks. How much is electricity? How long would it take to break even on this equipment versus the cost of electricity?

Again, that assumes sun every day. More batteries and solar to deal with clouds or supplement with grid power when needed.
 
Many thanks!!

From the estimation, I can see the most expensive part would be the storage of electricity.
 
May I know how the system (EG448VDC/120VAC Model: UL1741) handles the case when supply W (from solar panels) is greater than demand W at a time?

e.g.
For an EG448VDC/120VAC Model UL1741, within the maximum W allowed (i.e. 8,000W), if input from Solar panel is 7,200W, at the same time the appliances (and batteries) only consume 6,000 W, how will it be handled?
 
Last edited:
Solar panels do not push power. The solar charge controller only draws the power it needs from what the panels are capable of providing. If you were drawing no power and the batteries were full, everything would be fine even if you had 20kW of panels sitting out in the bright sun. it would simply be a case of lots of potential solar power going unused.
 
The question you have to ask yourself is if it's worth it (if you gonna ROI). I don't know you mining hardware specific but from 8 GPU I can estimate 19$ a day if you're 8 GPU are 3090Ti. Meaning it will easily take over 4 years to ROI (assuming no difficulty adjustment and assuming ETH never switch to proof of scam). But they are supposed to totally switch soon meaning ETH will not be minable anymore and all the miner will switch to other shitcoins, diminishing there ROI too.

I do mine too, but I made my system differently : I built it for my need (excluding mining) and for worst case scenario... But normally I still have 8 month-ish without barely any clouds, located between the tropics. I just use mining to burn out the surplus of electricity I generate and I made a system so the mining switch when the battery is at 50% (under 50% = mining turn off, over 50%, mining turn on). That way the mining money is bonus without any ROI goal.

Lt.Dan Solution seam to be the best option for 24/7 mining. No batteries, get some grid tie inverter and just use the "grid" as a battery, Will divide the system cost by two and will bring you're ROI down to 2Yr+ instead of 4Yr+
 
Last edited:
To make it very simple, I have done this math before, and I currently GPU mine as well on my own solar. The only way to do this right is if you do a large 18+kW solar array and be Grid-Tied with no battery. Get all of your credits during the day, and use them up at night.

Even getting a professionally installed 18kW Grid Tie system is cheaper than buying all the batteries you want to get.
 
I made a mistake for my calculation

it's more 8 years ROI for full off grid system and 4yr ROI for using the grid as battery
But then the ROI should be calculated taking account how much you pay in electricity (calculate the ROI based on the Bill you don't have to pay).

So if you are paying 10$ a day to run the rig, then the actual ROI for offgrid would be more, 16 years (and half of that for using the grid as battery)
 
I made a mistake for my calculation

it's more 8 years ROI for full off grid system and 4yr ROI for using the grid as battery
But then the ROI should be calculated taking account how much you pay in electricity (calculate the ROI based on the Bill you don't have to pay).

So if you are paying 10$ a day to run the rig, then the actual ROI for offgrid would be more, 16 years (and half of that for using the grid as battery)
This is very true, your ROI will depend on your current electricity cost. My cost here is $0.36/kWh all in with fees/taxes. This makes my ROI far faster than the national average of $0.10-0.15/kWh
 
You're missing many other items. Your A/C requirements are going to double/triple since you're essentially running a space heater 24x7. This gets into PUE (power usage efficiency). E.g., you need may need 2.5 watts in order to consume 1 watt of power for your mining rig (A/C cooling, inverter inefficiencies, wiring losses, voltage drop, shading/clouds, bad weather, battery losses, additional lighting/LEDs, additional surge suppressors, and many other things I'm not even considering). For example, that 25kW solar system you may be planning really should be 62.5kWh. At this size, you're at the point of a "commercial" solar install, since you're looking at hundreds of panels, dozens of inverters, and enough copper wire to circumnavigate the globe a few times.
 
You're missing many other items. Your A/C requirements are going to double/triple since you're essentially running a space heater 24x7. This gets into PUE (power usage efficiency). E.g., you need may need 2.5 watts in order to consume 1 watt of power for your mining rig (A/C cooling, inverter inefficiencies, wiring losses, voltage drop, shading/clouds, bad weather, battery losses, additional lighting/LEDs, additional surge suppressors, and many other things I'm not even considering). For example, that 25kW solar system you may be planning really should be 62.5kWh. At this size, you're at the point of a "commercial" solar install, since you're looking at hundreds of panels, dozens of inverters, and enough copper wire to circumnavigate the globe a few times.
Don't ever air condition a mining rig. High CFM is all you need.
 
Don't ever air condition a mining rig. High CFM is all you need.
It's going to produce a lot of heat for the rest of the house. Unless you put it in an outdoor shed. Even then, you need some sort of fan/cooling/air-flow device. Which is additional electricity overhead.
 
It's going to produce a lot of heat for the rest of the house. Unless you put it in an outdoor shed. Even then, you need some sort of fan/cooling/air-flow device. Which is additional electricity overhead.
I would never put something like this inside of a home, its especially not feasible because not everybody has a 240v outlet in their bedroom. Its assumed once you get past the 1000-1200w mark, it gets placed elsewhere. Mine is in my shed. Pumping out 1800w, in 130* ambient temps, and it works just fine. 1x 16" Ventilation fan running at 80w is all thats required.
 
I am going to place the rig in a separate shed in my farm, it is an off-grid one.
.
Other cost (my case) :
I will have a separate shed for the installtion of
.
1. 2 sets of EG4 48VDC/120VAC Model: UL1741
2. 12 sets of EG4 48V 5kWh battery model: UL1973
3. 2 sets of EG4 Server rack (max 6 server batteries)
.
Also
4. Few acres of land to install the solar panels
Option 1: 60 sets of Solar panel (assuming 200W panel): 71.04kWh / 6h * 1000 / 200W (rounding from 59.2), or
Option 2: 21 sets of Solar panel (assuming 570W panel): 71.04kWh / 6h * 1000 / 570W sets (rounding from 20.77)
.
All these for the 24x7 running of a rig with 8 GPUs, say 3080, or 3080ti or 3090 (not 3090ti)
.
End up, I might be of running a commercial grade solar farm, but a bit different: I am the electricity producer and the consumer.
 
Last edited:
Thanks DestroyFX for your comment!!
Thanks Lt.Dan for your comment!!
Thanks Cyanlite for your comment!!
Thanks rmaddy for your comment!!
 
Last edited:
It's going to produce a lot of heat for the rest of the house. Unless you put it in an outdoor shed. Even then, you need some sort of fan/cooling/air-flow device. Which is additional electricity overhead.
Thats why I am thinking
.
1. Day-time, it will be hot:
Use (extra) solar (panels) to directly power the cooling system during day time
(like this video:
)
.
2. At night, it is not hot
(do nothing)
 
Hi,
I am planning to build a DIY solar system for my small mining rig:
.
Requirment:
It is a rig with only 8 GPU cards with max 4,100W power requirement, I wish to have it running 24 hours a day, thus it would need 4,100w x 24h = 98.4kWh / day
.
I am do planning basing on and by referring to this video:
.
My place has sunshine hours ranging from 6 hours in Winter to 9.5 hours in Summer, I have some questions:
.
1. For planning, should I use 6 hours of sunshine or to use average say 7.8 hours or to use 9.6 hours for the calculation/estimation?
.
2. Within the maximum W allowed, how will the system (EG448VDC/120VAC Model: UL1741) handle the case when supply W (from solar panels) is greater than demand W (from the rig) at a time?
.
3. Since the batteries are expensive (the most expensive part), how to correctly calculate the number of batteries needed (to be specific, I plan to use EG4 48V 5kWh battery model: UL1973) so to have the rig running 24 hours a day for 365 days a year?
.
Can I use the following estimation?
Total power needed a day: 98.4kWh
Sunshine a day - 6 hours (directly power supply from solar): 4100w x 6 = 24.6kWh
Batteries requirement: 98.4kWh - 24.6kWh = 73.8kWh, thus 73.8 kWh / 5kWh = 15 pcs of server battery (rounding from 14.76)
.
My estimation:
EG4 48VDC/120VAC Model: UL1741 => 2 sets (max 16,000w input from solar panels)
EG4 48V 5kWh battery model: UL1973 => 15 sets
EG4 Server rack: 3 pcs (max 6 server batteries)
Other parts
.
Please help and advise and comment. Many thanks
Sorta off topic but can you explain to me how this mining thing works and why so intense ? And gpu? I don't get it
 
Sorta off topic but can you explain to me how this mining thing works and why so intense ? And gpu? I don't get it
Distributed computing
GPU have many shaders units usually more than 2000 of them. They normally where designed to compute pixels but instead in the case of mining it allow processing > 2000 computations per cycle. Massively parallel compared to just using a CPU.

So it come done to run the hardware at full power, full time, The more computation you do, the more reward you get.
Different internet magic coins have different mining algo. Say, SHA-256 for Bitcoin and most shitcoins, Scrypt for Litecoin, Kawpow for Ravencoin, RandomX for Monero, etc etc. Some Algo are more suitable for ASIC mining (specialized hardware that can only do that), Some are more suited for GPU and some for CPU.
 
Back
Top