Decided to run my ebikes on solar

[scrambler]

It will help once data centers are moved out into space

Data centers currently consume approximately
4.4% to 4.5% of total U.S. electricity as of early 2026, driven largely by the rise of artificial intelligence. This usage is rapidly increasing from roughly 1.9% in 2018, with projections estimating that data centers could account for 6.7% to 12% of total U.S. power demand by 2028

Yeah but there are a lot of hurdles to overcome, heat dissipation being a huge one.... and of course maintenance....

 

[usclassic]

Yeah but there are a lot of hurdles to overcome, heat dissipation being a huge one.... and of course maintenance....

Maintenance yes but chips are reliable after the burn in period which they will do on earth but space is cold and cooling is free there. No water needed.

 

[usclassic]

SpaceX will get them there pretty soon.

 

[Chargeride]

Starlink sats burn up constantly, they just keep sending up more, they only last five years, two of them deorbit every day

 

[stompandgo]

It will help once data centers are moved out into space

Data centers currently consume approximately
4.4% to 4.5% of total U.S. electricity as of early 2026, driven largely by the rise of artificial intelligence. This usage is rapidly increasing from roughly 1.9% in 2018, with projections estimating that data centers could account for 6.7% to 12% of total U.S. power demand by 2028

They need to either be responsible for generating their own electricity, with surplus going to the grid, or off the grid entirely. This situation is already out of control, and it just started, with no end in sight.

 

[scrambler]

but space is cold and cooling is free there. No water needed.

Unfortunately NOT, space is a perfect vaccum so getting rid of heat is VERY HARD (no convection, no conduction nothing to take the heat), it acts as a thermos on satellites....

 

[scrambler]

They need to either be responsible for generating their own electricity, with surplus going to the grid, or off the grid entirely. This situation is already out of control, and it just started, with no end in sight.

Data centers and crypto are becoming the two biggest threats to our never ending ever increasing energy consumption

 

[stompandgo]

Data centers have always eaten power. Once corporations started migrating their business applications and data to the cloud, everything ramped up. Crypto mining has always used the same GPU's as AI, but the "datacenters" were distributed all over the world. Still a lot of power, but spread out. Building datacenters exclusively for AI is concentrating the power crush into your neighborhood. That's the problem. Every grid is not the same. AI datacenters should never be located in major metropolitan areas without their own power source, but they are, because stupid politicians think that they will be good for the economy and bring hundreds of jobs. During construction, maybe. After getting their CO, a dozen or two at most.

 

[usclassic]

Unfortunately NOT, space is a perfect vaccum so getting rid of heat is VERY HARD (no convection, no conduction nothing to take the heat), it acts as a thermos on satellites....

Cooling a space-based data center in a vacuum requires
using large radiators to radiate waste heat as infrared light, as convection is impossible. The primary solution involves mechanical pump fluid loops (MPFL) circulating liquids to transfer heat from computing hardware to external panels located in the shade

Key Cooling Strategies:

• Radiative Cooling (Primary): Since vacuum is an insulator that prevents convection, all heat must be radiated away. Spacecraft use large radiator panels that act as heat sinks to discharge heat into the -270 degree void.
• 
  Mechanical Pump Fluid Loops (MPFL): A mature, standard technology where a pump drives coolant to cold plates directly installed on high-power NVIDIA H100 GPUs or other processors, removing heat directly from the source.
• Massive Radiator Surface Area: Due to radiation being less efficient than terrestrial cooling, large-scale orbital data centers require massive radiator arrays to handle high heat loads.
• Heat Pipes and Thermal Materials: These passive systems are utilized to transport heat from the chips to the radiators.
• Managing Thermal Cycling: To avoid overheating, the data center or its radiators may need to be oriented in the shadow of the structure or Earth to avoid direct sunlight

 

[PedalUma]

Im starting to feel they might just turn them off and have a fire sale of chips.

With fish and chips on newsprint. But newsprint is so rare now, maybe served on your laptop. Cocktail Sauce with Fish & Chips is so much better than mayo based whippy slime.

 

[stompandgo]

Back to the subject. I have been looking into a system like this to power my fish pond. In season (soon), the filter pump runs 24x7 and draws 65W according to the manufacturer. I also use a small disc aerator that draws 15W. That's 80W total, 24x7. I have an area to mount a panel. I also have AC power, and a weather-resistant cabinet nearby. If I went with @usclassic's EcoFlow Delta 3 classic, a 1024w unit, with their 220W panel, at 80Wh, that would run them for 12 hours. That's overnight. However, it would be flatlined in the morning. The pond would still be drawing 80Wh while the panel was charging the unit. Would it catch up by nightfall?

I could double the panels, but I don't want my backyard looking like a satellite, and would 400W of panels be able to keep up?
I could plug the unit into AC power, but how does the unit balance the charging between the panel and the AC input to minimize the AC input?

These are the questions that I have had trouble finding answers for. I don't mind spending $700 for a system that doesn't take the pond completely off the grid, but I'd want to know by how much.

 

[Chargeride]

Non steered solar has a generation factor at the most 20%,.

 

[stompandgo]

Non steered solar has a generation factor at the most 20%,.

Sorry, I am not familiar with this terminology.

By "non-steered", do you mean fixed, in place panels that do not follow the sun?
By "generation factor", would that imply that for 80W I would need 320W of panels just to run the pond and have nothing left to charge the power bank?

 

[scrambler]

The current efficiency of solar panel is between 20 and 25%, this means that out of 100 watt that hit them they only produce 25W.
That said the rating of a panel describes the actual power it can output under ideal sun exposure. So a 320W panel can output 320W under ideal sun exposure and conditions.

The effective power it will produce will depend on the orientations, and the temperature.
Typically for me in Norh California, with panels oriented full south with a 20% slope, the highest production I can get is in full sun at mid-day, during the spring or fall when it is not too hot out there, and it is about 87%.
Under these conditions, my 320W panels will output around 280W.

Side note:
Panels can actually produce more than their rating under fringe condition, like during a cloud edge effect.
I have seen my 320W panels go up to 340W during this short term transition

 

[stompandgo]

Thanks, that helps. Take a look at this scenario. Let's say that a 200W panel can output 160W on a good day across 10 hours. With the 80W constant load, that leaves 80W/h for charging. That would leave me 224W short from grid independence every good sun day. I could add another 200W panel, giving me 320W on a good day, leaving me 240W/hr for charging, which should do the job. My question, and it's probably unit/charge controller dependent, is if I use AC power as a backup for charging on bad days, or to use just one panel, does it prioritize the solar input, i.e. use zero wall power, until it senses that the solar cannot keep up? This is the info I have had a hard time finding.

 

[scrambler]

First, the solar production throughout the day looks like a bell curve, so you dont get the full power all day.
Under my conditions, during the longest day of the year a 320W panel that produces 280W at best would produce about 2,200 Wh or the equivalent of almost 8 hours of full production (this is the best possible conditions)
During the shortest day of the year but still under optimal sun condition, that would drop to 1,200 Wh or the equivalent of about 4 hours of full production.
If I look at the full year I get 530,000 Wh for the panel, which averages to 1,450 Wh per day or the equivalent of 5.2 hours of full production daily.

As for your question regarding a multi-source charger prioritizing the solar input over the AC input, it is definitely possible but depends entirely on the Multi charger you are using.
I believe the Victron Multiplus can be configured to do that.

 

[usclassic]

Hot Coffee and noodles cooked with solar today. Hooked up the lights and 3D printer as well.

The new 900 watt microwave uses 1494 watts when cooking. It has eco mode that turns off display to save power when not in use.

 

[stompandgo]

First, the solar production throughout the day looks like a bell curve, so you dont get the full power all day.
Under my conditions, during the longest day of the year a 320W panel that produces 280W at best would produce about 2,200 Wh or the equivalent of almost 8 hours of full production (this is the best possible conditions)
During the shortest day of the year but still under optimal sun condition, that would drop to 1,200 Wh or the equivalent of about 4 hours of full production.
If I look at the full year I get 530,000 Wh for the panel, which averages to 1,450 Wh per day or the equivalent of 5.2 hours of full production daily.

As for your question regarding a multi-source charger prioritizing the solar input over the AC input, it is definitely possible but depends entirely on the Multi charger you are using.
I believe the Victron Multiplus can be configured to do that.

Thanks. This helps. I shut down the pump based on the freeze cycle, which varies year to year. Normally, it's from November to March, roughly the duration of Eastern Standard Time. It does not see the shortest days. During that time, I only run the 15W aerator.

I realize that the system I'm looking for has multiple variables that a simple solar array/power bank does not. What I don't want to end up with is a setup that can run off the grid during the day, but must be charged by the grid at night, defeating half of the savings. I'll look into Victron.

 

[scrambler]

Some of the integrated battery backup systems (like Ecoflow, Oupes, Bluetti, Anker and others) may also have such options in their managing App

 

[scrambler]

Below are an Excel sheet and word files I created a year ago for a friend (so not fully up to date) on these integrated solar backup generators and their use for trailer / Tiny living, in case it can be useful..
Solar equipment.xlsx
Sizing solar for RV.docx