diy solar

diy solar

How to have off grid inverter supply power only if able to?

hidegety

New Member
Joined
Jul 27, 2023
Messages
14
Location
EU
Hi, totally new to PV. I have rather unique (maybe) situation: Got office in building, KWH costs are high, would like to power our server rack from solar as much as possible, optimized for money saving. Catch is we ourselves do not have contract with grid operator, we just pay rent (of which power costs are part) to building owner. The only option is to have strange off grid setup, where PV would be our main source and grid (stable and clean) our backup (grid plays a role of backup generator here). Goal is to produce and use as much power as we can, while having the PV 100% sandboxed, separated from gird and from office building power infrastructure.

Idea is simple: Panels connected to off grid inverter/mppt (or 2 separate devices, doesn't matter), some batteries mixed in, inverter supplies AC to ATS (which I got already anyway), if load exceeds inverter's ability to supply -> inverter AC output goes down (just output, not inverter itself, it needs to continue running to be able to charge batteries), ATS sees this and switches to grid. That far is more or less clear and simple. Now -> when inverter is able to supply the load 100% on its own (remember, there is no grid tied), it should turn its output back on, ATS sees the preferred source (inverter) is back on and switches back to inverter's AC output and keeps it as long as inverter is able to fully supply the load.

But how to "inform" inverter with output switched off about current load, if the load is on grid and inverter does not see it? Inverter needs to know if it will be able to supply the load (which varies). I imagine some kind of power meter between ATS and load feeding current consumption info to inverter. But what kind of meter? What is the inverter feature that enables this? What is the protocol? Is there some documentation for this?

I attach drawing, just simplified draft, forget about internals of PV part (grey marked), what I ask about is the ability of inverter to supply AC only if able to bare the load on its own.
Tx in advance for shedding any light on this topic.
 

Attachments

  • OfficePVDraft.jpg
    OfficePVDraft.jpg
    185.5 KB · Views: 7
I have just purchased a cheap all in one inverter , 5500w, 48v Edecoa

It (and I assume most others aswell) has priority settings, two options , SBU or SUB

In order of priority:

SBU: solar, battery, utility

SUB: Solar, utility, battery


SUB should be suitable for your needs, running directly from solar PV when available, if not then loads of fed by utility power . And only in the case of a power outage will the inverter draw battery power




Please ensure model is suitable for your locality
 
Last edited:
I have just purchased a cheap all in one inverter , 5500w, 48v Edecoa

It (and I assume most others aswell) has priority settings, two options , SBU or SUB

In order of priority:

SBU: solar, battery, utility

SUB: Solar, utility, battery


SUB should be suitable for your needs, running directly from solar PV when available, if not then utility power . And only in the case of a power outage will the inverter draw battery power




Please ensure model is suitable for your locality

There are many AIO (all in one) inverters available, 'MPP solar' are a good brand name to consider.


Although your solution with the transfer switch may be a viable option, an AIO inverter would be the simpler choice
 
That does not sound like an ideal situation.

First, the setup is not off-grid under European regulations. Usually the inverter "hands over" to the grid when solar power is not available. For that, it is connected to the grid as well as to the solar arry (via own MPPT or external charger) and/or battery. Since you're in the EU says your avatar, such a grid-parallel setup can not be done DIY and if you do you must find a professional to configure, connect and legalize the installation. That is ... difficult to impossible under the current shortage of manpower.

Even a manual transfer switch is illegal if not done professionally at least in some countries (Germany for instance, they do renewables very ... thoroughly).

If there is any chance to completely power your servers from pv power without any connection to the grid, that would be the way to go DIY. But such an installation needs to be big enough and needs all components including batteries to power it, which may negate the roi. Winters can be dark in the north, and the house owner will probably not approve of it because it is so ... individual, unique, special:)
 
Thanks for your replies and your time.

From legal pov there are 2 codes that apply here:
1: General technical regulations for safety and security of any electrical device or installation.
2: Energy related. This one is there to protect the grid.
Off grid solution is not affected by energy regulations, only by technical norms.

AIO inverters are nice but in my case the whole solar shenanigans must be off grid. Inverter can not see the grid, cannot be plugged in, can be powered only by solar or batteries. As @SpaceJellyfish says, inverter with internal ATS is not considered off grid. Inverter directly messes with power transfer in order to monitor it, it is therefore grid tied.

ATSes are not illegal in EU (Slovakia in my case), I personally have one from APC (AP7723) which works flawlessly, I use it to power the rack from 2 different phases, it is the one I plan to use for solar as well. It is precisely ATS that guarantees that whenever grid will be powering my load, it will be completely segregated from internal solar, it is either or.

Even though I will do this DIY, the inspection will be done by professional electricity inspection agent. The last thing I want is fire.

All I need is the way to turn on the inverter when it is able to bare the current load. I read somewhere something about inverters with external current limiting sensor -> is that what could "inform" the inverter about current load? Does someone know exact model?
 
Maybe I forgot to underline 1 detail: The decision which power source will be used lies not in inverter but in external ATS. And ATS uses only the power source availability to decide which source will be used. There is therefore no connection between grid and solar.
 
Probably not the most efficient option, but its simple and minimal monkeying to make it work well.

I run offgrid with a twist. all schneider equipment, Im sure other brands have the option to prioritize which power to use first.

2x 100-600 mppt hooked to panels, charging mode set to primary. this is used if its producing power.
60-150 mppt being fed by a 4kw telecom smps (fed by the grid) this is only used if there isnt enough power from solar, and only to make up the difference.

As far as the utility knows I am charging a battery with their power.

Some care has to be taken selecting the voltage of the smps, with lead acid at 0c you need about 59 volts to bulk charge and even more if you equalize. my mppt also has a voltage drop of about 1.5 volts

All this combined means even in summer my 58v power supply cant properly finish absorption mode.

You also probably dont want to go too high of a voltage since the psu is likely right beside the mppt with short fat wires and with all that capacitance on short wires (and minimal inductance) the mppt "could" fry potentially if its getting more than the bare minimum voltage it needs.
 
While looking around I found the awesome simple device - voltage monitoring relay. Very reliable. I may simply hook it up to monitor batteries / solar voltage, when dropped below some cuttof (say 42V in my 48V setup) and come back on again when voltages are above (say 50V). Inverter will always produce (given it has solar or batteries). Just the external relay will decide if that power produced by inverter will be used. And after this relay will sit already previously mentioned ATS -> whenever it will see inverter on (turned on by monitoring relay) it will use it. Otherwise it will use grid.
This way grid and solar will remain completely segregated, inverter will not need to know anything about the load and grid anything about my internal solar setup.
What do you think?
 
For a server, what you really want is a double conversion UPS. This has 0ms transfer time. With your proposed system, you are 3/4 of the way there. what you need is NOT a transfer switch, but a Battery Charger that has the wattage in excess of your server needs.

Instead of switching to Grid when batteries get low, you turn on the battery charger. It will power your inverter and charge the battery. You have it turn on at 20% SOC, and turn off at 30%. If an extended bad weather is forcasted, you can charge up the battery to full, just incase the grid goes down, and you are not producing solar.

Since you only use the grid to charge the battery, your system may avoid all the grid connection problems. If you have time-of-use pricing, you can do some fancy logic to recharge from grid when prices are low.
 
@DIYrich, if I understood your proposal correctly - are you suggesting just simple plain old UPS with extended battery packs? I already considered that and discarded the idea due to the price and max battery cycles. UPSes are intended to kick in when grid goes down, not to cycle regularly. Or do you mean replace UPS batteries with something more beefy, like LIFEPO4?
 
@DIYrich, if I understood your proposal correctly - are you suggesting just simple plain old UPS with extended battery packs? I already considered that and discarded the idea due to the price and max battery cycles. UPSes are intended to kick in when grid goes down, not to cycle regularly. Or do you mean replace UPS batteries with something more beefy, like LIFEPO4?
There is no way to tell if the solar can support your load, even if there was are you going to switch back and forth for each and every cloud that goes by?

What diyrich and I mentioned is the only way:

Server powered off of inverter and battery bank

battery bank charged by solar, and when solar cant keep up a grid powered battery charger supplies the power.
 
@DIYrich, if I understood your proposal correctly - are you suggesting just simple plain old UPS with extended battery packs? I already considered that and discarded the idea due to the price and max battery cycles. UPSes are intended to kick in when grid goes down, not to cycle regularly. Or do you mean replace UPS batteries with something more beefy, like LIFEPO4?
No, I'm saying your system is 3/4 of the way to being a double conversion UPS. Add a grid charger for the battery, and you have the equivalent of a double conversion UPS.

PV-Inverter-Battery <- Grid

Grid only goes to the Battery, so you have no issues with grid connected compliance.

"double conversion" because the Power from the Grid is converted to DC before being converted back to AC for the Servers.
In a stand-by UPS, the grid power is passed through, and the UPS interrupts the grid if it detects anomalies. This takes time 10-20ms typically.
With double conversion, any grid problems, and it stops charging the battery. Inverter continues to draw from the battery, no power interruption.
You might get some ripple, which is not great for the battery. Some capacitors should solve that problem (which may already be in the inverter).
 
I already considered this option. Technically it makes sense. But before I put together all parts of setup in sufficient capacity I would feed grid through batteries -> resulting in losses -> resulting in partially nullified savings. On top of it I would chew through battery cycles faster -> furthermore delaying ROI. If I am on grid I need no losses, just pass-through. Which makes it impossible to use inverter for that (since then legally it is not off grid anymore).

I am again and again comming back to this external SOC monitoring and inverter output shutdown.

Let's suppose that I do this with voltage monitoring relay. If there are 2 sources (FV and batteries) what would be good general approach to determining when to switch off inverter output? Just go above min and below max voltage of batteries (according to batteries documentation)? What would solar panels (with all their clouds and stuff) do to voltage that would mess up the measurement?
 
The inverters shut down automatically when batteries cross a threshold. Batteries with a bms have their own safety functions, like temperature, individual cell balancing, charge and discharge voltage and current. Make sure battery and inverters can communicate via a common protocol.

2 cents from a Spaniard with a DIY off-grid-system:
Do check your local regulations. Under normal EU regulations, which are made for a continet spanning network of connected installations:

Suggestion 1:
Size your system (panels, battery, inverters) so that they can independently power your servers. There must not be a connection to the grid, neither through a manual or automatic switch or any inverter that has an AC input. That means an own set of AC cabling for your case. Clarify with the landlord who might not approve of someone doing changes to their property.
Maybe there's a barn around for sale where you plaster the roof with panels and that does not even have a grid supply. You can do DIY without anyone asking questions. This is most certainly much more expensive than relying on grid supply.

Suggestion 2:
Do a grid parallel installation. You need an installer. Clarify with the landlord. Maybe you can bargain with them that you buy stuff and do the panel mounting and DC cabling because that's not rocket science. But it also cuts on their margin ...
You don't need a battery, possibly saving a lot of money. You need a grid connection, and when the sun shines you can sell your production, possibly lowering the bill.

Suggestion 3:
Similar to 2, but no sale of your production. If the whole installation is below a threshold (that's I believe 30kWp panels in Germany, 10kW inverter output in Spain) you'll have a somewhat simplified bureaucratic process to master:) which might be worth a thought or two.

Suggestion 2 and 3 also offer the possibility of subsidies (if you are the owner), which is much more difficult under 1 because you need to find a professional, registered installer or engineer to stamp your installation. My experience ist they don't, they all have more than enough work and some don't look fondly on DIYers.

In any case and independently there maybe a lot of saving to be made on the consumption side, the servers. Efficient processors and ram configurations or even a platform change might help saving energy and cost. Might be worth a thought or 2 ...
 
Last edited:
Let's suppose that I do this with voltage monitoring relay. If there are 2 sources (FV and batteries) what would be good general approach to determining when to switch off inverter output? Just go above min and below max voltage of batteries (according to batteries documentation)? What would solar panels (with all their clouds and stuff) do to voltage that would mess up the measurement?
For LiFePO4 batteries, I would set the disconnect at somewhere between 45-48v (5-10% SOC), and reconnect around 51.5-52.0v (30-40% SOC). Or whatever gives you at least 1 hour of run-time on batteries alone.

The battery would reach the reconnect voltage while charging. If it were just above the reconnect voltage, and then went dark (evening), you would have some battery run-time before cutting off for the night.
 
Thank you all for your suggestions and ideas.

@SpaceJellyfish
Unfortunately Scenario 1 would be too expensive. We would be covering peaks with capacity we usually don't need. Saving costs is our only motivation. Sure, we can put together setup for €15k and be 100% off grid, but ROI is like 15 years which is too long.
Scenario 2 and 3 are grid tied and thus off the table. We are not the end user, there is no contract with supplier or grid, just rent. Property owners are just forwarding their power costs to all renters, so there is no motivation from their side to block this. They just don't want us to mess with their installation (which I totally understand and respect) -> hence 100% off grid (with grid as backup, not as supplement).

@DIYrich
Inverter switching off is not a problem, all modern off grid inverters can do that (it's just solar/batteries voltage drop below trashold). So you suggest that inverter will not sense the load at all and just switch the output on, whenever batteries will be full enough to be able to power our base consumption (which is very stable) for some minimal time? That is simple and it could work. It would prevent starting with just panels + inverter and later supplement all with batteries. ... ... ... actually no, it would't. Just without batteries it would switch off and on for each cloud - which is not a problem, ATS will handle the swings just fine.
 
It would prevent starting with just panels + inverter and later supplement all with batteries. ... ... ... actually no, it would't. Just without batteries it would switch off and on for each cloud - which is not a problem, ATS will handle the swings just fine.
I would suggest some minimal amount of batteries. Constant switching is not good.
 
I just reread the whole thread and I must admit I missed (well not rly understood) what @2strokeforever suggested.
2x 100-600 mppt hooked to panels, charging mode set to primary. this is used if its producing power.
60-150 mppt being fed by a 4kw telecom smps (fed by the grid) this is only used if there isnt enough power from solar, and only to make up the difference.
If I understand this correctly you would make the junction in charging batteries, 2 MPPTs that talk to each other, one of them primary charger (solar), the other one secondary (grid). Then inverter from batteries to AC that powers the load. Only source here would be inverter, but batteries would have 2 sources, not the ATS. That is interesting idea. Only downside is that going always via batteries means relatively significant loss on {AC->Battery->Inverter->Load} double conversion. That woud prolong ROI, it is optimized for availability (I have separate UPS for that) and not cost saving.
Fresh approach nonetheless.
 
I would suggest some minimal amount of batteries. Constant switching is not good.
Why not? The ATS is designed exactly for that. It has also monitoring period, after the preferred source comes back online (cloud goes away) it "monitors" its stability and only after 30 seconds (or thereabout) switches back to solar.
Sidenote: The monitoring period is not always 30 seconds, it is just like maximum, sometimes it switches back after just several seconds (like 5-8). I always suspected that the monitoring is not rly monitoring, but rather sine wave synchronization. It must be done utilizing slowing down or speeding up the secondary sine wave, unit it matches the primary, the difference must be rly small, like delta 0,01 Hz -> not to go off the tolerances. The duration depends on how large the sine wave shift is. The switch is according to data sheet max 8ms, with 100% power source segregation, none of the server power supplies seem to notice (or have problem with that).
 
Everytime you switch, there is a glitch in the output wave form. Not good for sensitive devices.

Switches have a limited number of times they are designed to switch during their operating life.
 
Back
Top