Author Topic: DC Failover for Solar Power  (Read 6485 times)

0 Members and 1 Guest are viewing this topic.

Offline Lukas52Topic starter

  • Contributor
  • Posts: 21
  • Country: de
DC Failover for Solar Power
« on: July 22, 2023, 09:22:55 am »
Hello!

Im a bit stuck at my biggest project so far.

I am currently installing some solar panels, with the goal being to power my home servers and 3D printers straight of my massive 24 Volt LiFePo4 Battery. I already figured out the charging, the BMS and the power distribution plus a lot of safety stuff.
What prevents me from sleeping at night is the question: "What happens if my battery gets empty". Theoretically i don't have enough solar power during the winter months, so its a real problem.
I have come up with two plans, but don't know which one is the "proper" of doing it...

Plan A:
Use a AC powered battery charger that gets switched on via a relay if the battery drops bellow 20%. Possible by using a programmable output on my solar chargers, but I'm not sure if that stays on when there is no solar power... (I am planning in using two Victron MPPT 250/70). The efficiency would be kind of bad tho and could possibly put unnecessary cycles on the battery, also it would need to be a pretty big charger.

Plan B:
Use 24 Volt DC Power Supplies. Switched using the same programmable output as in plan A. Problem with that: I am struggling to find high current DC relays that don't cost a fortune, i would also need to add considerable capacitance to prevent the voltage from dropping to far down during switching, so i don't overwhelm my power supplies. 

How is this done in commercial installations?
Any recommendations?

Thank you :)
 

Offline rteodor

  • Regular Contributor
  • *
  • Posts: 122
  • Country: ro
Re: DC Failover for Solar Power
« Reply #1 on: July 22, 2023, 11:12:14 am »
One simple and reliable option is to use 2 beefy diodes: one from the LFP battery, another from a mains power supply.
Use a mains PSU with tunable voltage and set it to ~25V (that is for 3.1V/cell). In this way you have a soft power drain transition from battery to mains. Cheap solution and works satisfactory if done right. The BMS should protect the battery when it is empty and there is no mains.

The problem with this solution is that you have no automated control of the transition. You need the SoC value, either from BMS or from a Victron Smart Shunt or something similar. With that you can automate turning turning ON/OFF the mains PSU so you can properly control the remaining SoC (in case you want to control the depth of discharge and/or keep reserve for power outage).

Avoid high current relays especially for continuous use. Leave them for battery protection only and use MOSFETs. Or use power switches (Victron, 123Electric, etc.) yea, yea expensive ... I know.
 
The following users thanked this post: Lukas52

Offline Lukas52Topic starter

  • Contributor
  • Posts: 21
  • Country: de
Re: DC Failover for Solar Power
« Reply #2 on: July 23, 2023, 06:23:19 am »
That is elegantly simple!

Would i even need control over the remaining battery capacity?
If i understand this correctly the diodes would prevent reverse charging, so all i need to do is turn the AC PSU on using the programmable relay on the victron charger, and do some math regarding voltage drop at the diode and cell voltage.

For example: The Cells im using a rated for 6000 Cycles when charged between 3,65 and 2,5 Volt. I set my BMS to be a bit gentler to hopefully squeeze some more life out of them by using 3,50 Volt and 2,7 Volt instead. Should still give me 80-85% capacity.
To keep some reserve i would want to switch over at say 3 - 3,1 Volt. That would be 24 or 24,8 Volt, add too that the Diode Vfr of 600 mV -> Set PSU to ~25,2 Volt.

As soon as my battery voltage goes bellow this limit the diode should prevent current to flow from the battery shouldn't it? Since the rail Voltage would be higher thanks to the power supply than the battery voltage.

The only problem i can see is a scenario where the sun comes up on a cloudy day, causing a lot of switching once the battery voltage rises again...

Maybe a combination of both methods would make the most sense... as in disconnecting the battery and Solar Chargers using something like the 123\PowerSwitch until the battery has recovered a bit.
I would probably need two of those, assuming i can put them in parallel (unlike mechanical relays).


Is there something wrong with using my original Plan B, but with 123\PowerSwitch (es) instead of relays? It would eliminate the diodes as constant power draw (@ 0,6 Volt drop they would eat up almost 50 kWh per year). Capacity is a bit of an unknown to me. In higher voltages i would be very careful, but at 24 Volts the worse that can happen (in my mind) is some accidental short results in some welding, and potentially a tiny bit of fire. But i do have fuses in everything, including main battery fuses, wire protecting fuses for the distribution layer and another layer of fuses for the actual loads themselves.
 

Offline rteodor

  • Regular Contributor
  • *
  • Posts: 122
  • Country: ro
Re: DC Failover for Solar Power
« Reply #3 on: July 23, 2023, 08:49:47 am »
Quote from: Lukas52
Would i even need control over the remaining battery capacity?
I can not tell you from experience that you need the SoC because I do this thing with diodes and lead-acid batteries. That works very well but LFP is different.

From what I know about LFP you would get 6k cycles when cycled at 50% DoD (depth of discharge). But it really depends on your specific goals, requirements and cells specifications.

My example: hybrid installation with 5KWh LFP. I am cycling it at ~50% because of two reasons: (1) to get reasonable cycles and (2) have some reserve for occasional mains power loss during night storms.

Quote from: Lukas52
If i understand this correctly the diodes would prevent reverse charging
Yes that is their purpose.

Quote from: Lukas52
To keep some reserve i would want to switch over at say 3 - 3,1 Volt. That would be 24 or 24,8 Volt, add too that the Diode Vfr of 600 mV -> Set PSU to ~25,2 Volt.
Two remarks here:
  • 1. Schottky diodes have lower Vfr depending on current. What diodes I use, have 0.1...0.6V. That makes for a smooth transition of power consumption when both sources are available.
    So the diodes works also as equalizers, they will have variable Vfr depending on the current consumption on each branch. Unless the loads have a problem with that Vfr drop, I would not factor in the Vfr but consider the voltages before the diodes. It keeps things simple.
  • 2. 3-3.1 V could mean 5% or it could mean 20%. You could be fine with that large variation or go back to the problem of getting/measuring SoC.

Quote from: Lukas52
As soon as my battery voltage goes bellow this limit the diode should prevent current to flow from the battery shouldn't it? Since the rail Voltage would be higher thanks to the power supply than the battery voltage.
As mentioned above, it will be a smooth transition: current from battery branch decreases and current from PSU branch increases.
That is still the case but in oposite direction when the Sun comes from behind the clouds.

Quote from: Lukas52
Is there something wrong with using my original Plan B, but with 123\PowerSwitch (es) instead of relays?
Technically both of your plans are correct but ...
... with Plan A you can not maximize the use of solar
... with Plan B (assuming the use of 123\PowerSwitch-es) the cost is quite high. (BTW: if my memory is right they mention on their site that you could paralel them). 50[KWh/year] that you have calculated * 0.4[EUR/KWh] = 20 EUR per year. One 123\PowerSwitch is ~120EUR+VAT.

With diodes you can turn off the PSU on the AC side (with a relay or better, a SSR) and avoid the 123\PowerSwitch altogether.
I do not know how well it would work to command the SSR from MPPT charger because the charger does not know the SoC. You might get too much on-off cycling on the PSU ....
The problems I see in your setup are this:
  • (1) getting a proper 24V tunable voltage PSU,
  • (2) choosing proper diodes and
  • (3) more complex: getting the SoC and use it to optimize cycling of the battery and reduce idle/redundant power consumption from the PSU.

Many things to consider but I hope it helps.
 
The following users thanked this post: Lukas52

Offline rteodor

  • Regular Contributor
  • *
  • Posts: 122
  • Country: ro
Re: DC Failover for Solar Power
« Reply #4 on: July 23, 2023, 09:14:04 am »
Something to clarify: I mentioned the power switches because I did not want to hang to the diodes idea not knowing your project specifics. Also I wanted to point out to whoever might read this thread, that mechanical switches are better for the abnormal use cases (protection) while the solid-state power switches are better used for normal use case switching.
 

Offline Lukas52Topic starter

  • Contributor
  • Posts: 21
  • Country: de
Re: DC Failover for Solar Power
« Reply #5 on: July 23, 2023, 09:43:13 am »
The Cells I'm using are EVE-LF280K-280Ah, most of my values comes straight from the datasheet. I'm just not sure how accurately the Victron chargers can measure voltages. The Datasheet claims to know the state of charge rather precisely, but I'm not sure if that is with or without the Smart Battery Sense device (that's a sensor that goes directly between the battery terminal to measure the voltage and temperature there and sends it back to the charger).

My loads can (in theory at least) go all the way down to 18 Volt, even if at reduced efficiency, so they probably don't mind that extra voltage drop.

The 0,6 Volt drop was for a diode i found on mouser that can do 400 A with a 30 V rating, that is in fact overkill, since my loads, even in a worst case, will only draw 200 A only for a couple of seconds, 100-150 A is the "usual" load.

 A more detailed search seems to suggest that just putting more diodes in parallel might be the smarter choice, like the VS-42CTQ030S-M3 for example. 30 V / 40 A with a Vf of 380 mV and much cheaper than one big diode.

My PSUs (they are two running in parallel, but they have a sync connector to sync the output voltage) can be adjusted between 22 ~ 28 Volt


So bottom line, i will probably just get the Smart Sense thing or even a smart shunt so i have a proper battery voltage to go of of and use diodes, eliminating DC side relays.
All i need know is to find the best diode for this application :)
 

Offline rteodor

  • Regular Contributor
  • *
  • Posts: 122
  • Country: ro
Re: DC Failover for Solar Power
« Reply #6 on: July 23, 2023, 03:20:38 pm »
Quote from: Lukas52
The Cells I'm using are EVE-LF280K-280Ah, most of my values comes straight from the datasheet.

I have a pack of the very same EVE cells waiting :). My first attempt to put them to work failed because it seems I got too much resistance in the inter-cells connections (1.5 ... 3.5 mOhm). So I need some advice of my own :D.
The smaller system, where I used the diodes only works at 3...5A where I put diodes rated for 15A just because of availability.

Quote from: Lukas52
I'm just not sure how accurately the Victron chargers can measure voltages. The Datasheet claims to know the state of charge rather precisely, but I'm not sure if that is with or without the Smart Battery Sense device
It is not (only) the voltages, its the current. I will have to go back to the theory a little bit here.
This is the problem with LFP: it is not possible to determine the SoC in range 10%...90% from voltage alone. You could have two cells or batteries with the exact same voltages an one can be at 30% and the other at 60%. Voltage stays almost flat for most of the SoC range. It is at the ends of the SoC range where voltage varies fast enough to be useful.
That is how chargers work: in bulk charging they will dump as much current as possible while watching that voltage is lower than some limit: 3.45...3.55 (default for Victron) ... 3.65. When this limit is hit then they will go into constant voltage for some time or until the battery current drops under some limit. Then they will keep a float voltage. When float state is reached a charger _estimates_ that the SoC is 100%.

If the charger delivers current not only to the battery but to the loads too without knowing how much goes in each direction determining SoC from voltage is Mission Impossible. But by using the voltage limits and timeout the charger will know when the battery is near full or full and prevent the battery getting more energy that it could hold. So the charger does not know the SoC but it can figure out some thresholds of the SoC from the voltage.

Now, if the BMS measures the battery voltage and current (or whatever device does that) it is in the great position to be able to reliably compute a SoC value. This is what the SmartShunt does. The Smart Battery Sense is quite frankly next to useless in your case. You said you have a BMS. Does it have a shunt or a Hall sensor for measuring current too ? And if it does then how can you access/use it ? I mean do you have an interface like serial or RS485 or CAN ? Better, maybe your BMS has some programmable outputs that you can set from some software to be switched at some specified SoC level.

If your BMS does not have any of this capability I'd warmly suggest to consider the SmartShunt. You would need maybe a RaspberryPi with VenusOS installed to automate it. If you have the resources and time and desire to tinker then you could use an Arduino or the like instead of the RaspberryPi.

Quote from: Lukas52
The 0,6 Volt drop was for a diode i found on mouser that can do 400 A with a 30 V rating, that is in fact overkill, since my loads, even in a worst case, will only draw 200 A only for a couple of seconds, 100-150 A is the "usual" load.
For 200A for a few seconds I would look for at least 250A. But that's just me, maybe more experienced forum colleagues can give better advice.
Maybe move the thread to "Power/Renewable Energy/EV's" (or start another thread) so this discussion can get more attention.

Quote from: Lukas52
A more detailed search seems to suggest that just putting more diodes in parallel might be the smarter choice, like the VS-42CTQ030S-M3 for example. 30 V / 40 A with a Vf of 380 mV and much cheaper than one big diode.
I tend to think the same too. Still, at 200A balancing the diodes might become relevant. Again, more experienced people can give better advice on this than me.

Good to know that you have the PSU's already.
 
The following users thanked this post: Lukas52

Offline Lukas52Topic starter

  • Contributor
  • Posts: 21
  • Country: de
Re: DC Failover for Solar Power
« Reply #7 on: July 23, 2023, 05:39:12 pm »
Ah ok, now i know what you mean with the SoC problem :)

Looking at the discharge graphs i assumed that the Charger just measures the voltage very precisely, but those graphs where made with a constant load and still 80% happens withing 0,05 Volt or so.

Given that i am running two 8S Battery packs in parallel a 123 Smart BMS would be 900 Euro. The Daly BMS the seller of the cells recommends is only 400 Euro. It has both UART and RS485, however the documentation is horrible and apart from mentioning it in the pin out diagram never talks about it again... So getting a smart shunt from victron seems to be less painful :)

I probably also open a new thread in the mention subforum, just to check on the diode thing, or maybe someone knows some better value BMS or something.

Thank you for your help, your information have been (and will be) very helpful!
And you have already saved me from spending a fortune on relays :)
 

Offline rteodor

  • Regular Contributor
  • *
  • Posts: 122
  • Country: ro
Re: DC Failover for Solar Power
« Reply #8 on: July 23, 2023, 07:50:11 pm »
Quote from: Lukas52
Looking at the discharge graphs i assumed that the Charger just measures the voltage very precisely, but those graphs where made with a constant load and still 80% happens withing 0,05 Volt or so.
Temperature, age, usage and other factors will vary the voltage more than 50mV. That is why it does not work.

Quote from: Lukas52
or maybe someone knows some better value BMS or something
I imagine people are in vacations nowdays, otherwise to such a question you should get tons of replies.

I'll give one answer too (even if my practical BMS experience is next to inexistent): Basically there are 2 cost categories: low cost Chinese and high cost ones made elsewhere.
In the low cost category everybody praises JK while Daly is controversial. Most people hate it, others says it just needs to be set properly.
But the JK has the advantage of an internal active balancer. JK also has a stream of supporters that even took the task of translating and improving the documentation.

123Smart and REC are in the high cost category. I've chosen REC because I need to be able to work remote on it and because people say it has plenty of options.

There are many criterias to consider in a BMS. Here are a few that come to my mind:
  • power switch (relay, solid state MOSFET's; mechanical relay is better for protection)
  • balancer (pasive, active)
  • SoC monitoring (yes, some do not have it at all - choose the right one and you do not need a SmartShunt)
  • precharge resistor circuit
  • charge current limiter
  • interfaces (RS485, simple serial, CAN, etc)
  • programable outputs (relay, optocouplers)
  • software (standalone, Bluetooth or WiFi app)
  • options (because some come with locked options)
  • documentation

This being said I got much of my information from Andy's "Off Grid Garage" youtube channel.
« Last Edit: July 24, 2023, 04:33:36 am by rteodor »
 
The following users thanked this post: Lukas52

Online coppercone2

  • Super Contributor
  • ***
  • Posts: 9420
  • Country: us
  • $
Re: DC Failover for Solar Power
« Reply #9 on: September 03, 2023, 05:40:25 am »
you should put this thread in the power / green section because there is ALOT of people that read that which have/do solar panels
 

Offline rteodor

  • Regular Contributor
  • *
  • Posts: 122
  • Country: ro
Re: DC Failover for Solar Power
« Reply #10 on: September 03, 2023, 01:30:08 pm »
« Last Edit: September 03, 2023, 01:33:50 pm by rteodor »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf