diy solar

diy solar

DC coupling is not bad, I guess.

1201

Solar Wizard
Joined
Feb 1, 2023
Messages
2,829
Location
Texas
I love AC coupling with a passion: it's super cheap for the amount of pv you can connect, you take the load off your main inverter during the day, it's highly efficient at running loads, and it gets MORE efficient at high output- when my pv inverter is absolutely running 100% flat out the efficiency is 98%.

A battery inverter running flat out would be in the 80s for efficiency probably.

But, charging a battery from ac coupled pv is not very efficient. You have the losses from the pv inverter, plus losses in the battery inverter converting ac to DC again- I have seen as low as 84% !

So, maybe the answer is having some pv on a DC coupled charge controller? I did that recently and it's pretty nice I must say! I can have the ac coupled pv only run loads or export to the grid at high efficiency, and at the same time charge the battery at 98+ percent efficiency from the DC coupled charge controller.

Btw, victron os is the shiznit!
 
Last edited:
In my case with a SolArk, some DC coupled solar helps stabilize my performance when the grid is down. I have 9kW of AC coupled solar and unless my batteries are at 60% SOC my loads will not be enough to keep the AC coupled solar going. I do have 3kWh of DC coupled solar as well.
 
maybe the answer is having some pv on a DC coupled charge controller?
I have not tried this, but with your Growatt inverter I believe you could simultaneously wire the PV you have currently connected to the Growatt to the charge controller and set the Growatt MPPT inputs to “dc source”, that way the Growatt and charge controller would share the PV power until battery full then the Growatt would get it all.
I have set the mppt to “dc source” on mine and the power loss is small, barely noticeable- not even 5%, of course there is no documentation but “dc source” seems to turn off the mppt tracking, so that being the case the inverter and charge controller should not fight if connected to the same PV strings.
If you try it I’d like to hear how it goes. I’d suggest that first time make sure you have a breaker or disconnect ready to open in case they do fight.
 
In my case with a SolArk, some DC coupled solar helps stabilize my performance when the grid is down. I have 9kW of AC coupled solar and unless my batteries are at 60% SOC my loads will not be enough to keep the AC coupled solar going. I do have 3kWh of DC coupled solar as well.
It's this related to a setting? Do you have ac coupling set for example between 60% and 90%?
 
I have not tried this, but with your Growatt inverter I believe you could simultaneously wire the PV you have currently connected to the Growatt to the charge controller and set the Growatt MPPT inputs to “dc source”, that way the Growatt and charge controller would share the PV power until battery full then the Growatt would get it all.
I have set the mppt to “dc source” on mine and the power loss is small, barely noticeable- not even 5%, of course there is no documentation but “dc source” seems to turn off the mppt tracking, so that being the case the inverter and charge controller should not fight if connected to the same PV strings.
If you try it I’d like to hear how it goes. I’d suggest that first time make sure you have a breaker or disconnect ready to open in case they do fight.

My voltage is too high for my charger controllers
 
Last edited:
Since i have only used solar 100% off grid i have remained a bit weak on AC coupling.

But it seems to me like what you are saying is it is less efficient to do A than B.
A. Convert PV DC into AC and power an AC charger to convert it back to DC for the batteries.
B. Convert PV DC into lower voltage DC and feed it to batteries.

And if im understanding correctly option A probably has something to do with microinverters which don't directly interface to a battery.

In my system i have all in one inverter/charger combos, and in these units the MPPT alters the PV DC to the correct voltage to dump out into a DC 'bus' where a buck/boost converter will operate in buck mode to drop the voltage down to battery charging voltage. So in my system option B is already happening, if im understanding you correctly.

I wouldn't have option A unless my MPPT only fed an inverter and then the inverter fed a separate ac-powered battery charger.
 
Most DIY-friendly batteries are limited to 48V DC. Converting either grid AC or PV DC to 48V DC is necessarily going to have losses. AC to DC charging is ~90-95% efficient with the hope that future SiC/GaN tech brings that closer to 95%+.

Generally the focus has always been on inverting efficiency (PV DC or 48V battery DC). The reason being if you're in a position to store excess energy in a battery, that the solar power is likely "cheap" to you. Yes you lose some on the way but the battery holds the same amount of energy at 100% regardless of how much was lost during the charging process.
 
Since i have only used solar 100% off grid i have remained a bit weak on AC coupling.

But it seems to me like what you are saying is it is less efficient to do A than B.
A. Convert PV DC into AC and power an AC charger to convert it back to DC for the batteries.
B. Convert PV DC into lower voltage DC and feed it to batteries.

And if im understanding correctly option A probably has something to do with microinverters which don't directly interface to a battery.

Correct. Option A would be using a grid tie inverter's ac output to charge a battery.
In my system i have all in one inverter/charger combos, and in these units the MPPT alters the PV DC to the correct voltage to dump out into a DC 'bus' where a buck/boost converter will operate in buck mode to drop the voltage down to battery charging voltage. So in my system option B is already happening, if im understanding you correctly.

That is correct
 
Enphase (pure AC) is 98% efficient. If you have 50 feet of 10awg wire at 25 amps, that is 2.19% voltage drop per 120v line. Net efficiency is 96%.

Sol-Ark inverting DC->AC direct from panels (not via batteries) is 97% efficient. But, the voltage on the wires can be 400v at 25 amps for the same 50 feet is 0.66% voltage drop, for a net efficiency of about 96.5%. Sol-Ark has a high voltage bus and a low voltage bus (48v). The 97% efficient is because it doesn't have to convert to the low-voltage bus.

Factor in batteries, and there is no comparison. Much more efficient to go from high voltage DC to low voltage DC, than it is to invert the AC back to DC to store in the batteries.
 
Most DIY-friendly batteries are limited to 48V DC. Converting either grid AC or PV DC to 48V DC is necessarily going to have losses. AC to DC charging is ~90-95% efficient with the hope that future SiC/GaN tech brings that closer to 95%+.

Generally the focus has always been on inverting efficiency (PV DC or 48V battery DC).

Depends. If most of your loads are daytime loads then efficiency of DC to AC might be more important
 
Depends. If most of your loads are daytime loads then efficiency of DC to AC might be more important
Yes that is what I mean by "inverting efficiency". AC to DC is technically called "rectification" although most people are very loose with their terminology and say "inverting" for both.
 
Old time power supplies and modern cheap ones use diodes for rectification.
When it comes to battery charging, you need current regulation, so rectify, filter, DC/DC converter.

Our battery inverters or hybrids have an inverter for battery to AC which operate bidirectionally, AC to DC, using same sine-wave inverter.
Called "Power Factor Correction" if just a DC power supply, but might as well call it inverting when same circuit is bidirectional. It is a sine wave AC <--> DC switch mode power supply
 
It's this related to a setting? Do you have ac coupling set for example between 60% and 90%?
I have not tweaked my settings. It is not really an issue for me because I have several workarounds.

I do have about 3kW of DC coupled solar. My loads are normally pretty small and I have never had the grid go down when the batteries were less than 90%, so the DC coupled solar covers my loads. I did add a 14-30 receptacle to my essential loads panel and when I tested the system and charged my EVc at 4.8 kW the AC coupling micros fired up, so that is another one of my workarounds. I also installed a new pack and am drawing it down to about 60-70% every night so I should test it now since my new pack charges at 6-7 kW. That is probably enough to fire up all of my AC coupled solar at least for a while based on my observations from Solar Assistant.
 
Last edited:
Enphase (pure AC) is 98% efficient. If you have 50 feet of 10awg wire at 25 amps, that is 2.19% voltage drop per 120v line. Net efficiency is 96%.

Sol-Ark inverting DC->AC direct from panels (not via batteries) is 97% efficient. But, the voltage on the wires can be 400v at 25 amps for the same 50 feet is 0.66% voltage drop, for a net efficiency of about 96.5%. Sol-Ark has a high voltage bus and a low voltage bus (48v). The 97% efficient is because it doesn't have to convert to the low-voltage bus.

Factor in batteries, and there is no comparison. Much more efficient to go from high voltage DC to low voltage DC, than it is to invert the AC back to DC to store in the batteries.
I would like to know the efficiency of the solark when it's selling 12kw from pv to grid. I doubt it's 97% efficient but it may be.

My grid tie inverter selling max power to grid today didn't go below 98% efficiency. Same 98% efficiency for daytime loads.
 
I have not tweaked my settings. It is not really an issue for me because I have several workarounds.

I do have about 3kW of DC coupled solar. My loads are normally pretty small and I have never had the grid go down when the batteries were less than 90%, so the DC coupled solar covers my loads. I did add a 14-30 receptacle to my essential loads panel and when I tested the system and charged my EVc at 4.8 kW the AC coupling micros fired up, so that is another one of my workarounds. I also installed a new pack and am drawing it down to about 60-70% every night so I should test it now since my new pack charges at 6-7 kW. That is probably enough to fire up all of my AC coupled solar at least for a while based on my observations from Solar Assistant.
It's nice to have the workarounds for sure but all nice if you can figure out why, just in case.
 
but all nice if you can figure out why, just in case
I agree and definitely know why my small loads were not enough to enable the algorithm in my SolArk to fire up 8 kW of micros. I was in a rush at the time and the SolArk was not in my planning horizon because I already had an Outback Skybox which I thought would fulfill my needs. The Skybox did have a better AC coupling algorithm but had many other limitations which drove me to migrate to the SolArk several years later.
In retrospect I would have started with a DC coupled array and added AC coupled additions incrementally after that. Adding micros in small increments is far easier than with MPPT constraints of voltage and orientation.
 
I agree and definitely know why my small loads were not enough to enable the algorithm in my SolArk to fire up 8 kW of micros.
Maybe I'm misunderstanding, but if the soc is less than 60% even with low loads the inverter should push all available ac coupled power into the battery
Adding micros in small increments is far easier than with MPPT constraints of voltage and orientation.
Can you expound on this a little. I'm struggling to understand which constraints you are referring to
 
If you've only got room for 3 or 4 panels on one roof slope, you can install any number of microinverters.
String inverters, need to match voltage of other parallel strings or at least MPPT range of a spare input.
 
Maybe I'm misunderstanding, but if the soc is less than 60% even with low loads the inverter should push all available ac coupled power into the battery
That is the way my battery usage is now configured and I have not tested that but expect it should fire up all the micros. The grid down scenarios have not been my highest priority.
Can you expound on this a little. I'm struggling to understand which constraints you are referring to
I only have two MPPT controllers and want to add small increments of panels at various azimuths and tilt around my home. Not to mention that DC has to be in metal conduit versus with micros Romex will suffice. With micros I can add four panels at one place and two at another and so on with different tilts and azimuths. With MPPT controllers I have to worry about minimum voltage and consistant panel size as just some of the the constraints. The micro constraints are there too but I can incrementally add one micro at a time and use Romex.
As I mentioned, a more planful approach would have been to start with DC coupled arrays in the beginning where I had one big pitch that could easily accomodate one or both MPPT controllers. Then I could add micros later incrementally with less complex wiring and conduit.
I agree with the premise of your first post.
 
Last edited:
That is the way my battery usage is now configured and I have not tested that but expect it should fire up all the micros. The grid down scenarios have not been my highest priority.
So in regular operation, connected to grid your micros are fully operational though right?
 
So in regular operation, connected to grid your micros are fully operational though right?
Yes because they just pass through the SolArk and serve the loads and battery charging and export any excess to the grid. Actually their output gets mixed with the DC coupled solar that gets inverted and some mix of both gets exported.
 
If you've only got room for 3 or 4 panels on one roof slope, you can install any number of microinverters.
String inverters, need to match voltage of other parallel strings or at least MPPT range of a spare input.
Yes, that is exactly how a mix of DC coupled and AC coupled solar can be implemented optimally in my situation.
 
Yes because they just pass through the SolArk and serve the loads and battery charging and export any excess to the grid. Actually their output gets mixed with the DC coupled solar that gets inverted and some mix of both gets exported.
Nice. What are some things you liked better about the radian?
 
Nice. What are some things you liked better about the radian?
Actually it was a Skybox, but did have a Radian at another home and the Skybox offered everything All in One and much better TOU scheduling than the Radian. In the end I could not get Outback to accomodate closeed loop communication with my BMS and that was important to me. I did accomplish that with the SolArk which also improved on the TOU scheduling. I recently migrated from my DIY pack with an Orion JR2 BMS to Pytes V5 batteries and as I mentioned I am now using my pack more actively than with the DIY pack. They are almost the same size and I don't have a rational explanation.
 
Yep.

If most of your usage is PV to grid, AC coupling makes a lot of sense in terms of efficiency.

If most of your usage is solar to battery then battery to load, DC coupling makes a bit more sense.

I would also add that the primary driver of efficiency in a standard nonisolated converter is similarity in voltage, input to output. A 500 volt to 48 volt converter isn't going to be very efficient, but a 500 volt to 400 volt converter is a lot more efficient. Same thing with conversion to 240 volts AC. So high voltage systems will help with DC coupling efficiency.
 
Actually it was a Skybox, but did have a Radian at another home and the Skybox offered everything All in One and much better TOU scheduling than the Radian. In the end I could not get Outback to accomodate closeed loop communication with my BMS and that was important to me. I did accomplish that with the SolArk which also improved on the TOU scheduling. I recently migrated from my DIY pack with an Orion JR2 BMS to Pytes V5 batteries and as I mentioned I am now using my pack more actively than with the DIY pack. They are almost the same size and I don't have a rational explanation.
Copy that. Sounds like the solark is an improvement in almost every way. I think a lot of these boxes are really good, but hampered on the software side
 
If most of your usage is PV to grid, AC coupling makes a lot of sense in terms of efficiency.

If most of your usage is solar to battery then battery to load, DC coupling makes a bit more sense.
You're missing the factor of price/TOU. It's increasingly worth it to take the conversion losses for battery storage because the $/kWh value of that energy is so much higher in the evenings for typical TOU structures.

You can thank non-DIY solar for that!
 
I would like to know the efficiency of the solark when it's selling 12kw from pv to grid. I doubt it's 97% efficient but it may be.
97.5% per the spec sheet.

A string inverter will be as good as a microinverter. They both do the same job, PV to AC, so why wouldn't they be similar in efficiency?

My grid tie inverter selling max power to grid today didn't go below 98% efficiency. Same 98% efficiency for daytime loads.
Are you accounting for the clipping since microinverters are often less power than the panels they are attached to.

For PV to AC, microinverters and string inverters are going to be basically the same. String inverters will less often be clipping since they are less often installed derated.

For PV to battery to AC, string inverters will be more efficient. They convert to AC only once, not twice, and AC conversion has loss.

Mike C.
 
97.5% per the spec sheet.

A string inverter will be as good as a microinverter. They both do the same job, PV to AC, so why wouldn't they be similar in efficiency?

The solark has charge controllers so they charge the battery and the solark then has to invert from battery to AC to run loads or power the grid.


Looking at the images In this thread, the efficiency is around 93-94% at high output
Post in thread 'Sol-Ark 15k Current Capabilities' https://diysolarforum.com/threads/sol-ark-15k-current-capabilities.78308/post-1002422
 
It's increasingly worth it to take the conversion losses for battery storage because the $/kWh value of that energy is so much higher in the evenings for typical TOU structures.
Yes that is where my biggest economic return comes from. This time of year I get credit for export at those peak rates and the SolArk helps me avoid importing any kWhs at those rates.
 
You're missing the factor of price/TOU. It's increasingly worth it to take the conversion losses for battery storage because the $/kWh value of that energy is so much higher in the evenings for typical TOU structures.

You can thank non-DIY solar for that!
maybe the answer is to have both ac and dc coupled pv :p
 
The solark has charge controllers so they charge the battery and the solark then has to invert from battery to AC to run loads or power the grid.
That is not how it works when PV is available.

The MPPTs make the panel voltages into the desired voltage for the AC inverter (usually about 350 volts DC which is optimum for making 240 VAC and why making strings that produce about 360 volts DC is best for MPPT efficiency). There is no going through the battery for PV to AC.

The battery charger works from the MPPT output voltage. The battery discharge runs through a booster to make the 360 volts DC for the inverter. The battery path is not involved in the PV to AC direct path. In this way, the string inverter is really just a big microinverter, same energy conversion process, but with a battery charger and booster to the side.

This is one reason why inverters with high voltage batteries are more efficient, the battery buck and boost conversions are more efficient than doing it at 48 volts, and cheaper since current is what drives cost for the most part.

This design is also why an inverter can have 12 KW AC output (the size of the AC inverter) but take in 18 KW of PV (like the EG4 18KPV). 6 KW is being used by the battery charger separately from the AC power path.

Basically, the AIO hybrid is NOT a battery charger followed by a battery inverter. There is a direct path of PV to AC.

When there is no sun, then the system has to charge the battery during the day and then use the battery during the night. But this is done with more efficient DC conversions than doing AC twice like a microinverter.

Mike C.
 
That is not how it works when PV is available.

The MPPTs make the panel voltages into the desired voltage for the AC inverter (usually about 350 volts DC which is optimum for making 240 VAC and why making strings that produce about 360 volts DC is best for MPPT efficiency). There is no going through the battery for PV to AC.

The battery charger works from the MPPT output voltage. The battery discharge runs through a booster to make the 360 volts DC for the inverter. The battery path is not involved in the PV to AC direct path. In this way, the string inverter is really just a big microinverter, same energy conversion process, but with a battery charger and booster to the side.

This is one reason why inverters with high voltage batteries are more efficient, the battery buck and boost conversions are more efficient than doing it at 48 volts, and cheaper since current is what drives cost for the most part.

This design is also why an inverter can have 12 KW AC output (the size of the AC inverter) but take in 18 KW of PV (like the EG4 18KPV). 6 KW is being used by the battery charger separately from the AC power path.

Basically, the AIO hybrid is NOT a battery charger followed by a battery inverter. There is a direct path of PV to AC.

When there is no sun, then the system has to charge the battery during the day and then use the battery during the night. But this is done with more efficient DC conversions than doing AC twice like a microinverter.

Mike C.
Fair enough. However my point stands that a battery inverter is not going to be as efficient as a pv inverter at high output.

Also, while power may be coming directly from the mppts without hitting the battery first,my belief is it's still being converted to 48v DC for the inverter section to do it's thing, unless you are telling me that the inverter can convert both 360v DC and 48v DC to AC, which I doubt.
 
Last edited:
In an all in one the inverter is feeding from high voltage DC bus. It's impossible for the inverter to work off 48v without stepping it up first. But, in situations where the solar power can cover the entire load, no or almost no 48v power is being stepped up and fed to the inverter.
 
In an all in one the inverter is feeding from high voltage DC bus. It's impossible for the inverter to work off 48v without stepping it up first. But, in situations where the solar power can cover the entire load, no or almost no 48v power is being stepped up and fed to the inverter.
ok i think i understand. so the inverter is always converting from high voltage dc to ac. either directly from mppt, or by stepping up 48v to the high voltage and then inverting that to AC?
 
Yes, exactly. The MPPT has its own buck/boost converter function to always spit out power onto the DC bus at correct voltage. Then there is a buck/boost converter which can either a: step voltage from AC input down to charge batteries, or b: step voltage up from batteries onto dc bus to feed inverter.

Thus, most all in ones cannot power loads from AC input and battery power simultaneously because each process requires use of the buck/boost converter and there is only one to share, so it can power loads from either AC input or battery power but not both simultaneously.
 
Last edited:
You're missing the factor of price/TOU. It's increasingly worth it to take the conversion losses for battery storage because the $/kWh value of that energy is so much higher in the evenings for typical TOU structures.
Right, I am just speaking for efficiency. Things like load shifting, zero-export times or even a good sale on microinverters will change the economics.
 
Yes, exactly. The MPPT has its own buck/boost converter function to always spit out power onto the DC bus at correct voltage. Then there is a buck/boost converter which can either a: step voltage from AC input down to charge batteries, or b: step voltage up from batteries onto dc bus to feed inverter.

Thus, most all in ones cannot power loads from AC input and battery power simultaneously because each process requires use of the buck/boost converter and there is only one to share, so it can power loads from either AC input or battery power but not both simultaneously.
Ok I get this now. One more question- what path does pv DC take to battery? So it first goes from mppt to hv DC bus, then is there a converter that steps that down to 48v DC?
 
Back
Top