GoodWe Hybrid 14KW Solar / Tesla Model 3,Y Battery Module 225KWH ESS

I joined the forum a couple of months ago, but I've been doing my own solar and ESS builds for years. I've decided to post an overview of my system I installed at my lake house a few years ago. I'm also currently most of the way through a complete rip and replace at my main house, but I'll potentially cover that in another post later. It's similar to this install but has some nice updates and newer equipment.

First thing I should cover is my goals for the system. They are pretty simple really.

• Power backup during an outage - the lake house is in a rural area with a lot of above ground power lines. We lose power at least 2 times a month on average and I didn't want to invest in a backup generator
• I wanted the payback on the system to be 5 years or less - partly just because I wanted to see how economical I could build a system
• My wife and I are big into being environmentally conscience. We also drive a lot of miles because we go back and forth between the main and lake house at least once a week, and we have kids and grandchildren that we drive to visit often. We have 2 Tesla EVs, and we average about 35k miles a year on them. And I'm also almost done with a EV Ski Boat, but that's a project for another post. Anyway, I wanted to cover all the EV charging with Solar and ESS systems at the two houses.
• Keep the system as standard as possible. By occupation I'm a design engineer working in the industrial products market, but for this build I wanted to keep things as standard as possible such that if I were to sell the house, it could be maintained by anyone familiar with the standard solar installs. That means keep the custom-built electronics to a minimum and only where needed to save very significant costs.

Here's a diagram showing the topology



In terms of equipment selection, the first set of panels I bought were 20 Heliene 360W panels, then later I bought a pallet of 38 Canadian Solar 400W panels on a clearance sale. I used Ironridge mounting hardware. Here's a couple of pictures of me mounting the panels on the roof.







I started out planning to use SMA inverters. I had an earlier system based on two Sunny Island 6048 inverters along with Victron MPPT charge controllers and Tesla Model S battery modules. So, for this system I was planning to use a SMA Sunny Boy 7.7kw PV inverter and a Sunny Boy storage 6kw ESS inverter. I also decided early on that this system would be based on a high voltage battery system. My previous system was a 48V battery system and I wanted something more efficient and without the complication of heavy wiring, busbars, fuses etc.

But two things caused me to shift directions. First the order for my SB 7.7kw inverter was put on back order because SMA decided to suddenly phase them out and introduce their new line of all-in-one hybrid inverters. It was also turning out to be time consuming to reverse engineering the CANBUS battery interface on the Sunny Boy Storage inverter due to the proprietary nature of SMA.

For the inverter I came across GoodWe. GoodWe inverters aren't that well known in the US, but worldwide they are one of the biggest suppliers of inverters in the world. They are also on the approved equipment list in many states including California. And I checked with the local authorities where my houses are and there wasn't going to be any problem with getting them approved for grid tie. GoodWe inverters are also sold by GE in the US under their brand and they are now making a much stronger push into the US market. Net, net, I was convinced it was a good quality product that was well supported.

The other factor is GoodWe is not super proprietary about their stuff like a lot of other inverter manufactures. That made it much easier to figure out the battery interfaces.

I ended up deciding on the GoodWe GW9600A-ES all-in-one hybrid inverter. It has four 15 amp MPPTs that support 150% DC oversizing with up to 600V input. It supports grid-tie and off-grid operation as long as you install the associated autoformer. It has built-in energy monitor and tigo rapid shutdown support. It supports grid-zeroing if you want that and it has an economic mode where you can define time periods throughout the day and define exactly how you want it to behave during each time period. I also discovered just by use, that unlike some inverters that say they support DC over-provisioning the GoodWe will go way over 9.6 kw with my 14kw of panels. Since it's a DC coupled system between the panels and the batteries not all the power has to go through a DC to AC conversion. It's really common for me to see 13kw of power coming from the panels with some of that charging the batteries and the other part going to power the house or be fed into the grid.

It also has an acceptable mobile app and web based monitoring and reporting. I say acceptable because I've been able to get it do everything I need, but it could be better. But I will say it's gotten better and better over the time I've had it. Also because GoodWe isn't super proprietary I've seen all the interfaces needed to intergrate it into Home Assistant which is something I'll likely do in the future.



For the batteries for ESS I pretty quickly decided on using surplus Tesla Model 3/Y battery modules. I bought my Tesla Model Y new when they first came out. But for my 2nd Tesla I decided to buy 2 wrecked Model 3's and rebuild one good car out of them. Through that I was impressed with the quality of the Tesla batteries, their refined BMS system, and the long term durability.

But what really sealed the deal is the cost factor. There's simply no other way to get such high quality batteries for such a low price per KWH. I picked up my first set of 4 Tesla Model 3/Y modules for $1k because one of the modules had a bad cell group. Which rendered them useless for car, but perfect for fixed ESS storage. The cell group was bad because some of the fuse links in the module were broken. I just tig welded shunts across that cell group to make it a 95S pack instead of a 96S pack.

In terms of the amount of battery storage, I also knew I needed a bunch. Mainly because I was keenly following the debate going on between our power company and the utilities commission on a complete revamp of the Net Metering rate plans. And importantly the fact that the proposals were to only grandfather existing net metering customer into 1:1 credit for power sent to the grid until 2027. Also, in NC, anyone doing solar and net metering is going to be required to be on a TOU rate plan. Which I saw as another opportunity to reduce my payback period on the system by only buying any power from the grid at the lowest discount rate during the night. The discount rate is only $0.065 per KWH where the standard non-TOU rate here is about $0.12 per KWH. But the biggest driver of needing a lot of ESS capacity is I pretty much split my time 50/50 between the main and lake house along with a lot of KWH of EV charging. Basically, when I'm not at one of the houses, I bank all the extra power in the ESS system. Then when I arrive there is plenty of power to dump into the EV.

Here's a couple of pictures of the battery. This is showing just one dock box with modules, I have 2 now and changed the orientation of the modules to fit 6 in each box. It was also a requirement that the batteries not be inside the house for fire safety reason. But with a HV systems and the 10Ga wires required. The batteries don't need to be anywhere near the inverter.



In terms of panels, rapid shutdown, and ESS shutdown I don't think it's worth going into a lot of details because it's pretty rudimentary stuff, so I'll mostly just post some pictures. I do have a completely integrated shutdown system. One switch handles a 100% shudown of all aspects of the system. Main panel on right, and main backup panel on left.



Garage backup sub-panel



Dock backup sub-panel



Here's a graph of my last month's showing % of the power covered. This includes EV charging. The big dip at the beginning is when I had the system offline for a few days for some upgrades.

I should probably also mention I've made updates to the house to make it super energy efficient. I have 7 high efficiency pioneer minisplit systems installed so every bedroom has individual temperature control and only needs to be heated or cooled when in use. I have heat-pump based hot water heaters and heat-pump based clothes dryer.


So what did it cost me:

20 Heliene 360W panels $2,400
16 Canadian Solar 400W panels. $1,695
GW9600A-ES $3,700
GW Autoformer $315
12 Tesla Model 3/Y Modules $7,500
Mounting hardware and Misc $1,800
Tigo Optimizer RSD Modules $1,200

Total $18,610

Savings per year
Electricity $4,200 (includes EV charging)
Payback Period 4.4 years

For EV Fans
Yearly gasoline saving over
the 2 BMW that Tesla's replaced. $5,600

I would move that battery breaker into external metal box because they catch fire and install rubber boots on those exposed battery terminals. But everything else looks amazing (except those annoying red stickers, hate them).

I already got rid of the breakers and replaced them with fuses. The breaker turned out to be unreliable and were giving false trips.

Yeah, good idea on insulating those battery connection points.

The required warning decals are ridiculous and I seriously doubt they provide much real value because so few people would understand the implications of what the different required warnings really mean and what you should do as a result. But if you want to pass the inspections, you have to have them....

Pretty amazing you got 225 kWh for $33/kWh. This is the way to go for those that can do it. And seeing how you got it approved is encouraging. Installing battery away from house is a necessity in my opinion. Do you have BMS on those modules? I don't see any in your picture. Can you show remaining battery modules from your system? I only see 2 or 4 of them. I have 6x Tesla model S modules in 2s3p config giving me 27kWh 44V 630Ah. I plan to convert them to 6S1P 132V 210Ah in the future and install them outdoors with water cooling and inside fire suppression enclosure. Thinking to dip entire module in conformal coating and let water spray over each module in case fire is detected. This way false triggering will not ruin the module.

I remember watching Jack Rickard (rip) with EVTV on YouTube using Tesla batteries for solar backup and trying to go hvdc. He was looking at some huge inverters from China. This seems like a much better solution on the inverter. Awesome setup.

Pretty impressive. I scanned through and couldn't find this. Does the goodwe need battery comms to work?

AntronX said:

Pretty amazing you got 225 kWh for $33/kWh. This is the way to go for those that can do it. And seeing how you got it approved is encouraging. Installing battery away from house is a necessity in my opinion. Do you have BMS on those modules? I don't see any in your picture. Can you show remaining battery modules from your system? I only see 2 or 4 of them. I have 6x Tesla model S modules in 2s3p config giving me 27kWh 44V 630Ah. I plan to convert them to 6S1P 132V 210Ah in the future and install them outdoors with water cooling and inside fire suppression enclosure. Thinking to dip entire module in conformal coating and let water spray over each module in case fire is detected. This way false triggering will not ruin the module.

Click to expand...

The picture I had on hand are actually fairly old and part way through my install of the system. That dock box only has 4 modules, and the Tesla BMS slave board is missing from the top module because some of the interconnect wires to the module were damaged and I had that BMS slave on the workbench reverse engineering the ISO-SPI interface to it. I'm actually traveling for the next few weeks, but when I can I'll take some new pictures of the current setup where the modules are rotated 90 degree and in 2 layers of 3, such that I can get 6 modules in each dock box.

When I get some time, I should post lessons learned about buying and working with Model 3/Y modules. As well as all the different variations of modules available. Including the new CATL manufactured LFP modules. The don't have the same energy density as the NMC Li-Ion modules that Tesla makes in the US, but they have the somewhat better safety of LFP vs Li-Ion, and they are more robust with less fragile parts that the Li-Ion modules have. And for fixed ESS usage, energy density isn't really all that important like it is in an EV. The LFP modules are used now in the lower range Tesla Module Y's and 3's. That was originally only the case for Tesla made and sold in China, but now it's the case for the standard range cars sold worldwide.

It's an interesting idea of coating the modules and using a water system in case of fire. But I wonder if it's really needed. Especially if you have the modules outside your house which I agree is really the best idea. Especially when you consider the robust safety measures built into the construction of all Tesla modules.

There are really two fundamental scenarios that lead to Li-Ion thermal runaway and Tesla has them both covered quite well. The most common cause is metal particles precipitating out of the electrolyte paste used to make Li-ion cells. That happens if the process controls on the production of the electrolyte paste isn't controlled very precisely. Tesla and Panasonic really have that process down to a science. This is exactly what happened to GM/LG that cause them having to recall and replace every single Chevy Bolt battery pack. It was also the root cause of just about all the laptop fires on airplanes a bunch of years back.

But to be super safe, that's why every single cell in a Tesla module is connected to the power plates with individual fusible laser welded links. The typical failure scenario is that cell does develop a short due to a metal particle in the paste between the anode and cathode layers. That cell essentially shorts out all the other parallel cells in the brick. Then the whole brick of cells gets really hot and goes into thermal run-away. That's enough concentrated thermal energy to then cause the neighboring cell bricks to overheat and go into thermal runaway. By having every single cell connected via a fusible link, if a cell shorts the link blows and a single cell doesn't have enough thermal energy to heat all the adjacent cells to a high enough temperature to go into thermal runaway.

The second cause of failure is where a brick of cells is allowed to get out of acceptable voltage or current range. That's where the very refined Tesla BMS system comes into play. Also, because the cells are assembled into groups of cells as a brick. That tend to average out slight cell to cell variances. Tesla is extremely regimented about a pack being built out of cells, and bricks that were all made in the same production batch as they come off the cell production lines. So, all the bricks in a Tesla pack are extremely consistent relative to each other. Outside of physical damage to a brick of cells, it's really rare to see a Tesla pack to have hardly any variance in brick-to-brick voltage even if the BMS system isn't functioning. The BMS is kinda the last resort, but if you build a pack out of extremely consistent cells, it'll never tend to get out of balance in the first place.

1201 said:

Pretty impressive. I scanned through and couldn't find this. Does the goodwe need battery comms to work?

Click to expand...

Yes, the GoodWe inverter requires a data connection to the batteries BMS system. I've never seen a high voltage inverter that doesn't require comms with the BMS system. From an engineering point of view, there are two reasons why that's typically the case.

First is because by nature a high voltage battery has to have quite few cells or cell groups in series. Therefore, it's pretty important to make sure the battery pack is staying in balance for safety reasons. And in high voltage and power battery systems, it really not the best solution from a cost point of view to have the BMS control the current flow to/from the battery since the inverter typically has to have relays or FETs to control the current flow. So, the BMS on an HV battery does all the monitoring and balance functions, but it tells the inverter when to start and stop the flow of current for error scenarios.

Second reason is to open up the flexibility of the design and support of different battery technologies in the HV battery system. By delegating the control of the battery safety logic to the BMS built into the battery system. The inverter no longer has to have much knowledge of how the battery works or what kind of battery chemistry is used.

With all that said, the GoodWe has the widest battery support I've seen of any HV hybrid inverter. It supports both CANBUS and RS485/Modbus battery interfaces. And it has choices in the setup for almost a dozen different battery makers. Also for a fair number of the protocols it supports, there is enough info available online to build your own interfaces to the inverter.

jpwhit said:

Yes, the GoodWe inverter requires a data connection to the batteries BMS system. I've never seen a high voltage inverter that doesn't require comms with the BMS system. From an engineering point of view, there are two reasons why that's typically the case.

First is because by nature a high voltage battery has to have quite few cells or cell groups in series. Therefore, it's pretty important to make sure the battery pack is staying in balance for safety reasons. And in high voltage and power battery systems, it really not the best solution from a cost point of view to have the BMS control the current flow to/from the battery since the inverter typically has to have relays or FETs to control the current flow. So, the BMS on an HV battery does all the monitoring and balance functions, but it tells the inverter when to start and stop the flow of current for error scenarios.

Second reason is to open up the flexibility of the design and support of different battery technologies in the HV battery system. By delegating the control of the battery safety logic to the BMS built into the battery system. The inverter no longer has to have much knowledge of how the battery works or what kind of battery chemistry is used.

With all that said, the GoodWe has the widest battery support I've seen of any HV hybrid inverter. It supports both CANBUS and RS485/Modbus battery interfaces. And it has choices in the setup for almost a dozen different battery makers. Also for a fair number of the protocols it supports, there is enough info available online to build your own interfaces to the inverter.

Click to expand...

That's the big problem for hv inverters and diy. Not many easy solutions for hv batteries. For example I'm not aware of a bms that can do 95v

How were you able to get the AHJ to sign off on non UL listed DIY batteries? This is probably the most difficult problem with DIY solar forced to buy very expensive UL listed batteries. My utility even requires it to be California certified which is even a step above UL I'm in Arizona.

The trick is to get the system inspected before adding batteries. The inverter doesn't require batteries to operate, at least mine doesn't, it'll do PV Net Meeting with just panels and the inverter. You just say you want the ability to add batteries later, if asked.

The GoodWe inverter is UL Listed and on the California approved list. That's one of the reasons I went with it.

jpwhit said:

The trick is to get the system inspected before adding batteries. The inverter doesn't require batteries to operate, at least mine doesn't, it'll do PV Net Meeting with just panels and the inverter. You just say you want the ability to add batteries later, if asked.

The GoodWe inverter is UL Listed and on the California approved list. That's one of the reasons I went with it.

Click to expand...

Just have to make sure never have a fire who knows if home owners insurance will cover a non UL non permitted battery. IT gets so hot here batteries don't do well outside in the summer without AC cooling.

k490 said:

Just have to make sure never have a fire who knows if home owners insurance will cover a non UL non permitted battery. IT gets so hot here batteries don't do well outside in the summer without AC cooling.

Click to expand...

That's why my batteries are outside the house and a good distance away. But I'm in a wooded area so they never get much direct sun exposure and it's not hot enough here for it to be a problem. It's also why I used white dock boxes. So, the little bit of sun exposure the battery boxes get wouldn't generate much heat.

In hot climates, it's really pretty simple to setup the liquid cooling loops built into the Tesla modules. You can just buy the Tesla coolant reservoir with integrated 12v pump and the radiator fan assembly off ebay from a wrecked car. You just have to put the radiator assembly somewhere reasonably cool like a crawl space or air conditioned utility room. This is also a good solution for locations where it gets pretty cold. As long as the radiator is somewhere that stays at moderate temperature all year round.

If you don't have a place to put the radiator assembly, then I would just buy a pre-fab utility shed and put the batteries in there. And I'd install a mini-split in the shed.

But at some point, if there are too many obstacles for your given situation, and the remedies for those obstacles add enough time and expense, then buying commercial batteries may be the best choice.

1201 said:

That's the big problem for hv inverters and diy. Not many easy solutions for hv batteries. For example I'm not aware of a bms that can do 95v

Click to expand...

Here's one. And it comes for free already mounted to the Tesla module and it's even pre-wired to all the cell bricks for you. It doesn't get much easier than that

jpwhit said:

Here's one. And it comes for free already mounted to the Tesla module and it's even pre-wired to all the cell bricks for you. It doesn't get much easier than that

View attachment 220960

Click to expand...

And it talks to the inverter?

Looks like s2s has this inverter for $2850

1201 said:

And it talks to the inverter?

Click to expand...

No, not directly. I built a custom BMS master that understands how to talk to the Tesla BMS Slave via the ISO-SPI interface and then communicates with the GoodWe inverter via it's CAN BMS interface.

If there were signs of enough interest, I'd possibly consider having PCBs made and make that available to other DIYers.

1201 said:

Looks like s2s has this inverter for $2850

GoodWe GW9600A-ES 9.6kW Split Phase Hybrid Inverter with 4 MMPTs| Compatible with BYD HVL Battery Certificated UL9540

The GoodWe A-ES Series is a split-phase hybrid inverter designed to increase the self-consumption of your generated solar energy. GoodWe A-ES is compatible with high voltage (80-495V) batteries with a power capacity ranging from 5 kW to 9.6 kW. With up to 4 MPPTs, the A-ES inverter seamlessly...

www.self2solar.com

Click to expand...

That's a good price. They may be clearing stock because the new GW ES-US 5-11kw hybrid inverter is supposed to start shipping about now.

Really enjoyed the write up and details, I'd ask if you can keep adding your notes and experiences in this thread. You've done a lot of good work here and others can benefit.

jpwhit said:

No, not directly. I built a custom BMS master that understands how to talk to the Tesla BMS Slave via the ISO-SPI interface and then communicates with the GoodWe inverter via it's CAN BMS interface.

If there were signs of enough interest, I'd possibly consider having PCBs made and make that available to other DIYers.

Click to expand...

the extra complexity is what kills hv for most people, including me.

jpwhit said:

That's why my batteries are outside the house and a good distance away. But I'm in a wooded area so they never get much direct sun exposure and it's not hot enough here for it to be a problem. It's also why I used white dock boxes. So, the little bit of sun exposure the battery boxes get wouldn't generate much heat.

In hot climates, it's really pretty simple to setup the liquid cooling loops built into the Tesla modules. You can just buy the Tesla coolant reservoir with integrated 12v pump and the radiator fan assembly off ebay from a wrecked car. You just have to put the radiator assembly somewhere reasonably cool like a crawl space or air conditioned utility room. This is also a good solution for locations where it gets pretty cold. As long as the radiator is somewhere that stays at moderate temperature all year round.

If you don't have a place to put the radiator assembly, then I would just buy a pre-fab utility shed and put the batteries in there. And I'd install a mini-split in the shed.

But at some point, if there are too many obstacles for your given situation, and the remedies for those obstacles add enough time and expense, then buying commercial batteries may be the best choice.

Click to expand...

I probably could do that with lifepo4 battery not a nickel battery don't have the room being in a tract home.

jpwhit said:

If there were signs of enough interest, I'd possibly consider having PCBs made and make that available to other DIYers.

Click to expand...

Enough interest to control and reuse the most popular EV batteries out there... I bet yes!
Do you need direct access the slaves, and so open the battery, or you can use the connector on the battery penthouse?
That can do a big difference.
Here a 82 kWh Tesla battery on the side. Don't take much place.

I've done it both ways. Whole pack vs taking the modules out. There are Pro's and Con's both ways.

The main con of the whole pack approach is simply the logistics of moving around a 1000lb pack without anyone getting hurt. The other con for the main pack approach is then you have to start dealing with a lot of variations in the firmware versions on all the different Tesla controllers in the Penthouse. And there are a bunch of them. And the APIs do change between the firmware versions. There's the BMS Master, Charge Port controller, PCS 1, PCS 2, and the HV controller. Most of the solution you find out on Github don't handle all the different firmware versions of these controllers. They only work the firmware version that match the pack of the person that created the github project. So, in practice, they can be very hard to get working and usually require you to modify the code you found on github and/or reverse engineer the system yourself. And figuring out how to change the Tesla firmware is pretty impractical w/o help from someone inside Tesla.

To use the main pack, you also have to hook up and use the liquid cooling system even if it's not needed by the batteries themselves due to low charge and discharge rates typical in solar/ESS. That's because the penthouse electronics are liquid cooled. Lastly, it's actually kinda hard to keep all the Tesla safety systems in the main pack happy all the time. And that can lead to some reliability issues with the system getting unhappy and opening the HV contractors at the worst possible times.

The main pro is you're getting all the safety systems engineered into the Tesla packs. And I could write a really long post detailing all of them, but it's probably a lot more than most people realize.

The main con of taking the modules out, is you have to really understand how to handle them to not damage any of the per-cell fuse links, or the laser welded links between the BMS slave board and the flexible circuit board interconnect to each battery brick. But assuming you know how to prevent that damage. They are much easier to handle since 2 people can pick them up and move them around. The other con is you're giving up a lot of the safety engineered into the overall pack. But there is still a good bit of safety systems you're gaining over most DIY assembled battery packs engineered into the modules.

The Pro is the slave BMS boards don't have firmware on them. So, once you figure out how to interface to those boards, it'll work for pretty much every flavor of module that Tesla has produced over the life of the Model 3 and Y.

There is a hybrid approach that I'll probably use soon. That's to disassemble the whole pack so things can be moved into place in much more manageable size pieces. Then reassemble the pack in the desired final location, but don't reinstall all of the penthouse electronic modules. I would just reinstall the HV contactors and the HV connectors along with the interlock system that keeps the HV connectors from being energized when the appropriate cables are not connected end to end like it works in a Tesla vehicle. Such that you can never have any high voltage where someone can touch a live contact. Then I would have my custom controller run the contractors and HV cable interlock system, and it would talk to the BMS slaves to manage the batteries. There would be no dependence on any Tesla firmware that can vary.

I also think the Tesla modules / packs of choice are the newest LiFePo4 (LFP) modules that are now available from standard range vehicles. They don't have any of the handling issues of the Li-Ion modules and all the fuse links. They have the natural increased safety of LFP vs Li-Ion cells. The trade-off is lower energy density and as such being a 60kwh pack instead of 82kwh. But for fixed location home usage, it's worth that tradeoff IMO. And it'll still be the cheapest per KWH battery system a DIY person will be able to get IMO. And you get the safety of the Tesla case's fire and hot gas containment engineering. You also wouldn't need the liquid cooling system.

And I don't think you'll see post on here where the pack is done after 3 or 4 years with swelled cells like you do on here from LFP cells sourced from 2nd tier manufacturers. So, in the long run, since these packs have a history now of consistent 10+ years of usage, it's going to be order of magnitudes cheaper per KWH times years of viable service of the pack.

jpwhit said:

I've done it both ways. Whole pack vs taking the modules out. There are Pro's and Con's both ways.

Click to expand...

Wow! Thanks for all those details. Really nice!
I'm also a big fan of reusing high quality EV battery.