diy solar

diy solar

Battery strategy for beginner

Orangesauce

Seeker
Joined
Oct 25, 2021
Messages
31
I am building a 10kW solar system for my residential home in Colorado.

Last night. @Hedges was super helpful in steering me on panels, and also got me thinking more about my rather lazy approach to batteries.

My past experience has been with a APC Matrix 5000 for computer usage. It suited the purpose fine, but I want to back up the house with protected load panel.

By the time I am ready to purchase batteries, I will have a load plan. For now, I'm just trying to decide on a strategy between expensive plug-and-play or custom and cheap. I need to get comfortable with the different responsibilities of various configurations so I can pick the battery strategy that matches my level of interest.

What I like about systems like the Pytes v5 is it has integrated heat and a fully integrated modular approach. I read documents about "runaway cells" and posts about exploding batteries and imagine unkept systems with corrosion and cables that have come loose over time. I don't want to be the guy that is way out of his safety zone and doesn't realize it until too late.

Where should I spend research time to learn the bigger picture well enough to discover what niche I'm comfortable with long term?
 
Current Connected (Dexter) is a vendor people seem quite happy with.

He has SOK, EG4, others.
Lithium batteries have a BMS with algorithms and communication, several protocols. So they come with all the joys of computers. Search the forum for what you're considering, so you buy with eyes wide open.

 
He has SOK, EG4, others.
I think I realized I'm going to spend the extra money on Sol-Ark after our conversation last night. They have a battery list and I'll cross-reference that with line card of suppliers such as CC, thanks :)

Lithium batteries have a BMS with algorithms and communication, several protocols. So they come with all the joys of computers. Search the forum for what you're considering, so you buy with eyes wide open.
I'm more comfortable with these kinds of things, it's pretty easy for me to debug this stuff.

Am I making a mistake by just pressing the easy button from the start though? I feel like I ought to learn what I'm not going to do before I just ignore it (and potentially regret that thousands of dollars later...)
 
Research the benefit of battery communications vs controlling via voltage. If you have a shunt, it provides similar info. Unless you are doing stuff like shutting down loads at various SOC, controlling with voltage is usually more than adequate. That opens up more options for batteries.

Look at what app is available for the battery. And what info it provides. Being able to watch individual battery health is helpful. Also if Solar Assist can read the info.

Look whether you need or want UL9540 listed batteries.
 
Look whether you need or want UL9540 listed batteries.
This is an issue for me.

I went a lower cost route with repurposed EV batteries. But I can't permit my battery system.

I may end up buying 200 or 300 AH of UL9540 listed battery modules to get my system permitted and inspected, then switch back to my EV battery system. But I am kind of betting on a friend doing a system to sell them the batteries after. My brother has talked about adding storage for a couple years, but he keeps dragging his feet.

My home built battery enclosures work great and they were fairly cost effective. They are also completely safe and a bit over built. Make solid connections, use the right fuses and a quality BMS, and it is not too hard to do. For just the batteries, I have about $4,000 USD on the cells and another $1,500 in the cabinets, BMS units, cables, bus bars, fuses and holder, etc. So call it $5,500 for 36 KWHs (720 AH at 52 volts) of battery capacity. I would need 7 of those 100 AH server rack batteries to match the capacity.
 
Research the benefit of battery communications vs controlling via voltage. If you have a shunt, it provides similar info. Unless you are doing stuff like shutting down loads at various SOC, controlling with voltage is usually more than adequate. That opens up more options for batteries.
Aha! This makes perfect sense, thank you! So not just about connection hygiene, but knowing where to monitor.
Look whether you need or want UL9540 listed batteries.
Good point, my AHJ probably requires them as I'm in a metro area. And if I need them, then my decision kind of gets forced before I learn about the details.
That's something I was also scratching my head over. Well placed breakers / fuses along with the monitoring can make the systems pretty failsafe.
So call it $5,500 for 36 KWHs (720 AH at 52 volts) of battery capacity. I would need 7 of those 100 AH server rack batteries to match the capacity.
Another light bulb. I need to look at my math again, I think I lost something in (or forgot) the "AH at voltage" vs kWh conversion.
 
I think I realized I'm going to spend the extra money on Sol-Ark after our conversation last night. They have a battery list and I'll cross-reference that with line card of suppliers such as CC, thanks :)

People are happy with SolArk, especially US based support. Only grumbles we hear are that price is so much more than similar models by same overseas manufacturer.

But keep learning. Take a look at Midnight, which has new US built HF inverters with high surge. They hope to keep adding functionality by firmware alone, though not guaranteed. These are the guys who build the off grid industry starting decades ago, and designed key products for several big name brands.

I'm more comfortable with these kinds of things, it's pretty easy for me to debug this stuff.

Even if you can sniff bus traffic and understand it, what do you do if BMS or inverter doesn't do what it ought to? For instance, if BMS decides to block discharging or charging after sitting for a while at float voltage?

The new EG4 PowerPro was found to have protocol compatible with SMA Sunny Island. I think one member had it sitting in a grid-backup configuration, and when grid disconnected Sunny Island shut off because battery voltage dipped by 50%. (my AGM would never do that.)

Or, inverter doesn't reflect correct SoC of battery? Current Connected tried to get SOK working with Sunny Island, but had problems with that.

GXMnow uses Schneider with both their DC coupled SCC and AC coupled microinverters. Had to add a uP giving commands to make inverter charge from microinverters while on grid.

Am I making a mistake by just pressing the easy button from the start though? I feel like I ought to learn what I'm not going to do before I just ignore it (and potentially regret that thousands of dollars later...)

The more you know, the better selection you can make.

Once you buy something expensive you have painted yourself into a corner.


Research the benefit of battery communications vs controlling via voltage. If you have a shunt, it provides similar info. Unless you are doing stuff like shutting down loads at various SOC, controlling with voltage is usually more than adequate. That opens up more options for batteries.

What's missing with just voltage/current of battery connection is individual cell voltage. With communications, BMS can tell inverter to reduce charge rate and allow balancing. It is expected that as battery ages, there will be more runners, so might be fine initially but have problems later.

If you have limited production, I think controlling loads based on SoC, time, light should be done. But doesn't have to be through the inverter, maybe a different monitor could query BMS.

Look whether you need or want UL9540 listed batteries.

UL listed battery, or UL listed ESS including inverter and battery.

EG4 has 18kPV paired with PowerPro as listed ESS. (make sure the one you buy comes with stickers, not just a loose piece of paper, if required by AHJ. Vendor said sticker can't be added in the field.

There are a few others. At this time, often quite expensive, e.g. I'm going to use Sunny Boy Smart Energy when it receives lists including from CEC. If I add HV battery, BYD, it is twice the cost of server rack types.

I went a lower cost route with repurposed EV batteries.

So call it $5,500 for 36 KWHs (720 AH at 52 volts) of battery capacity. I would need 7 of those 100 AH server rack batteries to match the capacity.

Time heals all wounds? Or opens new ones?

At $1000, 7x server racks might be $7000 today.


Although I find the 14kWh PowerPro package and features more attractive. $3800 for 14kWh, $7600 for 28kWh, $11,400 for 43kWh

 
Time heals all wounds? Or opens new ones?

At $1000, 7x server racks might be $7000 today.
When I decided to build my batteries, the cost for (the yet to be released) SOK 100 AH server rack battery was $1,700 each. Competition brought it down to about $1,500 each about 2 years ago. I figured it would have been $12,000 vs my $5,500 for the battery. And I saved about $1,000 off my electric bills each year over the 2.5 years since the battery bank went online. Waiting until now, would have still cost a bit more, but no, the savings is not as much now. And there is certainly risk going with any used cells. I paid a bit more than surplus pricing and my first 18 KWHs were actually new cells. Then I paid even a bit more per KWH for the second 18 KWHs as the demand had gone up a bit. They were newer cells and rated a little higher in capacity, but they did have a few cycles on them. The car they came out of was totaled with under 100 miles on it though, so how many cycles could it have had?

My hope is to get 7 years on these cells. With the increases in the electric rate here, that will have "saved" me about $10,000 which should then buy me UL rated batteries. OR... With all the new EVs on the market, I might be able to double my capacity for even less ;-)

I actually tried several routes to buy a Tesla Powerwall2. No Tesla licensed installer would work with my existing Enphase system. And no dealer was willing to just sell me the unit for self install. What I ended up with is more powerful and double the capacity. I would have really needed 2 PW2's.
 
My original goal was to get bidirectional charging set up with something like the dcbel r16. They got UL mid-last year, but aren't in general circulation yet. And few cars support the necessary ISO standards. I think the VW id.3 is the closest to something I'd want to drive at the present time. Hopefully the electric Macan / Q6 e-tron will support it.

Although I find the 14kWh PowerPro package and features more attractive. $3800 for 14kWh, $7600 for 28kWh, $11,400 for 43kWh
Or $22,400 for 86kWh? I ask because if I can get a Tesla 3 around that price, then I can drive the batteries for repair when they fail. Or drive them across town to recharge during an extended outage. Or drive them to the grocery store to pick up milk. None of these racks seem to be able to do this yet :ROFLMAO:. But of course neither does the Tesla support bidirectional, and Elno seems to be very reluctant to do so.

My hope is to get 7 years on these cells. With the increases in the electric rate here, that will have "saved" me about $10,000 which should then buy me UL rated batteries. OR... With all the new EVs on the market, I might be able to double my capacity for even less ;-)
Much like driving cars with CarPlay, I replace my phone more often than my car. And I'd expect to replace my car more often than my home ESS. Or at least when I replace my car, I don't want to *also* have to replace my ESS. Getting them in the same package seems like a win-win. While cars do spontaneously combust in garages, it seems like it occurs less often than it does to home ESS systems.
 
Go light on batteries for now. Some inverters are battery optional, even batteryless backup while the sun shines. 4x used automotive starting batteries in series is all that's needed for huge starting surge. (Hopefully can set low battery cutout to avoid cycling them, something around float voltage but with several seconds delay when lower.)

Some forum members report savings having used batteries, at whatever price they were. Like GXMnow, above. Same goes for PV panels and everything else. When cheap enough to give significant savings, time to stop worrying about missing even lower prices.

Firmware functionality is something to look for. Some like SolArk and Midnight Rosie, expect support and upgrades. I use Sunny Island, one I just commissioned has 2007 firmware. I won't update unless I need something available later, but they still don't incorporate time of use shifting beyond connect/disconnect from grid. (other products offered for that.) But I got great price (thanks, DC Solar and Warren Buffet!) and arguably the best, most robust off-grid inverter ever made. (Early Trace that was truly low-frequency with multiple transformer taps, and "Warpverter" DIY today, also have a claim to that title.)
 
Go light on batteries for now. Some inverters are battery optional, even batteryless backup while the sun shines. 4x used automotive starting batteries in series is all that's needed for huge starting surge. (Hopefully can set low battery cutout to avoid cycling them, something around float voltage but with several seconds delay when lower.)
That's kind of what I've been thinking too, but wanting to learn about it all the same to help with inverter selection.

My ideal inverter would have some kind of external integration options. For instance, Home Assistant internally uses something called "Event Driven Architecture". This is a software construction where there is no explicit wiring or dependencies between software modules, rather all software modules subscribe to a stream of messages. If one module wants to communicate with another, it puts a message on the bus, knowing the other module will "hear" it. These streams can be externalized so people can write their own listeners that do things when they see events (alarms, conditions, state changes, etc are all events...)

This can be taken further by standardizing the hosting of the modules, for instance with containerization (such as Docker). Again with Home Assistant, the software plugins can be docker containers and they still just listen to the event bus. But now the creation of the plugin is standard fare for developers and more importantly the container orchestrator provides significant security guarantees for the host.

I guess where I'm going with this is if I could find an inverter that could host docker containers, I'd be throwing endless money to get one. Then the best place for my home automation to run is inside the inverter itself – right where all the uninterrupted power is!

Are there any inverters out there that are open source hardware or software?
 
This is a software construction where there is no explicit wiring or dependencies between software modules, rather all software modules subscribe to a stream of messages. If one module wants to communicate with another, it puts a message on the bus, knowing the other module will "hear" it. These streams can be externalized so people can write their own listeners that do things when they see events (alarms, conditions, state changes, etc are all events...)

I inherited some Labview written with a bit of that. Layers of routines (layered on the flowchart page) which put text strings on a software buss, and routines could insert field of text. Made debugging a lot of fun. It manipulated I/O drivers feeding pins of an image sensor and read ADC cards that digitized outcoming analog voltages. All in a PXI chassis running Windows. Not directly driving the pins, but filling memory of a pattern generator which then clocked through at hardware speeds.

Might have been a lot simpler wiggling I/O pins of a uP or gate array.


Are there any inverters out there that are open source hardware or software?

See "Warpverter" based on the clever design of Trace, from the engineers now at Midnight.

 
See "Warpverter" based on the clever design of Trace, from the engineers now at Midnight.
That was a very satisfying read, thanks. You have some serious analog chops. I imagine your remote compound is close to a long enough runway for the occasional turbine powered air taxi to whisk you to emergency jobs that few can comprehend. Please don't ruin the fantasy if that's not true ;)

Anyway, I now understand your previous post much better as well.

I've seen that before where an EPROM is used for waveform generation. That's far superior to having a real-time OS running repetitive code that provides no value. Code running on something like an inverter should be able to reboot in a loop with no ill effects to the critical functionality.

The software end users load on to a box like I am envisioning would have no business telling the inverter how to operate unless the code is signed by the manufacturer. The last thing I'd want to see is my inverter infected with SCADA malware due to some community container that did great things for home automation but had a zero day software supply chain vulnerability. Malware in this case could start your house on fire and possibly damage the grid or kill linemen if the attack was synchronized with a local blackout. Yuck. But watching events that the inverter is sensing on the grid or weather feeds from the internet could go a long way toward predictive SoC planning.

Much in the same way all-in-one inverters simplify the install for weekend warriors, running home automation containers inside the inverter would aim to further reduce the sprawl of electronics in the typical smart home. Running Home Assistant on the inverter means that (in effect) the inverter can very reliably turn off the loads in the home during outages – there are no external devices in the control plane that might have lost power.

I will spend more time looking at Victron in the coming days for sure.
 
Yeah, I forgot that Warpverter steps through a table in memory.

For my project, memory dump pattern generation with 10ns steps made it difficult for me to do the setup and hold timing I wanted, while also reading out 1 megapixel. Maybe if I slowed it down to 50ns steps. I wanted counters for the delays, as I tuned it to get good data from pixels.

This was just a test coupon; the real camera was 16 Mpixel. I used it to check sensors exposed to "environments."

It's another guy building Warpverter. I've only done mental exercises fitting utility transformers as the "Papa Bear", "Mama Bear", "Baby Bear" (names given by Trace engineers.)

I did develop boards for a quadrupole atomic mass spectrometer a.k.a. RGA. That was fun, sniffing air and then other guys used it for the isotopes of Kr. It used a chip with another waveform generation technique, DDS. I think it computes values rather than using a table, should be better for a clean signal. It was the 11 MHz sine wave source.

Now I play with electron optics and interference. That's where my eye for electromagnetic fields comes in.

I got out of one emergency overseas assignment by catching Covid, another by electrocuting myself. But a nice domestic trip I accepted. My verdict was a description of something that turned out to have a name: Chromatic Aberration.

I have a number of half-finished compounds. Working on a couple presently.
 
Back
Top