Sol-Ark charge/discharge efficiency?

[cm119]

I have Foretress eFlex batteries on a 15K, and notice that the charge/discharge efficiency seems to be pretty bad according to the data. I do have some cooling fans coming directly off the battery bus so that is part of the issue, but it can’t be using that much power. It looks like my efficiency is 70-77% or so. What does everyone else’s look like?

 

[pvgirl]

There have been reports of those batteries having a very high self consumption, that could be causing your losses: https://diysolarforum.com/threads/eg4-batteries-and-sol-ark-15k.66995/post-892197

 

[reardencode]

I get 76% for the few months I've had solar assistant recording data. Sol-Ark 15k + HomeGrid Stack'd 38.4kWh.

I've come to understand that the Sol-Ark is significantly less efficient when it is both inverting and charging, so changed my TOU settings to limit the time where it's doing both. Will see if that makes an improvement over the coming months.

 

[Hedges]

There will be some no-load losses, I would think not much in battery but moderate in the inverter. You'd have to gather data at higher power levels to get a reasonable number, since efficiency goes to zero at no-load.

Other people have reported seeing twice the expected loss, like 82% round trip vs. 91% calculated from specs. Quite a bit of their load was LED lights, which have no power factor correction. poor PF will boost losses considerably, e.g. at 40% load the inverter's efficiency might be what's expected at 80% load. Losses in transistors and inductors go as current squared; double current causes 4x loss.

After testing with actual loads, try shutting all those off and repeating tests with purely resistive loads like space heater.

 

[2TrevorJ]

82-92% here for Feb‐Oct. Lower efficiency in months with lower throughput.

 

[cm119]

There have been reports of those batteries having a very high self consumption, that could be causing your losses: https://diysolarforum.com/threads/eg4-batteries-and-sol-ark-15k.66995/post-892197

Yikes, I’ll have to look more into that. 30w/battery, so 90w for me, 24 hours a day, for 30 days, 64kW/month just for the batteries to run themselves?

 

[cm119]

I get 76% for the few months I've had solar assistant recording data. Sol-Ark 15k + HomeGrid Stack'd 38.4kWh.

I've come to understand that the Sol-Ark is significantly less efficient when it is both inverting and charging, so changed my TOU settings to limit the time where it's doing both. Will see if that makes an improvement over the coming months.

I just got PTO a couple weeks ago, before that I had the sol ark setup with ‘load first’, so it would start inverting power to load as soon as the sun came up, and only started charging after pv exceeded load. Since getting PTO I’ve set it back to ‘battery first’, and the battery gets 100%of the PV until it’s charged, then the sol ark starts powering the loads and selling back. If this is true, hopefully this will help moving forward.

 

[cm119]

82-92% here for Feb‐Oct. Lower efficiency in months with lower throughput.

I just got PTO a couple weeks ago, so most this data is with low throughput doing only self consumption. Now that I’m selling back maybe it will change some?

 

[Skypower]

Original mpp LV 6548’s, three banks of diy batteries with JK bms’s. Letting the Solar Assistant “maintain battery state of charge” supplement from grid as needed during off peak hours. Seems to be a bit more efficient during the warmer months and deeper depth of charge. A lot of loss in conversion to load.

 

[TedK]

Solark 12K with 6 AG4 batteries, 97% battery charge discharge efficiency over the last 12 months.

 

[SHTF Power Anyway]

Yikes, I’ll have to look more into that. 30w/battery, so 90w for me, 24 hours a day, for 30 days, 64kW/month just for the batteries to run themselves?

Sorry I am just now reading this thread. I have 10 eFlex batteries and a very long thread on this forum including gaslighting responses from a rep of the manufacturer who now sadly works for Sol-Ark (I have two of those as well). When the batteries are actively powering loads, 30W/battery seems to be what I'm seeing. This little problem destroys the use of my system half of the year. Ask me if I'm angry.

 

[Lt.Dan]

When the batteries are actively powering loads, 30W/battery seems to be what I'm seeing.

So I just want to emphasize that part. Is it only when powering loads? Or all the time? What if there is no load?

 

[koppted]

Sorry I am just now reading this thread. I have 10 eFlex batteries and a very long thread on this forum including gaslighting responses from a rep of the manufacturer who now sadly works for Sol-Ark (I have two of those as well). When the batteries are actively powering loads, 30W/battery seems to be what I'm seeing. This little problem destroys the use of my system half of the year. Ask me if I'm angry.

You didn't end up selling those batteries? I remember reading that long thread. It was quite obvious that the Eflex rep was blowing smoke. I'd bet you before that people thought the Eflex batteries were Tier1 batteries because of how expensive they were and the so called "support" you would get if anything went wrong.

 

[Tulex]

15K with 2 DIY 304ah W/ Seplos BMS. Looks like around 94%. This months is a little off because I changed some settings.

 

[Hedges]

When the batteries are actively powering loads, 30W/battery seems to be what I'm seeing.

What do you mean by "actively powering loads"?
I assume this is not 30W "idle", with no current drawn. 30W is what an inverter would consume making 120V or 240V sine wave, but surely a battery/BMS doesn't dissipate that much.

How much power is being delivered, what percentage does 30W represent?

There is resistance everywhere, especially in FETs. Power dissipation will go as the square of current.

 

[SHTF Power Anyway]

So I just want to emphasize that part. Is it only when powering loads? Or all the time? What if there is no load?

No load, not even powering the inverters, working with MurphyGuy we figured it was a bit over 12 watts. Thankfully, we're still connected to the grid so when my bank gets low, I put all loads on the grid and allow the inverters to recharge the bank. But this defeats the purpose of the system for us.

 

[SHTF Power Anyway]

You didn't end up selling those batteries? I remember reading that long thread. It was quite obvious that the Eflex rep was blowing smoke. I'd bet you before that people thought the Eflex batteries were Tier1 batteries because of how expensive they were and the so called "support" you would get if anything went wrong.

No, I ended up sitting on them long enough to make it extremely painful to resell them. I received an offer I should have taken but it was only for 4 of the batteries at the time. Without packaging, any warranty, and with recent price drops I'm in a pickle. I also wanted to be sure I picked a good replacement. I almost pulled the trigger on Trophy batteries but Dan got sick and didn't have anyone supporting his business so I was scared away. We had funds all ready to go to rework the system at the time. I was going to move to Victron inverters & charge controllers with Trophy batteries. I was super excited and would have dumped the eFlex at a huge loss at that point but... Here I am with my overpriced, shit system. I just didn't want this person to think he was crazy. Honestly, dealing with this forum and trying to figure out where to go from here tore me a new one for months. I stopped sleeping. I'm extremely angry.

 

[SHTF Power Anyway]

What do you mean by "actively powering loads"?
I assume this is not 30W "idle", with no current drawn. 30W is what an inverter would consume making 120V or 240V sine wave, but surely a battery/BMS doesn't dissipate that much.

How much power is being delivered, what percentage does 30W represent?

There is resistance everywhere, especially in FETs. Power dissipation will go as the square of current.

Some of your question is over my head so I'll just give you my simplistic answer. At absolutely nothing attached, not even the inverters, they dropped more than 12 watts an hour each.

When the inverters are turned on, there is always a draw on the batteries when no solar is being generated because the inverters themselves have a draw. I stopped trying to chase calculating the actual loss a long time ago. The calculation relies on SOC which everyone screamed wasn't reliable. When using the BMS software I could see my SOC was all over the place but can't get any support from Fortress because there's no red warning light on any battery and they refuse to support non-failure investigations (their words, not mine). I have one battery that has now cycled 6 times more than the others. Why? Hell if I know! smh

I decided to post my 2 cents worth for this poor person seeing the same crap I did. I hope they fare better than I did in somehow rectifying the situation.

The eFlex battery are contactor based not FET. My understanding from everyone who responded to my huge thread in March is that the FET based batteries consume far less than the eFlex.

 

[Hedges]

So apparently 12W consumption with no load, maybe 30W when feeding something else?
But only known within the accuracy of SoC estimation.

Contactors do consume power, there is current going through a coil. 12W, doesn't really surprise me. I was thinking it was FET, and that there shouldn't be any power consumption except while switching, or current flowing.

"Economizers" reduce that. Voltage and current needed to close contactor is high, to hold closed much less. If you're into hacking them, you might be able to add something into the coil drive circuit.

SOC requires tracking current over time. A clamp ammeter could at least show current flow. Clamped outside, wouldn't necessarily show what coil consumes. Clamped inside, between cells and circuitry, it should show BMS/contactor consumption as well as inverter consumption (if connected). 12W at 48V would be 0.25A, and some clamp ammeters have 0.1A resolution. Some have 0.01A resolution, better for this task, but still about 0.08A error. If zeroed just before use, should be reasonably accurate for a quick measurement.

 

[SHTF Power Anyway]

So apparently 12W consumption with no load, maybe 30W when feeding something else?
But only known within the accuracy of SoC estimation.

Contactors do consume power, there is current going through a coil. 12W, doesn't really surprise me. I was thinking it was FET, and that there shouldn't be any power consumption except while switching, or current flowing.

"Economizers" reduce that. Voltage and current needed to close contactor is high, to hold closed much less. If you're into hacking them, you might be able to add something into the coil drive circuit.

SOC requires tracking current over time. A clamp ammeter could at least show current flow. Clamped outside, wouldn't necessarily show what coil consumes. Clamped inside, between cells and circuitry, it should show BMS/contactor consumption as well as inverter consumption (if connected). 12W at 48V would be 0.25A, and some clamp ammeters have 0.1A resolution. Some have 0.01A resolution, better for this task, but still about 0.08A error. If zeroed just before use, should be reasonably accurate for a quick measurement.

I appreciate the tips. I just don't have the skillset to make modifications. Measurements at this point won't get me far. I just wanted to leave my own personal experience on this thread. I wish I had more DIY in me. We DIY'd the install but that's as much DIY we're actually capable of given current circumstances.