One of the things that seriously annoy me about being on the Internet, particularly around the communities heavily influenced by Doctorow and Stallman, is that most of those members, confronted with business decisions influencing design and implementation of devices and services, can only respond with one of two discussion-terminating memes: enshittification or defective by design. I find these answers annoying because they neither attempt to understand how those decisions are made, nor identify a working strategy to make things any better.
I’m putting this preamble here because I totally expect that what I’m about to write here is going to elicit those two responses, and honestly, I don’t have time for those.
When we got the keys to our new house last year, we started almost from bare walls — it took six weeks to get an Internet connection, we (still) have no mobile network signal, and receiving any delivery whatsoever for the first few months was more than just a bit complicated, as our street didn’t exist on any navigation app. In addition to this, before we started actually living in the house, we realized that the electricals were… less than stellar. There are a number of reasons for this and I’m sure I’ll have content fore more posts in the future, but for the time being let’s just say I realised very early on that I wanted to have a battery backup for the NAS, Home Assistant, and network equipment.
While our house, like basically any new build in the UK over the past few years, was built with solar panels on the roof, they amount to a fairly low 3kW at full production, and didn’t come with whole-house batteries. Again, this is something I’ll probably get back to in the future, because I have Opinions on the complaints about grid stability, while having set up companies to skim and scam left and right.
UPSes are definitely not a new topic for me. I have (co-)maintained both NUT and APCUPSD for years in Gentoo Linux, as having multiple UPS for different rooms was a necessity, when I lived in Italy — since my mother’s house is located in a terrible zone for power supply, and still getting multi-hours maintenance blackouts at least once a month every winter. But instead of going straight to APC (pardon, Schneider Electric nowadays), I decided to take a look at EcoFlow — a company I had not bought anything from before, but that I kept hearing about both from YouTubers (paid and not) and from Ukrainians.
My primary interest here, was to have a long-running battery so that, even in the event the power was lost while we were out of the house, we wouldn’t have to hurry back to be able to monitor it. Most UPSes, even in the professional or server space, are designed to deal with short interruptions, the interim between a power loss failure and the bringing up of separate supplies such as diesel generators, or just the time needed to save open files and turn off the computer.
But there’s a different class of devices that became common in the last few years: portable, large-capacity batteries, which EcoFlow (but also Anker, and a bunch of other companies) is quite well known for. These are designed to operate for a much longer time with lower drain requirements, which felt a much better fit for the networking gear, rather than the workstation-oriented high-drain, low-capacity UPSes. And indeed, there’s no contest in terms of “bang for your buck.” With a whole lot of caveats, of course.
Conceptually, there’s very little difference between an UPS, a portable battery, or many of the demand-shift batteries that are being sold: you get the battery of cells itself, a BMS (battery management system) to handle the charge/discharge and monitor the health, and an inverter to supply the 230V output. Admittedly an UPS also tends to have a user-serviceable management interface, but as it turns out, some of the EcoFlow batteries come with a NUT-compatible USB HID interface for that as well — and pretty much all of them have Bluetooth of WiFi connectivity nowadays.
The inverter is a bit of a weird sticking point for me — basically everything I power from an UPS or a battery takes the 230V AC and will then linearize it to DC anyway. I find it annoying that we haven’t really standardized on a high-power 48VDC bus for home/small-office equipment.
To put some numbers in, at the time of writing, with £600 you can either get the EcoFlow DELTA 3 Plus (the one I have) or, depending on the supplier, the APC BR1500GI. Comparing these is hard, but I believe this is close to a fair comparison in terms of equivalent purchasing power. At the time of writing, Amazon’s price for the BR1500GI is a third down compared to Schneider Electric’s own website, showing it as effectively quite cheaper than the EcoFlow, but prices vary for both of them during the year.
The APC one actually shares the same “battery cartridge” with at least another model, which is a rackmount version that, on Schneider Electric’s website, is priced at over £1000. This is a major difference between the portable batteries and UPSes, which I’ll get back to later — when you buy an UPS, you usually expect to be able to replace its battery after some time, while you can’t really do that with a portable battery.
The APC model is rated for 1500 VA or 865W — that’s power (drain), not energy (capacity)! If you want a video essay explanation, refer to Alec’s great video over at Technology Connections. When buying UPSes for offices, you often spec them for how much equipment you can back up, and mostly accept the runtime they provide it – this is by my experience both as buyer and seller of UPSes to small and medium offices – which is why the headline number, the one that is in the model number, reflects the maximum power, and you’ll need to dig quite a bit to find a capacity value. Conversely, most portable batteries are specified based on their capacity — the DELTA 3 Plus is a 1024Wh battery, which is the first number you see around it, with a maximum output of 1800W.
The next step, for a comparison to be valid, is to figure out what the capacity of the APC version is. For this, it’s easier to observe the battery cartridge (if you think this is a strange name, that’s what Schneider Electric calls them — “battery pack” is what I’ve always called them, personally): APCRBC124. This format of battery pack, with different capacity, is quite common for Schneider’s devices — as far as I know, though, you cannot just cheat and get a higher-capacity pack for a lower-power model to extend its runtime, even though the batteries are the same size, and the same voltage (24V), and they’re used on models with different maximum drain (1500VA and 1000VA.)
The Schneider Electric battery pack is rated for 9Ah at 24V: 216Wh — note here, while multiplying volts by amperes doesn’t quite work for AC (thus the difference between the VA rating versus the Watt rating on the APC UPS), for battery capacity this is the standard way to represent this. Compare this to the 1024Wh (51.2V at 20Ah) of the EcoFlow DELTA 3 Plus.
Comparing these specs (together with a video from, I believe, Functional Print Friday, which I can’t find again showing some maritime battery comparison) is what got me interested in trying to use the EcoFlow battery instead of a “classic” UPS — we’re talking about nearly five times the capacity and more than twice the sustained drain, for roundabout the same price (or a lot less if you compare it with the “professional” rackable versions.) Sure, you need to do quite a bit more work to replace the EcoFlow if it fails (re-wiring a bunch of stuff, which particularly for a rack setup is non-negligible work), but assuming a similarly proportional life span, the economics appear to be well skewed in front of EcoFlow style batteries over “old school” UPSes with their lead-acid batteries.
Is this all a scam? Is it intentionally designed to be bad? Not really — these are generally different markets. EcoFlow has been encroaching in Schneider Electric’s home customers with adding UPS functionality to their Plus series, but that doesn’t really mean much. I have… peculiar experience with UPSes which is the main reason why I don’t go for no-brand or white label UPSes anymore — that market, the one that EcoFlow entirely ate out already by now, hasn’t been on Schneider Electric’s radar for a long time, from what I can see.
So this is already a first point of market segmentation: APC devices have still a lot of valid use cases where their price is warranted. A rack-mounted UPS where you can replace the battery pack without having to touch any of the power wiring is indeed something that is very valuable in the right conditions. Schneider Electric is also providing a lot more certifications, which again have their own value, particularly if insurances are involved.
Now, even within EcoFlow’s own offerings, there are different markets covered. I briefly hinted at demand-shifting batteries as a different beast from portable batteries. Indeed EcoFlow sells STREAM-series batteries that are designed for demand-shifting: you charge them either during the day through solar, or – if you have a flexible tariff – during the night when it’s cheaper, and then discharge them when you run appliances and it would cost you more. I could have probably used one of those (particularly the STREAM AC Pro, which allows “off-grid” usage), but they have a lot of different parameters and usage patterns compared to the portable batteries.
Can you do demand-shifting on the DELTA series? Well, in theory yes — although EcoFlow’s own support people don’t recommend it, and indeed when they do sign up folks for testing new experimental integrations with, for example, Octopus Energy, they only do so for folks that are on their STREAM or OCEAN series.
I actually wanted to be able to do so with Home Assistant integration at some point, and looked into what options I had. As it turns out, neither the EcoFlow app, nor the unofficial Home Assistant integration, have a way to stop the battery from charging while there is AC on the input. This again is market segmentation: the BMS surely has a way to disconnect the input power, but that’s not a feature needed for this type of device. I “solved” this part by attaching it via a smart plug, which allows me to “artificially” cut the power to the battery, if I want it to discharge… which is a good thing to have, to be able to test your battery — because you test your backups, or you can’t be sure you have any!
And indeed, the first time I tested the battery for a real backup usage, it turned out to cut itself out at 70% capacity, weird! Turns out, it likely was overheating — the battery is at the bottom of a shelf in the closet, and there was no real air flow even with the built-in fan triggering. So even though the battery rating is to operate at 55°C temperature (and I assume this meant ambient temperature), and the closet itself is nowhere close to that, the BMS did jump to over 60°C just before cutting out.
I have then added a USB-powered fan, which is powered by the very same battery, just to move enough air away from the top of the battery that easily lowered the reported BMS temperature by 15°C! Running with half the usual load (turning off the NAS, because there was no reason to risk the disks on the first test), I could get the battery down to 50% without overheating or cutting off. Another test with the NAS appears to confirm the finding: a fan to move just enough air away is all it takes for the battery to operate, at least in the winter — I’ll see how this all looks like in the summer.
There is another issue: the Delta 3 Plus is actually designed to work as an UPS, intentionally. It does indeed have HID protocol support when you connect a data cable (best if a A-to-C) to one of the C ports in the front — annoyingly that does mean that you can only power one device via USB-PD. But when I tried setting this up on my TrueNAS, the kernel fails to enable the HID support:
hid-generic 0003:3746:FFFF.0004: unbalanced collection at end of report descriptionCode language: CSS (css)I saw a few other reports on a bunch of forums, and a few people saying that they managed to get it to work after a firmware update. Unfortunately I have not done any firmware update on mine in a while, because the largest design issue of all, is that the firmware update process for EcoFlow is completely broken.
It is reasonable to expect that during the firmware update, the power outlets on the battery would be cut off, causing it to turn off whatever it is connected to it. It is less expected that the battery would do so before it finished downloading the firmware image. It means that your Internet connection (ONT + Router + WiFi) needs to be outside of the battery’s power before you start the firmware update, or it will just abort midway through (thankfully, no bricking involved.)
If the firmware update coders had a little bit more care to download the firmware and necessary scripts first, then no problem would have appeared from that.
To conclude — do I regret not going for the proven and tested APC BR-1500GI? I think not. The operating temperature for that model is even lower than the EcoFlow, so it’s likely it’s going to have the same problem, if the ventilation is the problem. And running for half an hour is still a lot more than the full capacity of the APC would have been. I honestly think my only issue is that the NAS I’m using uses an AC-to-DC powerbrick, causing the inverter to do most of the work — unfortunately even though the NAS I’m using would barely fit into the USB PD specs (19V, max draw 280W), it’s well above the max power of the USB-C ports on the EcoFlow.
