I saw the same video, and had the same thoughts. I saw a comment mentioning flashing the Emporia with ESPHOME, of which I am a great fan. What I really like about that option is you can push all the logic to the "edge" device with ease (including hosting a webserver), or just merely expose the data and use e.g
homeassistant to manage it. I'm sure you could probably also have it post the data directly to clickhouse if you want. All this to say esphome is really great and I love its flexibility.
I was planning to build something similar with esp32 until i found the emporia.
It's a bit of a hassle to install (don't think anything can be done about that) and I didn't really like the official ui, but it works great with home assistant / esphome. [0]
I find it odd that they're using a Pi for this, it's not really a good solution for permanent installs, and is overly complex (and slow) for just monitoring a few sensors.
It could be done entirely on an ESP8266 or ESP32 for much less cost and complexity.
Other design choices that I find strange:
- Entire RaspberryPi available but only 2 CT inputs.
- Requires 2 plugs, vs monitoring line voltage from the same input that powers the Pi.
- LCD display on the front, when it's going to be installed somewhere I don't see it.
- Boots from SD card, which will fail eventually, or get corrupted when power fails and comes back on rapidly multiple times as is typical during blackouts.
- Custom 433mhz link, instead of using something standard like zigbee or wifi.
Right?! I was watching a video yesterday by Matthias Wandel and they were using the Emporia smart home monitor. This thing has 16 inputs and it seemed to be built upon an ESP32!
For some reason I can't find the video anymore...
Edit: ha! function_seven beat me to it... (7 hours ago), it's not on his "main" channel
Is it me or do the costs seem a little punchy? The heat pump monitoring system is 646 GBP. At local electrical prices, I'd have to save nearly 6 megawatts to see any kind of return.. and that's just on hardware cost, not including install.
My heat pump comes with an app for energy monitoring and remote control and stores 2 years of energy usage. Too bad it doesn't have an API to integrate with other stuff.
I see so many IoT start-ups coming up with sensor packages to retrofit onto existing equipment, however it almost always never makes economic sense due to the installation cost. The sensors and monitoring hardware ideally should be included in the machine/appliance at the factory.
I've had great success using home assistant's built in energy monitor dashboard and a ESP32 based power monitor from CircuitSetup. Home assistant supports exporting metrics via Prometheus which I then graph using grafana for fancier dashboarding.
I’ve had an IoTaWatt running for years without any issues. I’ve used it to pinpoint the exact time two refrigerator compressors failed. It’s been handy for all kinds of other stuff.
Zamel, a polish company, makes something similar called Supla MEW-01 [0] to measure energy consumption. It's got WiFi connectivity, a tiny webserver and a REST API. It seems it's all open source [1]. Their app can use either their server or a self-hosted server. If I understand correctly, you can also consume the meter's data by contacting it directly via the API.
I've ordered one on a rave recommendation from a friend, haven't yet had time to familiarize myself with it or test it. Planning to use it to accurately measure heat pump's energy consumption.
It's a nice project - I use it at home for a very simple setup to see what my house power consumption is; shows how I can go from 5-6 kWh per day to > 9 kWh just by using the oven. Worked out how to run the emoncms aspect in a LXC container, instead of in Docker, so it'd fit into my preferred way of doing things [0].
The forums can be a bit like posting into the void - a few views, no comments.
I've got a bug issue and bug fix PR up, and it's been there for a few months with no traction. It's the web UI rather than core code, but it's not even had a response from the core committers/project owners.
For UK users, the "Bright" product (see app stores) - which is also part-branded "Glow", almost like they want to set themselves up to fail - from Hildebrand allows for real-time access to smart-meter data (as in minute to minute, not delayed by 24 hours or even 30 minutes) via MQTT. It's essentially a modified consumer HID for smartmeters (the now ubiquitous little screens issued by energy companies to show consumption on an LCD display). It's cheaper, as well as more accurate, than the other devices mentioned here, albeit at the expense of providing a breakdown by circuit.
I've avoided smart-meter data so far as it shows zero watts no matter how much I export (I don't have an export contract, it's a DIY PV install and no point with current pricing) hence why I'm looking at ones with a seperate sensor. It frustrates me no end that my IHD doesn't show export, only import!
You do have something to offer: a little of your time. I would love to hear from you what you found amazingly interesting. Could you email me so I can ask?
Contact me. Have worked with designing/engineering/manufacturing similar measurement systems before and am quite interested in getting more done in the field. The e-mail address is in the profile.
We installed hundreds of RPi for energy monitoring in a huge project, and after 3 years of operation, we only had like 3 or 4 SD corruption errors which only killed the GUI, but the rest of the system kept chugging as usual (I wonder if there's a common bug in the OS, since the corruption errors killed only the GUI).
I kept hearing that argument about the RPi, but in our experience, it's just not the case.
We shipped a second batch with additional improvements a year ago (external SSD, external battery for UPS, and external WiFi or 3G/4G dongle), and we have been error free.
All my RPis had some kind of corruption sooner or later. At some point we blamed the shitty sdcard, but it happens with the supposedly good ones too. With the shitty ones I had unrecoverable boot and fs corruptions. With the good ones it is more subtle, some library is corrupted, some python module fails import, you can fix it by reinstalling the failing component but it always leaves me with some anxiety to trust it for serious stuff.
But regardless, things have moved on and for a reliable, serious system you can move the rootfs or ditch the SD card entirely and forget about this issue. Yes, I appreciate this is a more expensive solution.
I guess not everyone got the memo else this thread wouldn't even exist. Maybe the foundation need to release a version without an SD card. That would force people to take note of the various other booting options.
It's more expensive but also sometimes impossible for space constraints. If you designed a product with the sdcard in mind it can't be difficult to make room for an external SSD.
Several comments on that video say it can be Tasmotized. It runs on an ESP32.
Amazon’s got them for $165, and that’s with 16 circuits plus the main.
[0] https://youtu.be/siJZYFCmU68