Convert an existing wired doorbell into a smart doorbell using ESPHome
frenck.dev
frenck.dev
[1]: https://www.mattmillman.com/info/crimpconnectors/dupont-and-...
For my own crimping of dupont, I'm using the Iwiss SN-025 which works OK, but even then I've mostly switched to Mini-PV because I was tired of the dupont receptacles wearing out after 5 cycles.
The pictured SN-28B is definitely not good for dupont, which has a totally round cable holding ring. Amazon/Aliexpress sellers don't know anything about what they are selling.
For the task of connecting 2.54mm “pin headers” on boards you prefer mini-pv. A little hard to search for, but does this require a special crimper of its own?
As a hobbiest you don’t recommend doing your own crimping at all, and I am willing to pay money, but I do kinda need specific lengths due to a tight enclosure without much room for excess ribbon (I guess round cable might be bit better). Is there such an option? I kinda searched for it, and on electronics retailers like digikey I didn’t really see the length I wanted, though it is easy to miss stuff on there.
The IWISS SN-025 works well for the Mini-PV too. Which makes sense, since the dupont connectors are a near-clone designed for cheaper production.
> As a hobbiest you don’t recommend doing your own crimping at all
I just don't find it very enjoyable, especially getting to the point where I have tools that are actually suitable! The IWISS SN025 is fairly new[1] and before I ran across Matt's blog, reliable information on this subject was hard to find.
But I mean I have after all gone through the effort of getting tools and connectors, so it's more of a medium "recommend against" rather than a "not worth doing at all". It's almost like my preference with assembling PCBs. I don't mind that as much, and am good at it, but if I can just pay someone $20 to do it for me, I'll do that instead. Except I don't think there's any services at that price point for cable harnesses.
[1]: https://www.mattmillman.com/iwiss-sn-025-another-head-scratc...
I don't know of any either, but you can design around ready-made cables. Mouser and Digi-Key carry a pretty big selection.
Despite your link saying they "suck", I've found the light-touch, fully-manual nature of them give a much better feel for what you're doing (vs the cheap ratcheted ones; never used the official tool), and I got pretty efficient at crimping with them throughout my Voron 2.4 build.
I do recall that the design of the one I have, has little protrusions on one side of the die to encourage the wings to wrap under, so that they form a sort of m shape, with the ends of the wings "pressing" into the wire's sheathing for a nice grip. They cost somewhere in the region of £40-60 but I don't recall exactly.
I need to go and find out why they're not on their hanger on the pegboard I'm looking at, where they should be.
It's designed for an oval crimp around the insulation & a m-crimp on the wire. The typical cheap crimpers have two sizes of m-crimp, which is fine for a lot of JST connectors for example, but not this.
I've not seen an m-crimp on both sides yet, sounds like it wouldn't work very well at all unless you had two wings on the connector itself to m-crimp the ends of, which I've also never seen!
I used a Engineer PA 09 which, despite its cheap look, is f'n amazing!
I crimped all wiring on my Voron myself, took some time getting used to it - but was almmost Zen-like meditation afterwards :)
Cheap tools may not crimp as well as the official tools (I have no doubt the differences in finished crimp geometry are real), but especially for low voltage connections like these, I have a hard time believing the end result is dangerous, or even particularly unreliable.
And you're probably getting your connector and its mate from different manufacturers, too, which can have its own set of issues.
Why not go the other way entirely? Disable the chime, install a door knocker, and never worry about replacing outdated tech, dried capacitors, shorts or anything else for the rest of one’s life.
I have a [local only, no cloud] doorbell with a camera that does person detection. It's one of my favorite devices and well worth it to me, especially since my delivery person basically never bothers to actually ring the doorbell.
Would you maybe share with us how you built it/which parts you used?
The wifi connection for the doorbell isn't ideal for security purposes, but it's good enough and a lot easier than getting CAT5 to my doorbell. It is more or less 100% reliable with my home wifi setup, though again I'm not really counting on it for any sort of reliable home security.
OP is absolutely right tho that this setup requires work to put together and maintain. Configuring Frigate can be especially confusing, but there are other local options like BlueIris (paid, and Windows only which kills it for me but it is very capable). I've been in this game for many years so it's all easy to me, but getting started is definitely a learning curve. Very fun though, if it's your sort of thing. The DIY home automation community is in a bit of a golden age right now, IMO.
You also need a Protect NVR to record video, manage the camera and provide notifications - the Cloud Key Plus works perfectly for up to a dozen or so cameras (uses standard laptop hard drive) - but they have a variety of devices that can run Protect from the UDM Pro (also will serve as a router/firewall) to larger dedicated Protect NVRs with four or seven disks depending on how much storage you want.
It can be a bit pricy for just a doorbell camera, but if you add a couple of other cameras then the value gets significantly better. Cloud is optional other than being required for initial configuration and is used for notifications, remote access and remote administration.
You don’t want to know how many components are involved in your regular smart doorbell. It is most likely going through the cloud too.
I use Home Assistant, with hundreds of devices in my home at this point. I spent so long looking for the right video doorbell and chime that I ultimately ended up with something both costly and somewhat unreliable.
I use the Amcrest AD410 with a custom component that polls it (issue 1), with all of the cloud and online stuff blocked in my network, and an Aeotec Siren / Doorbell 6. The Aeotec siren is utter trash and I'm now on my third one; two units had the audio hardware fail, and a silent door bell is pretty useless.
The one I have now hasn't failed yet, but it can be extremely unreliable and get lock-ups requiring me to open it and short out or de-solder the battery connector to get it to behave again if someone presses the doorbell too quick.
The button part has never been the hard part; the siren is; given there seems to be so little, high quality, smart siren hardware available I'd probably build one half of this setup; bell/siren + ESP and use a Z-Wave or Zigbee button to "push" the action of pressing the bell and have HA "ring" it.
Aside: Since the door chime is so unreliable, I have HA set up to announce the door on speakers, and lights to blink in various places when a press is detected too.
I'd then finally replace the multi-purpose video doorbell with a dedicated IP camera and this simple button; As I live in the UK, RMAing my first AD410 was a hassle (had to send it back to the US when it failed), and I was without a doorbell for weeks.
Since my front door is south-facing, the combination of direct sunlight, then cold cycles, coupled with heavy-handed delivery people smashing the button day in day out, basically crumbled the (matt black) casing to dust within 4 months of fitting it. It uses pogo pins to deliver power from the rear of the casing to the electronics so the compromised integrity of the casing resulted in it power cycling constantly.
There are just too many unreliable and expensive parts to what is a simple yet essential function of being a door bell, and the lesson I've learned is that the video functionality is best _not_ integrated. I _do_ still want a smart door bell though for its many benefits.
Is there a way to link devices directly over the WiFi with ESPHome?
EDIT: I think I understand what you meant now; not specifically ESPHome, but the nature of the deployment in the OP takes the "smarts" out of the critical path, allowing it to fall back to being an electro-mechanically actuated doorbell. This is certainly ideal, but I don't have anywhere I can place/power a chime near the doorbell hence the desire for a wireless connection between them. Would be great if I could somehow link two ESPHome devices wirelessly without the server so they can communicate if it's down (I suppose I could with an REST API if nothing else).
Furthermore, you can have wireless communication between ESPNodes: https://esphome.io/cookbook/http_request_sensor
It'd be pretty amazing if ESPHome supported something like ZeroMQ[0], so you could talk between nodes in anything up-to full-mesh at a socket-level and not need to worry about the availability of a server to manage the traffic.
Is it me? Have I been misusing my 8266s all this time?
Otherwise, good article, nice idea, great conclusion!
“This relay module board for the ESP-01(S), It comes pre-soldered, and the ESP will just slide on top of it.”
So, bending that pin breaks a connection between the CPU board and the relais board, so that pressing the doorbell wouldn’t be able to send a reset signal to the CPU, or, as the article says:
FTA: “Without this modification, a doorbell button push would result in a reset/restart of the chip, which of course, isn’t what we want”
That made me wonder why the hardware is hooked up that way by default in a device that gets sold for this specific purpose.
A few searches learnt me that the device has a deep sleep state where it uses very little power. The only way to wake it up externally from that state is through a reset.
https://johnmu.com/quick-boot-button/ taught me that it can boot in about 100ms, and get WiFi on in a few seconds.
=> I think that keeping that pin in place is the intended way to use this hardware combo.
The device would boot when the bell is rung, first close the relais switch to ring the bell, and then do any home automation stuff a few seconds later. Latencies look acceptable to me.
I ideally would want to power it completely from the button push or, at least, from the same adapter as the doorbell, but both will be challenging. The first doesn’t deliver enough power for long enough, the second likely doesn’t run on 3.3V.
Charge up a capacitor, maybe?
That makes much more sense.
Yeah, I guess you would need to sound the doorbell every boot, and that might be inconvenient in places with a lot of power supply issues / loadshedding.
That said, that quick-boot-button link? That's some brilliant info right there.
Funny that the store bought wireless doorbell I have does exactly that. :D I guessed it does that to avoid/reduce returns from people who doesn't manage to pair it with the doorbell button and think that it is "not working". So pre-emptively they do a chime right after you plugged in to show that it is working.
It disconnects from wifi when rebooting, doesn’t it?
And now if this is wrong, hopefully I’ll be corrected and we’ll learn more. :)
So yeah, turns out it's nothing to do with the 8266's board, and everything to do with the chosen relay module. Since it has nothing to do with driving the relay, it doesn't need to be connected. But if it were connected, whenever it is high, the relay board connects it to the reset, so the chip gets reset.
So you are bending it simply to ensure that pin can't be plugged in to the relay
Anyway, I also realized my chime wasn't audible in some parts of the house, so what I did instead was disable the hardwired chime completely, then get a fairly cheap wireless doorbell with 3 plug-in receivers scattered around the house. An esp8266 module (Wemos D1) is wired directly to the doorbell button, supplying 3.3V to ring it.
I use Home Assistant to send the signal from the Unifi doorbell to the esp8266. The biggest downside is the need to have Home Assistant, Unifi Protect and Wifi all running for my doorbell to work, but at the same time, I have a lot of automations that stop without those, the doorbell is not that high on my list of critical priorities. People can knock.
Kind of wish I had this when I had a baby that slept during the day: it would be trivial to schedule the chime not to ring at certain times of day.
I did a write-up: https://imgur.com/gallery/wifi-doorbell-chime-QTYYQs4
It's no wonder delivery people just knock and leave nowadays.
https://github.com/IOT-MCU/ESP-01S-Relay-v1.0/issues/1
https://github.com/IOT-MCU/ESP-01S-Relay-v4.0/issues/1
https://rabbithole.wwwdotorg.org/2017/03/28/esp8266-gpio.htm...
Some models have an option of measuring temperature and/or power consumption as well.
...thing that is 2 inches across and 0.5 inches tall to use as a doorbell button?
The irony of the author responding to another request on his site with the sentence "well… it is just an on/off switch".
https://community.home-assistant.io/t/make-a-dumb-doorbell-s...
Many existing smart doorbells do actually use the existing wiring and chime of the doorbell. An example is the Google Nest Doorbell.
So in terms of "wasting what you already have" what you're wasting by going with a typical consumer product is basically a button that you can buy from Home Depot for $4.98.
Now, obviously, someone interested in this project is probably not interested in having a proprietary Google camera on their door. But in my experience most people who want their doorbell to be "smart" actually want it to be a security camera and a two-way communication device so you can relay information to the person who is ringing the doorbell. Otherwise, what good is it to know that someone rang your doorbell when there is no other information? That amount of information is just like a random timestamp, at that point I'd just keep my dumb doorbell.