Rant and ESP8266 - 19+ Months 18650 battery lifetime
Lets start this blog post with a rant. About a year ago I posted a blog, '1+ year 18650 battery lifetime'. This blog post was nothing special, but without my knowledge it got posted on hackaday. I noticed this because I suddenly got a lot of comments on it, tried to answer questions etc. At first I was happy about this, but this quickly changed. I got a lot of hate: how **dare** I post this without providing all source code, tutorial, etc. The entitlement is just shocking. So in the end I gave up, ignored all the comments and ended up not posting any code. I regret to say I have seen this grow in the opensource world and for me this seriously kills the fun and motivation to work on it.
I do have an update on the 1+ Year sensor. Another version of the sensor that was build around an esp-12F module on a small custom PCB that reached around 20 months on a single battery charge.
The setup is the same as the previous sensor, same battery, same LDO, same software.
Based on a suggestion I also started keeping track of waketime of some of the newer sensors. This revealed that distance to access point has the most impact of how long it takes to sample and send the data point. I currently see numbers between 1.8 to 7 seconds.
Plan for the future is:
- Switch to LifePO4 batteries, this removes the need for an LDO (and are in general safer to use.), downside is the batteries have smaller capacity. I got a few of them, and am currently testing one out.
- Try to reduce waketime some more.
- Only send data on value change. I started adding this, but still finding out what the right threshold is.
I am working on a new rofi release where I will remove the old configuration options that have been deprecated for years and directly remove the old ‘xresources’ format. After this release, the rasi format is the only valid one. All theme options are handled via this. This has been in the making for long time now. I suspect this will cause a lot of hate to flow my way, not sure how to deal with this yet.
I hope this move will simplify the code base and will allow us to improve the rasi format further.
Dangerous LiDL socket.
I recently bought a ‘SILVERCREST® USB-tafelcontactdoos - Zigbee Smart Home’ power strip from the LiDL store. When I tried to use it (with zigbee2mqtt), I found out that socket 2 and 3 seemed to be joined. No matter which one I turned on, both outputs where powered.
After opening one up and measuring about it was clear that the two outputs where shorted. After removing a few layers of gunk it was clear what happened:
There is a drop of solder (only making good contact on one side) between the two outputs. Removing this fixed it.
Because this can be a potential fire hazard if high-load devices are plugged in, I filled in a form with pictures to LiDL.
No reply for a month, then they basically ask me for the same information again and fotos (that I already included in the original form). Now I don’t have the information anymore or can take new pictures, so I just resend what I had. I wonder what will happen.
Shitty Zigbee devices
A last warning I want to give is for Sonoff zigbee devices, I bought a few (motion and door/window sensor) and they are very shitty. The door/window sensors battery does not make contact, for all of them I had to add a small felt pad on the lid to make sure they where pushed into place. The motion sensor was worse, it stopped working after a while and just reports occupancy true every few seconds.
I strongly advice the Xiaomi sensors over sonoff, well worth the extra money.
please do not link any of my posts to hackaday or other websites.comments powered by Disqus