I love some of the stuff Andreas had done. He's one of those hidden gems of UKposts.

Fantastic work!! Love when all the knowledgment of serveral people mixes and gets improved to this levels. I haven't tested that router OS but certainly i will soon. Seriously really nice work!

That's cool to see solutions accross channels being developed further.

Each of your videos is a precious gem. Keep on so we can dream about living in the future.

Awesome. Love it. Thanks for sharing.

Keep going, CN. The world needs more of this.

You're awesome, man. Keep at it!

Awesome idea with the google sheets. I hadn't thought of that and will definitely be using it for future projects.

Everything in this video is incredibly brilliant! Thank you.

This is awesome!! Great video. I stumbled onto your video through the YT rabbit hole. This is a great project to add to my project of custom BMS for lithium based off-grid power storage. A few of use are building the BMS based off of TI's BQ76940 IC and want to be able to have communications via wifi with minimal impact to the packs voltages. Even if this wouldn't work for the powerwall application, it'd be great for remote and/or mobile projects like eSkateboards and eBikes, etc. Thanks for the info and hard work!! 🤠

Stumbled across you😀 big fan of Andreas Spiess. Really caught my attention, you have another subscriber and thank you so much for sharing. 🙏

Awesome work - thanks for sharing!

Waking up the esp by a interrupt, it could become a responsive "raw-wifi" button using the router as mqtt client for home automation, skipping all the needed time for the handshake and IP protocol. Of course a kind of encryption to make raw packet "safe" for the system will be needed. As always...great job!

this is amazing. thanks

LOVE this!!

Pretty cool! It is mind boggling how many tools are available. This is the future, and it is a complex and fun future!

Really cool! I love your sneaky use of Google services for data acquisition. I would also like to thank you for being one of the few overtly Christian engineers on youtube that I have found.

Great stuff

You could probably save power by taking off the LED ;-)

Awesome!

Mesmerising

Holy cow! Just when I think you're done coming up with game-changing ideas.... BAM. Right now I'm messing around with HC-06 modules, but I promise I'll some day be working off your stuff to make ESP8266 driven sensor nodes.

OR you could use an attiny + nrf24l01+ and run it two years on a single CR2032 coincell (replacing the router with an ESP8266 with a nrf24l01+ slapped on it).

Very cool. In some applications we can live without 802.11 if it gives us a big drop in power consumption.

Great video!

Great video with a ton of things to ponder. Correct me if I'm wrong, but what I understood is that you'll have a little esp8266+coin cell or similar traveling around with assets, broadcasting periodically, and if in range of one of your openwrt routers, they can talk. So, each particular router would be given a location name, e.g. "Warehouse B," and each esp will be identifiable, probably by MAC address, so that your google sheet will display that this esp (and by proxy its associated asset) is at this location (the range of the router). Is that correct? One hurdle would be that you have locations near enough to one another to have overlapping coverage. What's the plan for that? Do you have the flexibility to send whatever you wish, not just a voltage figure? I assume so and I suppose your wakeup cycle would send its broadcast and wait a given time for a response, which could include the router's location name to be stored to memory before resuming deep sleep. (Incidentally, Andreas did a video on storing data to RTC so it survives deep sleep.) Thus, the esp's broadcast could include the last known location as well as voltage, which is naturally useful for detecting when it finally runs out of battery. I've played around with an Amazon dash button that was well priced, maybe $2 including its battery, and could sniff its broadcast packets with a device in promiscuous mode. I was wondering if the openwrt router has to be in promiscuous mode as well? Got me wondering if the asset tracking is overt or covert...if overt, you could easily just "check in" an asset by tapping its dash button when it arrives at a new site.

Super project, can't wit for a follow up video

CNLohr I watched this video when it first came out, fantastic video. Love your other ones too. How long did the esp go for in the end?

Great work, I was a little bummed that the ESP-Now bit did not happen.

This is amazing. I wanted to take a look at ESPnow as the bootup and wifi logon times are too long for a little home automation project of mine. This is just so much better. How would you go about adding encryption or at least have messages digitally signed so you know they come from your ESP?

super cool，as always

Really cool

Amazing..

impressive stuff. well, I also worked with esp8266 coupled with atmega328 as a controller that handles and stores data in its EEPROM and certain wakeup policies were defined. We go the power very low and (watchdog timer waking every 4/8 seconds)RTC (rough as we did not require accurate timing and could sync again after some time). we achieved low power operations that could work for a long. But it can be said it wasn't real-time and that project was later abandoned and we moved over BLE. though we started that wifi thing to work with 3 coin cells but it never worked so we moved to highly efficient DC-DC converters (don't remember exactly) but I have with me battery operation data. In middle, we improved firmware just before we had abandoned project. In future, I wish to make that project open schematics and program. You work is impressive. definitely reducing up time helps. thus DTMI option is also a good solution. I didn't know Google sheet thing.. Can I use it with ESP32/ESP8266 directly sending data to sheets(no need of low power)?

Thanks alot for the great video. That's so nice, but this is consider a modification into the esp and the router to work as Wireshark sniffer. Which I can understand it's not even connecting the router as the router catch the flying packets. I got an idea based on what you have done for people who have many of ESP's which battery powered and get it reporting to one ESP working as a sniffer (directly powered) and post thier data it may saves more power that the router. (I can think of it but I don't have the knowledge to do it) but you may search it out if you find it interesting.

im looking at a home automation project, was curious on why espnow didnt work? too much battery use? Also as a relative noob, this video is super helpful, but what if anything else would i need to add/change to get it to chirp out a sensor reading instead of just voltage? thanks!

Very smart. Where can I find the latest solution to this? I see PVVX's MinEspSDKLib repo was not maintained for past 5 years. And your fork is about 3 years old. Do you still have that as the latest best solution? Or is there other better alternative in handling the power consumption of the ESP8266? Please advise.

Seems like RAW wifi is a really interesting thing to look into... Would love to see it being used for FPV with drones. This might lower latency a lot and also enable digital video transmission in "real time".

Inspirational video! How is compared using ESPnow?

I'm looking at the post-transmission power usage at about 5 minutes in. Since it seems to be sending a small unreliable one-way packet, is there any reason that it has to stay powered up at the 70mA level for around 15 ms after the packet is transmitted? You mention it staying powered up "in case" it needs to do something, but unless it's waiting to receive something after sending, it should be able to power down quickly - perhaps dropping the energy per packet by another 30-50%.

Awesome. Does the phy layer still check, if the medium is busy before transmitting?

روعة !!! +متابعة Just amazing !

Andreas Spiess has a video about ESP Now, which is designed for such short messages. But i like your approach better and it seems to take half the time.

Have you thought about undervolting the ESP? This seems the kind of application where instability is tolerable (you're not computing any safety-critical values). Since power consumption scales as voltage squared, that could make a significant difference I imagine (I'm assuming the current also lowers with lower voltage, not sure if true). I believe the most important factor (other than voltage itself) affecting chip stability is temperature, so you might have to find out the worst case temperature and find a borderline-stable voltage (maybe you can accept like 30% failure to send package?) at such temperature. Edit: Oh if that power is being used to for wifi TX antenna (instead of digital processing) then I guess that wouldn't change much... although it might be worthwhile to see the minimum TX power you can get away with too.

4:16 - your scope was last calibrated how long ago? ;-)

This is really, really cool. Now if only I had chosen better for my routers ("IoT/IP Cameras" dedicated is a linksys E2500v3 IIRC, main is TP-Link Archer C8 aka untweakable unlike C7)... I don't think my airspace can tolerate too many more networks, performance on everything dropped when I was setting up and testing the new router for my parents. This approach would be a lot more flexible than my current plan of wall-power-bound ESPs talking MQTT. Are you using "old" OpenWRT or LEDE? I'll have to look into how much flexibility I can get with the Google Forms data collection...

Is it possible to use an esp8266 on the "receiving end"? I would imagine a "server" esp8266 that listens to the sender. When it gets a package, it changes itself to a wifi client and connects to the normal wifi network and send the data to the "cloud" and then goes back into server/monitor mode. This would remove the need of a modified/flashed router. Did you look into this? Is this possible? Thanks!

Wouldn't it be easier to just send a single ICMP or UDP packet straight to the cloud server, saving requiring a custom modified accesspoint? I currently wakeup, measure temp (DS18) and send the data via http to an influxDB server, and goto deepsleep all in under 400ms, I assume moving from TCP to plain IP would be even faster, the data handling part is about 200ms.

Oh, those Russians... =)

Have you tried using another esp8266 to capture raw packets and then do the requests to an available router? It should work even with those routers can't have openwrt, or can't set the interface to monitor mode.

It seems there's no more "matching voltage since 1/15 21:38 (By the way Impressive work, as always :)

I make weatherstation with sleeping anbd esp12. After sleep, it measure temperature, connect to internet and send data to thingspeak. It takes around 4 seconds - 3 seconds. But definetely, here is much more better. Thanks.

>watch it blink >cuts away from esp to a screengrab >aaaand blink yo...

Very cool. So how long can it last with the 2 coin cells?

Have you considered using ESP-Now? (takes around 200ms with avg. of 80mA, i know this is a bit worse than your solution and requires a second ESP as Gateway, but wouldn't it have been easier? Apart from that the time maybe can be improved?) ESP-Now gets rid of the whole TCP-IP Stack and there are already some library's out there.

Nice vid, as always. I'm hoping to use this to modify N64 and Dreamcast joypads with an ESP module. I already have a UDP joystick app on Android working great for controlling the N64, but I haven't had much luck getting an ESP32 working in an original joypad as yet. I was using the Arduino IDE for the ESP32, and while I could successfully send the request to the joypad (and see the returned data on the o'scope), I could NOT get interrupts to trigger on a pin change? I tried different input pins, made sure I set the pin as an input, and with the proper pull-up resistor, checked the GPIO matrix etc. Nope, no interrupt triggering. lol Probably something glaring that I missed tbh, but I'll try an ESP8266 next. I just need it to request the button / joystick values at say 100 Hz, then send a small packet. UDP seems fine for this so far, but obviously the methods in this vid would massively increase the battery life in the joypad, and it only needs to be picked up by the ESP on the N64, so no router needed. (I could probably just grab the response from the joypad in the main loop, but it made sense to me to use an interrupt routine to measure the bit widths. Also a modest data rate to receive, so I kept missing bit transistions when using a loop, believe it or not. Again, probably my crappy "code". lol) Andreas does some awesome vids on the ESP modules. Definitely a recommended channel for like-minded techie stuff.

How exactly would I send raw data from my ESP8266 to my computer without any network in between? Did you use a network connection in the video or are the packets just floating aimlessly through the air? Thank you

I'm so frustrated I don't have your skills. Cool idea to use the Google doc.

Just curious what are the application of this ?

Hello. Pls tell how to flash your Minimalist Sdk. Thank you

My old 465 scope I use to use at Sony back in 1979

My code is doing the whole wifi connect, bme280/MAX44009 measurement, http post, enter deep sleep stuff around 653ms with arduino! I use fixed wifi ip(with dhcp!), modded the HTTPClient lib to exclude the dns request and use ip to connect to the server. On server side i also close the connection before the db insert. Sometimes the wake time goes up to about 1.5sec for some reason but its much better than the 5-10sec mentioned in the video.

Supercool

Did someone test in this scenario like how many devices can successfully submit the data, for example 100 or more ESP connected to such router and if there was any data loss?

The problem is if you use this approach you will have to change batteries every 1.5 month for just that simple one packet tranfer. Current of deep-sleep = 20uA Current of booting = 30mA Current of loading = 40mA Current of pre execution = 20mA Current of execution = 70mA Current of going in deep-sleep = 10mA Time of deep-sleep = 9.9s Time of booting = 35ms Time of loading = 25ms Time of pre execution = 15ms Time of execution = 20ms Time of going in deep-sleep = 5ms Times of 10 second interval in minute: 60s/10s = 6 times Times of 10 second interval in hour: 6*60 = 360 times Time of deep-sleep per hour: (9.9s*360)/3600s = 0.99h Time of booting per hour: (35ms*360)/3600s = 0.0035h Time of loading per hour: (25ms*360)/3600s = 0.0025h Time of pre execution per hour: (15ms*360)/3600s = 0.0015h Time of execution per hour: (20ms*360)/3600s = 0.0020h Time of going in deep-sleep per hour: (5ms*360)/3600s = 0.0005h Power drained from deep-sleep: (20*10^(-3))mA * 0.99h = 0.0198mAh Power drained from booting: 30mA * 0.0035h = 0.105mAh Power drained from loading: 40mA * 0.0025h = 0.1mAh Power drained from pre execution: 20mA * 0.0015h = 0.03mAh Power drained from execution: 70mA * 0.0020h = 0.14mAh Power drained from going in deep-sleep: 10mA * 0.0005h = 0.005mAh Total power drainage: 0.0198mAh + 0.105mAh + 0.1mAh + 0.03mAh + 0.14mAh + 0.005mAh = 0.3998mAh R2032 capacity = 240mAh Lifespan with 2xR2032 batteries: (2*240mAh/0.3998mAh)/24h = 50 days without counting for external factors which can affect battery life.

what if you didn't need to send your data to the cloud. Is it possible to record the data locally at the router?

Did you tried "Node-RED" and "mqtt" , for making automatic scripts in no time.(Flow Based Programming.)

I wonder how long will it take for IEEE to come up with a WiFi-LE standard.

Fucking genius.

So what do I do if I just want to get notifications from an ESP8266 for over a few years without changing/loading batteries and I want to use my casual AVM Fritzbox router? Means as a noob I am as dumb as I was before watching this video. :/ A sketch or something like this would be very appreciated.

Is there similar approach with ESP32 and router in monitor mode? Anyone can point in the right direction?

Pair it with a op amp and you got yourself a wireless voltage meter!

I'm gonna give a laptop a new life and looking for which OS I should use

Nowadays it wakes up faster - less than a second while opening a TCP control socket and initializing OTA and uses about 16uA in deep sleep.

Hi, I can’t find the tplink code at, github.com/cnlohr/tplink-raw-wifi. Did you remove it?

ESP-NOW

have you (or anybody else) tried to get data back to the ESP8266 with this approach?

Here's the link of Andreas' video: ukposts.info/have/v-deo/gYmmjYqdhmyasKc.html

Charles seriously, how do you manage to do so much stuff? Don't you work, or sleep? Ah, I see. "We live in the future", huh? You build a time machine, didn't you?

It would be great if someone make a version where Instead of router use regular wifi stick in a pc - probably with mqtt support. 😀

Have you run into any WiFi.status() bugs?

me: d cells, many d cells ALL THE D CELLS

What do you mean by "acid tracking"?

hey. could anyone calculate the batterylife if using 3 rechargeable AAs (3.6V, 3900mAh) and transmitting once every hour? I'm too stupid edit: using this exact method of course

So the ESP basically just UDP's the shit out of its packets and hopes that there is an OpenWRT router nearby to look for its packets and recieve them, is that correct? Interesting approach :D

Thanks for this vid. Comes handy, maybe we'd swap stock fw w/ your minEspSDKlib. In docs.google.com/spreadsheets/d/1oay41nWtpytSJAVGPSehMkfOp5g8D0GKNWIzqUzic_0/edit#gid=1212757165 you mention full RF cal wouldn't be feasible on bat. Have you tried adding bigger caps? Here it is possible to use a single 18650 with a big enough cap (here:1500uF) bridging the PWR spike on bootup/RF-cal. The spike has been determined to be at about 370mA. Then another problem keeps creeping up: every few PWR cycles (we do complete ones using a digispark and mosfet circuitry, still using a single 18650 cell) the esp8266 is locking in in a state where it is sucking 370mA continuously. W/o breaker this ruins bat life and probably before that it would be killing the esp8266. We need a complete PWR disconnect for the esp to recover from this kind of hickup. There's also this interesting page: www.ondrovo.com/a/20170205-esp-self-destruct/ . Did you experience this behaviour yet? edit: Using fixed IP makes wake-up/RF-cal and reconnect a 1-3 sec ordeal on PWR on. Better than 6-10secs for DHCP, anyway. Additionally, using DHCP in poor WiFi Rx/Tx risks dropping packets and running into re-Tx w/ timeouts involved. You certainly do not want this on a battery-run project. So, either fixed IP or your approach. Another thing worth mentioning is: espSDK now supports a complete WiFi stack kill button (and by this I don't mean the sleep mode thingy). Just for gaining some extra battery life, this alone might be a great try (i.e. an offline data logger). For projects that needs some kind of radio com it's not that big a deal. ;)

I was working on something similar, and I managed to knock down the wifi connection time to about 0.3 seconds by saving last ip configuration and reconnecting as a 'static' ip (0.3 seconds after startup) This project is perfectly timed for me :D wait, you cheated by hacking the router, never mind :s

So in order to get this working someone has to have a tp-link router and hack it up?? Isn't there any other solution ?!

"Milliamps of power" ??? I thought amps were for current and watts were for power. Precision in language makes one seem more authoritative.

Why cant people say arduino ?

You really need to understand the difference between voltage, current and "power" which you use wrongly EVERY time you say it