Best posts made by jirm

@mtiutiu

I think you have the solution very close. I´m sure It is simply a matter of "fine-tuning" the Livolo's power circuitry so that they can provide enough energy (much more than designed) for the extra circuitry, something for which they are not prepared at all, but should not be too difficult to achieve if necessary. Considerations.

1.- The power circuit depends on the "consumption" that can be achieved over the load connected to the switch, so it is necessary to ensure that the load has enough "power draw" to allow pass through at least 100-150mA from AC which should be achieved with a load about 40/60W.
It is also something that can be compensated by adding a bypass capacitor (parallel to the load) of between 100 and 470 nf (of type X2 and 250V better 400 V to go safer), but all this is matter the try and error to achieve the target.

2.- It must be able to avoid the Livolo power circuit so that additional power can be consumed without being affected by the circuit status of the switch, either in standby or active mode. In theory that can be done through the bridges @DJONvl and @Tigroenot indicates in the post you mentions https://forum.mysensors.org/topic/2775/livolo-glass-panel-touch-light-wall-switch-Arduino-433mhz/63 , and whit that should be possible to have the necessary power and several of them seems have confirmed that it is possible.

Perhaps the problem arises to be able to correctly identify which are the correct modifications to be made in each Livolo switch since there are different types of circuits according to the world region to which they are addressed and different revisions of printed circuit so it is critical and not very obvious as Identify the jumpers and connections to be modified and how the power bypass in each case must be performed in order to extract the current to feed the additional circuit.

I think the latter is the most complicated and difficult to perform correctly, but nothing compared to all the work you have already done, so I'm sure you've almost got it.

If I could help you, I suggest some posts (unfortunately only spoken in Russian ;-), o. c. you need use google translator to try to understand) where you will see in the lower page intersting people discussions that are given some patterns and ideas to modify the circuit and that seems have managed to extract required current to power additional circuits.

Http://mysku.ru/blog/aliexpress/12956.html

I also indicated the analysis (I think original) of a person who performed reverse engineering to extract the schematic circuit from the assembled Livolo switch and where it discusses some aspects of its operation that can help.

Http://we.easyelectronics.ru/Shematech/preparirovanie-sensornogo-vyklyuchatelya-livolo.html

Please do not be discouraged and persevere a little more than you already have it!

Greetings.

@Oli
Good !
I like when people can make something work that nobody even knows what it is

For me it usually happens just the opposite I am not able to make work even the things I know perfectly well ... maybe it will be a matter of "bad karma"

@mtiutiu

Hi @mtiutiu, You are the genius !!!.
I'm really happy to hear that you've finally achieved the desired results. Believe me that joy is mine if on anything but minimal I have contributed to you can achieve the desired goal.
I firmly believe that your development is the one that with difference and from my modest point of view has the key to success both because of its general approach and the fabulous development that I see you are performing in the accurate design of the additional circuit and I am sure that this design will be the future reference to give the Livolo's the possibility about true bidirectional RF control.

Regarding the circuit that Livolo sells to avoid the problems of switch flickering by low power loads (less than 3W) this is a circuit with active components and is not a circuit designed to provide more power but to guarantee the switch of a " minimum " but regulated and constant current and providing it with better immunity because the current fluctuations caused by some electronic loads (led and similar) when they are at rest.

So it is not at all something that can be compared to the function of the capacitor in parallel to the load that provides an absolute increase of the current flow through the load circuit (and therefore of the switch) due to its behavior resistive in AC (more correctly its impedance), and this last is the characteristic that here we are looking for.

Just point out that a couple of details regarding the use of a capacitor in this way, and is that the consumption will be constant activated the load or not although we speak of an increase in very low consumption (100ma @ 220V = 2.5W aprox) but should not be neglected if we intend to "modify" by this way a relatively high number of loads and in this cases I think at least imposes a previous calculation of the constant power that we will increased because of add a "pile" of parallel load caps.
Also not should we lose on sight the fact that we will introduce a reactive component (reactive power) that can affect the character of our electrical installation and that as I say if it is used generically in many loads it is necessary to calculate its effect in our installation to avoid some displeasure with our electric company and their invoices.

Finally let's do some simple math to calculate a capacitor value that might be appropriate.
Thus, the capacitive reactance is expressed as Xc = 1 / (w * C) where w is omega (equals 2 * pi * f, where f is the frequency) and C is the capacity in farads.
So if we wanted to have 20ma at 220V we would need a resistance of 4K (obviously I not do here this calculation of the simple ohm law) and therefore in my country that we have 50Hz frequency the capacity we would need would be C = 1 / (w * Xc) => C = 1 / (314 * 4000 ) = 729nF

Summing up the capacitor you have tested on 470nf gives you a constant current of about 15ma when the circuit is in standby (load off), and don't forget we have a capacitor so we can achieve much more current from his reserve (this is for what was designed) and we can manage the current peak draw when RF transceiver are full active working on sending (Hope RFM69W/HW typically have current peaks over 30-40 ma on 100-200ms duration on sending cicles at full RF power) and seeing all seams should be enough to power any RF transceiver and a low power mcu if its consumption is well controlled by its proper use.

With this capacitor we are increasing the consumption of the load circuit in only about 0.5W, but let's not forget that this will be constant consumption in 24/365 hours / days a year.

So I keep close tuned with your evolution here.

Best regards

@DJONvl

Yesss @DJONvl . Thats is !!!! .
You got it !!!

Please can you post all the details about how you are do it ??

How you are managed to achieve enough power from Livolo ??

How you wired the esp8266 and where on the livolo switch ??

You can post the schematic ??

Please, all details you can...

Regards

@aand.
No dubt that NRF24 is detected and working on Armbian, but seems not properly detected on the build module process for the gateway installation process made for Raspbian. Sure because OS diferences from Raspbian and Armbian, raise this issue.
Why not directly use Raspbian on OPI to avoid that and sure other issues that will by found because all software are mainly build for Raspberry (Raspbian OS) and that is the only well tested?

I see Raspbian is available and updated for OPI trough his manufacturer.

http://www.orangepi.org/downloadresources/orangepipc/oragepipc_e930546e866b23585721e5d2a6.html

Regards

@marceloaqno
@mfalkvidd @Tag @pansen @Reza @mihai.aldea @hausinger

Wooow woow woow !!!

Only few days "off" and the problem to make work Mysensors on OPI are solved.

So many many thanks to all people here that have been involved to make that OPI good little boards can work with Mysensors.

This is the proof that when good people join can reach the most difficult achievements and all humanity can benefit.

Awesome work !

Regards

@mtiutiu

Okay. According to your observations 200mw (peak) is required to feed the additional RF plate and to that must be added the self-consumption of the Livolo's power circuit.

I believe that in this case, it is not so important to calculate the peak power because given the dynamic of our operation (the peak power transmissions usually need a few ms), I'm certain that this peak power can be easily provided by the own current reserve stored in the capacitor itself, and because that I think is necessary focus tries to guarantee the stable supply of current for the "normal" operation of both circuits (additional plate and livolo), and that requires at least 15ma (I think that is the min) and much better if we can guarantee about 20-30ma.

In the next link you can see a fairly exhaustive analysis of current variations vs RF output power in a typical RFM69HW operation, which shows that the average current is around 20-30ma and the peaks can reach 80ma:

Https://www.andrehessling.de/2015/02/07/figuring-out-the-power-level-settings-of-hoperfs-rfm69-hwhcw-modules/

So trying to make a global vision of the power needs for that project in general and taking into account that usually we will always ignore the true capacity to generate current of the several type of loads that can be connected, the different topology of housing wiring (self-capacity, spur, etc ...), the huge variations in impedance of the loads according to their type, their dynamics of operation, etc. I think is much more reasonable and closer to the real needs use capacitors of at least 470nf min (although I would opt for 680nf or maybe 1uf), to guarantee that there will always be a enough constant current supply capacity really closer or exceeds demanded in any circumstances.
Therefore I would not consider so much in calculations the capacity of the own loads for the power supply and only would calculate the capacity of supply by using the correct value of the capacitor that in any case will have to be installed.

I think it is very important to guarantee the stability of the circuit operation (speaking in the long term) given the "infinite" possibilities of characteristics so variable that can be found in the installations of any house.

Collaterall efffect: When installing this capacitor we removed the load limitation of > 3w for Livolo's Switch, so generally now they can be used "independently" of the power and type of the load.

We have just solved a serious problem for Livolo manufacturers

Regards.

@aand

Hi again!

I spend some time seeing and reading some of the files from the mysensors github build and I see some things that can help to understand why the error raised to @aand trying to compile on OPI.

First say that I´m not a coding expert, so please be nice with my if I´m wrong and say something stupid

I see in the .configure file that the machine detection have H3 Soc and this are fine detected , but then I see that selecting type and CPU flags seems not properly selected to match OPI specs , because on OPI soc H3 have a cpu armv7 (not armv8) and is a A20 and not A53.
So maybe that cpu "selection" is not correct for H3 soc , and is so easy try if anyone that have OPI and have time can change only that values in the config file and try to compile.
All the values on the A20 line seems correct for the OPI H3 soc, so only needed copy them from the A20 line to the H3 line, and give a try.

File Configure

.......

function gcc_cpu_flags {
local soc=$1
case $soc in
BCM2835)
flags="-march=armv6zk -mtune=arm1176jzf-s -mfpu=vfp -mfloat-abi=hard"
;;
BCM2836)
flags="-march=armv7-a -mtune=cortex-a7 -mfpu=neon-vfpv4 -mfloat-abi=hard"
;;
AM33XX)
flags="-march=armv7-a -mtune=cortex-a8 -mfpu=neon -mfloat-abi=hard"
;;
A10)
flags="-march=armv7-a -mtune=cortex-a8 -mfpu=neon -mfloat-abi=hard"
;;
A13)
flags="-march=armv7-a -mtune=cortex-a8 -mfpu=neon -mfloat-abi=hard"
;;
A20)
flags="-march=armv7-a -mtune=cortex-a7 -mfpu=neon-vfpv4 -mfloat-abi=hard"
;;
H3)
flags="-march=armv8-a -mtune=cortex-a53 -mfpu=neon-vfpv4 -mfloat-abi=hard"
;;
*)

@mtiutiu

Hi all

Can I suggest some improvements for that project?

For sure I agree with @tonnerre33 about make a version board for only one gang switch.
I buy regularly (once a month or so) some parts from livolo and from last 4 or 6 months I see that the switch plate boards (for the EU version) are the same on hardware specs at least from 6 or 9 month ago.
But we can expect that in near future (maybe few months) Livolo manufacturer make some changes and updates on his designs, because they regulary are doing that in past. So we need keep prepared for that and for make the propper updates to this project to mantain it working with the next version Livolo switchs.

1. -One gang switch I think is most common switch people have installed or at least I think is needed too with this two gang project version.

2.- Maybe will be better use (or make another plate board version) for the RFM69HW (high power), because the pinount on HW not match with RFM69CW you use, and HW type have same working specs but with the plus that HW type have high power possibilities and is most easy to find and both have similar price.

3.- How can be little better documented all the changes (wiring) we need do on the power relay plate board?
At least for me is hard to do correct wiring only seeing the photo and is easy to do some mistake trying to solder the wiring in the correct pins and places.
So I suggest trying to do some more work on that and maybe include some scheme and plan and take all best pic from each wiring bridge is needed to do that we can see without doubts the correct place to solder it.

PD: @tonnerre33 I think one or two way switch function not be affected with this project plate board, because that function only differ on the pic (MCU) Livolo switch firmware and litles changes on power relay plate board to wire the additional pic pin output two way function to the COM connector.
Nothing of that function should be affected by anything that this plate modifies on the Livolo switch.

My best wishes !