I've been using a 12V supply and a 5V linear regulator to power a RGBW+W strip and a pro-mini for over a year now. And I use the 3.3v regulator on the mini to drive the radio. No problems and while a buck converter might be more efficient, there's no significant (or even noticeable) heat on the linear regulator and it's a pretty small amount of power in general.

Mostly power. Check how much power the ESP & wifi use vs NRF & atmel. For non-battery things it's probably not much of an issue though.

Also the mesh network and repeaters, again less of an issue if you have solid wifi coverage everywhere you would want a sensor.

There's also a more robust set of libraries and tools for working with various physical sensors on the arduino / atmel side than there is ported to the esp side.

But if you are just embarking on this now, and none of this items are an issue, there might be some logic in going all ESP/WiFi...

Check the power at the radio and make sure it's getting ~3.3v . Are the nanos 5v or 3.3v?

How do you envision the system working?

Were you just thinking that turning the knob would rotate through a pre-built palette of colors, which would be shown on the knob's RGB first then transferred to the final device or something where you could control the mix of RGB by pushing the encoder's button to switch from controlling R, G, and B until you get the desired color on the indicator LED then maybe a double push to transfer it to the final device?

I don't think either would be too difficult, but given the differing methods for specifying a color to the various RGB devices, I'm not sure how accurate the preview color will be on the knob's LED.

The speaker is probably being sent an AC waveform, and the meter is just averaging it out so you see 4.7v. Likewise there may be some DC power from one side of it to gnd all the time since it's able to energize the Arduino.

Best would be to trace the circuit further back and see if there is something else you can tap into ahead of the speaker and it's driver.

If not you might be able to tie the speaker outputs together through a diode and a small resistor ( say 16Ohm ) and hook that to pin 3 -- or not depending on how it's driven. Or you could build a voltage dividing circuit to keep the AC voltage down to ~2v and use an ADC pin to read the waveform being sent to the speaker.

Probably the Vera, could try power cycling it to see if it speeds things up a bit?

Could also ssh in to your vera and run:

tail -f /var/log/cmh/LuaUPnP.log |grep Arduino

Then push the bell and see how long it takes for it to show up in the log. If it show up there quickly, but you don't see the change on the web or other i/f the slowdown is either the vera itself or the system/device doing the display.

I have a doorbell sensor and a trigger which sends a growl notification using the push plugin. The sensor also flashes a bright LED directly.

I just tried 5 doorbell pushes while holding the laptop and phone. One took 4 seconds to hit the log and another couple to hit the phone. All the rest took only around 1 second to hit the log and another 1-2 for the growl notification to hit the phone.

I am using the ESP8266 gateway, and was using an ethernet gateway before that. I originally used the serial gateway, but that was before I had the doorbell sensor.

Dunno if any of that helps though.

Perhaps re-word the last note from:

A user experienced data transfer compare to USB-to-Serial(ttl)<-->MySensors Gateway connection. especially on OTA firmware update. If you think this will cause a problem for you, an ethernet gateway might be a better alternative.

to:

A user notes that data transfer speeds are comparable to that of a USB-to-Serial(ttl)<-->MySensors Gateway connections. If you think this may be an issue for your application (e.g. OTA firmware updates), an ethernet gateway might be a better alternative.

thanx,
JpS

I'd been using the ESP gateway for several weeks, but recently was having trouble adding a new sensor I am developing.

Since the copy of dev I built it with was now out of date, I went ahead and updated... still wasn't working right so while I had the serial link to it I checked and noticed it was crashing and cycling. It would work for hours, until I fired up my new node, then crash about immediately. A few dozen debug(PSTR())'s later, I found a cause.

In core/MySensorsCore.cpp the result of _msg.getConfig() is dereferenced in a comparison, and it's possible it could return NULL ( like when it's a new node ). I put a check around that and it's not crashing anymore. Which is more than what I can say for my new sensor node....

I have not tested this change with an AVR.

I submitted PR 264.

I haven't actually yet tried to run one at lower voltages, but I recall from looking into it a bit that you can't run at 8Mhz below 2.4v. I think 4Mhz was the max at 1.8 (but that's from memory so might check) -- but to do that would require a new crystal. Hence why most often people will jump down to 1Mhz using the internal osc. But at that speed you probably can't do 57600, and will probably want to go lower, 19200 or 9600 or something.

Google should find you some more detailed info, or even just a search on this forum for low power and 1Mhz or similar.

I'll be there at 6. I'll have on jeans and a black shirt with a tiny apple logo on it and maybe a greyish jacket. I'll leave my conference badge on my belt as well for easy identification :)

@mfalkvidd Thanks for all the tips and suggestions. I'm quite familiar with procrastination myself, and am putting the finishing touches on the material for my talk just now.

I have been told there is some conference-related event every evening and but it's not yet clear if they are strict start-stop things or just drop-in ones. Regardless, the one on wednesday doesn't start until 19:00, so I can be at the Steampunk bar at 18:00. I will PM you my contact info, and if you send me yours I can let you know if there is any last minute change of plans (and visa vera).

-JpS

I got this working on an 0.9 version of the nodeMCU ( the Yellow board ). It has some code upload issues, which are slightly worse on OS X than Linux, but usually a couple of tries with holding down the button as @hek mentioned usually does the trick. I do not have any TCP/IP clients pointed at it currently, but I also get lots of watch dog timer resets, which seem somewhat random, as it will run and log messages for somewhere between 15 mins and 10 hours.

It is just being powered by USB and I have not swapped NRF modules yet, so one of those may likely be the culprit, but just wanted to chime in with a (mostly) successful setup.

thanx

I'm visiting Sweden next week. Specifically Göteborg, for macsysadmin.se. Since it seemed like a fair number of readers of the forum here are from (somewhere in) Sweden, I thought I might ask for any recommendations of things to see, things to do / not do, etc.

And if anyone is in or near center city Göteborg and wanted to meet up at some time and discuss projects, etc. that'd be great too!

I'll be there from mid-afternoon sunday the 27th thru the following friday.

thanx!

The 2550 is pretty old, dates back to the 90s I think. You could pick up a newer panel and IT-100 for not much $ and fit it in place, possibly using the same keypads. The serial protocol is semi-well documented. I'm using an 1832 and IT-100 though the VERA ( plug-in exists ), and it works reasonably well. I do not know if they use the same 4 wire keypad and same protocol, but I think that has been decoded and hacked as well. That all said it might not be a huge amount more work to just try to use the existing sensors with something home-grown and either get rid of the keypads entirely or roll your own there as well.

Maybe you could post your entire sketch here?

Sounds like you might want to use an array. Something like :

 int myVars[] = { 123,456,789 };
 
 for (int i = 0; i < (sizeof(myVars)/sizeof(int)) ; i++) {
     Serial.println(myVars[i]);
 }

should work, though depending on exactly what you want to accomplish, and what controller you are using, etc. there might be some better approach.

I'm using this as a gateway w/ 1.5 also with no issue. I do have the LEDs hooked up as well, using an oversized 'recovered' power supply and the PA+LNA radio with no problems. All I need is a nice case...

I do pretty much this right now with a Yale 'Real Livinig' z-wave keypad lock. It sends an even when a pin code is entered, an 'incorrect pin' code if an invalid one and a 'correct' pin code with the name of the code when a valid one is entered. I can then make as many scenes as I want using the appropriate trigger. Right now I have a couple of lights come on when I enter mine. I have a special one which opens the roll up garage door, and ones for other people send me an alert, etc.

I think one could do probably craft something cheaper that didn't do the door unlocking, but making it weatherproof and damage resistant like the Yale might be more of a challenge.

@hek I have seen it happen once before with them -- wait a few hours, or at most a day and it'll be corrected if history repeats...

I just tried to order something from them and for some reason DHL is now the only available ( US ) shipping option, and it's quite expensive. The sensorbenders I ordered last month were ~$6 shipping, the other components I looked at today via DHL (similar size) were $46, so might be worth holding off a bit until they restore the air-mail option.