Remote controlled light switches and power sockets

Jee Labs Talk » Success stories

Remote controlled light switches and power sockets

(120 posts) (19 voices)

Tags:

No tags yet.

shapeshifter

Allereerst een gelukkig nieuwjaar aan alle Nederlandssprekenden en op JCs gezondheid, dat hij nog maar veel gave dingen mag verzinnen.
Having said this, a friend and I would like to make remote controlled light switches and power sockets. We had been looking on the web for a while when I stumbled onto JeeLabs. We think JCs stuff will make our lives a little easier.
We already had some circuits going for the switching part (see pictures). The first uses just a relay and the second uses a solid-state-relay constructed out of an optocoupler and a triac (DTR is the data terminal ready pin of our serial connector).
(Yes I know, wires out of the socket is sort of dangerous, I'm being very careful)
The third schematic shows globally how we want to use the JeeNode to make this switch wireless. The left-hand part is a zener diode hack to supply the JeeNode from the mains and the right-hand part is the solid-state-relay. We are also debating the need of a manual switch, which could be fitted on the JeeNode (so the system is also aware of analog switching).
This post is partly intended to keep you apprised and partly to hear what you think (also if JC is maybe working on this already, some posts on his blog suggested that he was).

Posted 03/01/2011 18:34:26
jcw

Cool! (and thanks)
Be sure to work out how much current you'll need while driving the opto-coupler and/or the RFM12B. The direct-coupled power supply you show may need some hefty resistors (or a big "X2" cap) to deliver that sort of current. A supercap and proper timing might help.
I'll definitely go into these types of circuits on the weblog. No need to wait or worry though - this is neither a race nor a "zero-sum game", more a way to share and learn from each other, as far as I'm concerned!

Posted 03/01/2011 19:20:35
jeroen

sort of dangerous ? the wires are isolated, so I don't see the problem ;-) (i am working on a 230 volt solution to, but this is exactly the reason why it is postponed)

Posted 03/01/2011 19:39:45
tankslappa

I'm not sure what the insulation voltage for those wires is rated at. Remember, just because it's got a plastic covering, it doesn't mean it's rated for 230volts.
I'm working on one too ;-)
So we're all waiting for the first person to get curly hair, then we learn from them... No point waiting for me, there's no hair on my head to go curly!

Posted 03/01/2011 19:48:10
cowboy

Cool!
My new house will be built in the next few months and I was already looking into some Domotica system to achieve remote switching... Ofcourse I was already thinking about z-wave devices to achieve this but also the Jeenode stuff...
I wonder how small this device can be made so it fits the space behind the wall sockets... Also I wonder if some kind of power metering can be done like the Plugwise circles...
I'll keep watching this space....

Posted 03/01/2011 19:48:55
tankslappa

I'm building one into a 4 way UK 13A extension lead. JeeNode, 4 triacs and various other bits. It's a bit of a tight squeeze and I haven't even figured out the heat-sinks for the triacs yet!

Posted 03/01/2011 19:52:03
cowboy

I will tell our electrician to make sure he uses extra-deep wallplastics that allows me to get some additional space behind the sockets......

Posted 03/01/2011 20:04:44
jeroen

Until there is a safe solution I use kaku devices (if possible)

Posted 03/01/2011 20:18:49
mqrius

Hiya, Shapeshifter's friend here.
First, a small correction; in the top drawing of the relay, the emitter of the transistor actually goes to the ground of the serial port, so the switching part of the relay is isolated from the switched part. So, not connected to the other side of the mains power, that would be a Bad Idea. :)
Jean-Claude: "Be sure to work out how much current you'll need while driving the opto-coupler and/or the RFM12B." The optocoupler is a nice MOC3062 triac driver with zero crossing detector, at 10 mA maximum trigger current. A quick look at the RFM12B specs state a maximum required current while sending would be 25 mA. I'm not sure how much a JeeNode uses in general? But yes, that's getting a bit high, but should be attainable with a transformerless power supply, I believe.
Cowboy: "Also I wonder if some kind of power metering can be done like the Plugwise circles" I believe this should be possible. A large resistor voltage divider to measure the voltage on an analog input of the JeeNode, and a small specialized resistor (~0.1 Ohm) in series with the load to measure the current with another analog input should do the trick. If the JeeNode is powered with the transformerless power supply, it should be fine to wire everything directly to the JeeNode (if you wire it correctly), but if not, linear optocouplers could be used, but I'm not sure if those are cheaply available...
On an unrelated note; my test rig is built into a power block:
Image link
It looks like a mess, because this is the third circuit I've put in there :P Also a video of my light blinking, controlled by the computer.
Video link
I use this, together with some noise, to wake me up in the morning :)

Posted 03/01/2011 20:21:58
shapeshifter

The JeeNode is already pretty small (it could be smaller if you detach the RFM12B and place it perpendicularly, but that's cheating :P).
In principle power metering can be done, but the question is if it's practical. For "real power" you need to measure the voltage (which you can do directly on the JeeNode or via the analog plug, as JC has shown us already: http://jeelabs.org/2010/02/27/voltmeter-with-gui/). Of course, you can't just measure voltage levels up to 230V, so you need a kind of voltage divider to bring it down to about 3V (corresponding with the JeeNode inputs). You also need to measure the current, which is a little bit trickier. You could for example use a small current-sensing resistor, say 0.2 ?. This would create a small voltage difference across the resistor and since we know the voltage drop and the resistor value we can calculate the current. The resistance is very small, so very little power is dissipating through it, this does seep some power from your lines though (about 0.6 W at 1.7A).
Those Plugwise circles are pretty cool, I wouldn't buy them though, because the business model of the company (which is Dutch, shame on us) dictates that they harvest as much data about your power usage as they can (which is why I would rather make some myself)...
Lastly, I intend to embed the switches into my wall in such a way that no switch is visible... at all (the best part of this will be the guests looking for the switches, which they can't find and me operating them like magic). For this I want to use the proximity plug. Can someone tell me if the capacitative switches are included in the kit (and if not, where I can get them) and if they are sensitive enough to detect say, a light press on a piece of plastic / paper / whatever glued to the sensor (my light switch). Thanks in advance.
//EDIT: Hi friend :P Should have refreshed before posting, thanks for correcting my interpretation of the scheme, in reality I have it hooked up as you suggested of course. Well, the bit about power metering should be extra clear now :P

Posted 03/01/2011 20:25:46
louis

Interesting topic, but this makes me think I am missing something (I am a software guy learning about atoms). I was planning to use a free port on an existing Jeenode with roomboard to switch on a light for some time when motion is detected in the dark (the light is 230Vac 40W). I intend to us a relay plug for this purpose, I am currently working on the roomboard code extension (which shall support a local switch input and remote override/control of the light as well). Would a relay plug do the job, or are the options discussed in the above posts better/smarter/safer?

Posted 08/01/2011 20:08:39
shapeshifter

On the contrary, the relay plug would actually be the first approach to use when switching mains with a JeeNode. This is reflected in the first sketch that I showed (Note: as Mqrius mentioned, this sketch is not quite correct and the signal part in my case the DTR pin, in your case the JeeNode
However, there are some concerns with using relays. Generally, they're big (which is not convenient when fitting them in wall sockets) and they're noisy (they make a clicking sound when they are switched). Therefore Mqrius and I wanted to use solid state relays (which is the optocoupler and the triac in the second sketch). Add a transformerless power supply to feed the JeeNode and you've got the third sketch.
Currently we are working on adding power metering support, but we have exams in two weeks (yes, we're still Master's students), so this has priority. This is a quick sketch of our work in progress (it probably contains some mistakes, but the idea is there):
Note that we probably have to replace the resistive transformerless power supply with a capacitative one to accommodate the power requirements of the JeeNode (and especially the transceiver). We also need to calculate all the values for the resistors and capacitors. We also added a push button with which you can manually switch the lights (this is still routed through the JeeNode though, so it's still tracked by the system).
You can already see that power metering involves more components, there is a voltage divider for measuring the voltage across the load, but since the AIO of the JeeNode can only measure positive voltages we have to use an offset circuit off the +3V line of the JeeNode to correct for this. The same is true for the current measuring resistor. Long story short, it will take some more time. We will probably have more in about three weeks... Until then, good luck with your project.

Posted 08/01/2011 21:36:11
bgrr

Hmm powermetering. Would like to do that with a jeenode :-) Already logging the kwhm meter with a K8055. but logging stuff like pc's servers and other individual stuff would be great.
What kind of components are used in cheap "wattmeters" like the Cresta RCE1106 ? Which should be pretty accurate for its price. They can be bought for 15 euro or so. So looks like the components aren't that expensive and complicated. Link to cresta ( http://www.cresta.nl/index.php?Itemid=2&option=comzoo&view=item&categoryid=44&itemid=473&lang=en )
Is there a way to hookup the watt and voltage part of a cresta to a jeenode ? Or something else that generates a pulse per watt or so.
Found this chip on the www a while ago but my electronic knowledge is very low so i haven't got a clue is if it is useable with a jeenode and how to build it :-)
http://www.analog.com/en/analog-to-digital-converters/energy-measurement/ade7566/products/product.html

Posted 08/01/2011 23:00:30
shapeshifter

To be honest, I'm not familiar with the insides of the power meters you mentioned. I figure that they will either use the scheme I've outlined, the chip you mentioned or something resembling a CT sensor (like what is done at: http://openenergymonitor.org/emon/node/58).
A quick look in the datasheet of the chip you mentioned tells me that you cannot hook that one up to the mains directly, so you'll have to bring the voltage from the mains down to the input of the chip (essentially doing the same thing as we are, only more accurate), this would explain the cheap units (resistors and capacitors are generally very cheap).
I don't think it's very practical to try to hook up an existing device to the JeeNode unless it has some function which transmits the acquired data wirelessly on a frequency that the JeeNode supports (actually this kind of hooking up is frequently used on this website), but then again, I haven't got such devices, so I can't open them up to look into it (neither am I sure I will grasp it when I do)...

Posted 09/01/2011 00:39:53
rikkertkoppes

I am currently working on controlling light through a jee node. Idea is to create a wall switch with four soft buttons and a relay plug attached. Since everything doesn't naturally fit into a wall box, I need to adjust the form factor a bit. Most ports are removed, switches are added.
I currently finished the first half of a broadboard prototype. Relays and wireless still needs to be added, but I a waiting for smaller print connectors (without detachable part) to replace the ones delivered with the relay plug.
Nest step is to create a custom board that will contain everything and will fit into the wall.
In the end, I will have a remote switchable relay and remote buttons that can be connected at will.
by the way, here is the schematic I used thus far: pdf
and the led buttons test sketch: pde

Posted 15/01/2011 09:22:44
smihai

nice! just add a battery and you've got a wireless switch (i would also add in a temperature probe, just in case you want to get crazy in the future ;)) )

Posted 15/01/2011 09:35:31
jcw

Great stuff... do you want me to make a board for it? I haven't looked into the exact dimension needed for this, and don't even know whether there is a standard across countries.
One important aspect though: have you thought about the safety implications? What happens if there is a fire and one of the insurance sleuths finds one or more of these units in your electrical circuitry?

Posted 15/01/2011 11:29:00
rikkertkoppes

I'd really like to cooperate on the board, having it made along with ( lots) of jee stuff, will probably cut costs.
I am still working on a setup for the board, I will probably go for a layered approach. One squarish board with all low voltage parts on it. Then an insulation layer of foam board, then the relay plug, mounted back to back.
Maybe, the relay bit can use some form factor redesign as well, to fit a little better, a fuse may be added as well
Everything will be surrounded by a 5cm PVC plumbing end cap, which will fit just fine in a standard wall box
As for safety/insurance: some sparks, arcing may occur in the relays, the rest of the circuit will all be encased. I am really confident about the standard safety measures (fuse, differential switch).
In case of a fire: I hope that will just destroy the thing then :)

Posted 15/01/2011 11:57:19
jcw

Ok. What's your plan for the power supply, if I may ask?

Posted 15/01/2011 12:13:26
mqrius

Looks nice :) How do you fit 4 switches and also some relays on there? Doesn't a JeeNode have just 4 ports?
Jean-Claude: "What's your plan for the power supply, if I may ask?" I'm wondering as well. Putting a transformer in is pretty bulky, making a transformerless power supply is hard to get enough current, and also it has a lot of static dissipation, and using a battery just feels wrong with the full main line power at your disposal...

Posted 15/01/2011 12:35:15
shapeshifter

Very nice! From the picture I take it that you took the JeeNode concept, but aren't using the PCB? This certainly allows for some more convenient sizing...
A few questions though, why would you use the relay plug (concept) when you can also use the solid state relay I mentioned earlier, this will also decrease the size of the circuit (and maybe prevent sparking and reduce costs)? Secondly, I figure that you use 4 switches for stability and maybe sensitivity, rather than having 4 actual switches right? In principle you could bring this down to 1 in the middle and make a frame for stability instead; this will save you space as well as costs...
JC is right to mention the safety concerns, not that I had implemented them though, so I won't go into that :P For power you could use batteries or some sort of transformer(less) power supply. It would be really great to make a board out of it!
PS s.mihai, if you are content with a rough temperature measurement, there is actually a hidden temperature sensor build into the ATMega chip (see: http://talk.jeelabs.net/topic/701) PPS Soldered my first JeeNode yesterday, it seems to be working :)
@Mqrius: we have to stop posting right after each other :P

Posted 15/01/2011 12:42:05
rikkertkoppes

Power supply is a bit of a headache. Maybe a battery, maybe I'll add some external power. I've got hollow walls, so it's fairly easy to add some low power wiring. However, this may not be a solution if others want to use it as well.
As for the ports, I'm using 2 ports for 4 buttons and LEDs (same trick as the blink plug, see docs and my sketch and schema). Furthermore, I'll be using one port for one or two relays. So one port to spare. I thought of using 6 switches, but the pcb will be pretty crowded in that case. Might be solved by moving components to the other side as smd. Not quite sure about that one though
The switches are really four separate switches, which allows for switching just about anything, including the relay. I imagine having one of those for switching off every other light in the house and have it near the from door. The key is that the switches and relays are only coupled through software, which allows to switch everything
These are some images for the enclosure

Posted 15/01/2011 12:47:56
rikkertkoppes

@shapeshifter: your power from mains and triac solution also seems nice. I have too little experience with those things to judge or design such things, but maybe we could adjust my scheme and come to a useful design in the end. I'll be happy to incorporate it.

Posted 15/01/2011 18:51:47
shapeshifter

Credit where credit is due; it was actually Mqrius' idea, but we're working on it together :P I'm sure we're all working towards a common goal here and there is bound to be an optimal solution, so yeah, let's just do the best we can...
As I mentioned before Mqrius and I are having some exams in the coming 2 weeks, so information from us maybe somewhat concise and thin spread. On the other hand, we are infamous procrastinators, so who knows :P
The power bit is not completely solved yet; the transceiver needs quite a bit of current, so either the capacitor or resistor has to dissipate a lot of power (which means more expensive components which meet the high requirements). Maybe we can optimize it a little bit by using a bridge rectifier (which boosts the current), but still... Luckily, the triac solution works perfectly well, so we can use that for sure.
Somewhere along the line we intend to do some SPICE simulations to figure out how the circuit behaves under different circumstances and then we can build the first prototypes...

Posted 15/01/2011 19:18:22
rikkertkoppes

Just found a nice circuit for a transformerless power supply that is decently isolated and spike protected: http://www.brighthub.com/engineering/electrical/articles/77929.aspx

Posted 15/01/2011 20:46:22
rikkertkoppes

This is the general idea for the housing
from left to right: - bezel. - pcb with buttons and jee node stuff. - foam board (or similar) insulation and support. - relay / triac board, 230 v part, also with print screw connectors. - plumbing 5cm end cap, with holes for 230 access. - standard wall box.

Posted 16/01/2011 10:22:12
mqrius

A small note about the transformerless power supply; They say it's safe, but I'm not convinced. The JeeNode and everything will still be at roughly half of your mains power. It's true that if you touch it, the current always has to go through the capacitor, but I'm not sure how much current is needed to hurt you...

Posted 16/01/2011 14:39:22
tankslappa

I've been watching this thread with some interest, the transformer less power supply is the route I am going too, although I'm using triacs for switching (I have dimming in mind). Interesting to see the different designs in that link from rikkertkoppes.
Mqruis, whilst I appreciate that it's still not 100% safe for small children to play with, your whole project is about switching mains AC on and off, so even if you made the PSU with a transformer in the traditional way, you'll still have "real" AC on the relays which you could poke your fingers into! You can't avoid it short of wiring your house for 12v lighting :o)

Posted 16/01/2011 18:36:16
mqrius

tankslappa: Of course, definitely agreed. I just think that claiming something is safe when it's not might prove rather dangerous.
Note: I'm not entirely sure if it is really safe or not. Just not touching it will do fine though.
Also, you say you want to use triacs as well. I'm wondering; have you looked at cooling? I've got TIC206M's here, going up to 4 Amperes, but I think they need cooling ribs to operate at that capacity, no?

Posted 16/01/2011 22:37:33
gjniewenhuijse

Nice idea, next step is to make it fitting and working with default switches from example jung and gira

Posted 17/01/2011 16:19:15
tankslappa

@Mqrius - Quite correct, a heat sink is required. I'm going to fuse mine down to well below the capabilities of the triacs (8A, old stock, going cheap... so bought 20!).
I'm not building mine into a standard single socket, although if I did, my standard would have an advantage... UK surface boxes look a lot bigger than the round one in the pictures above that you continental types get. ;-)
I'm building mine into a 4 way extension lead. Like this http://www.lambda-tek.com/componentshop/index.pl?origin=gbase17.5&prodID=B472070. I'm looking at the surge protected model, not due to the surge protection, but because it means the strip is longer, so room for the JeeNode etc. The surge protection stuff will probably have to come out!
That was I can turn the TV, stereo and everything off completely by remote control, and dim the lamp in the corner - Or maybe just blow the fuse when it all goes wrong!

Posted 17/01/2011 17:57:32
rikkertkoppes

I just added wireless to my breadboard prototype:

The back really is a spaghetti, but the whole placement of components is where I like them to be on the final board
Small setback with regard to the relay part: I got the wrong print connectors from Conrad. They fit on the relay board (to replace jc's detachable ones to save some space), but they are not beefy enough to hold a 2.5mm^2 cable. But on the other hand, the relay bit may just be replaced by triacs anyway.

Posted 18/01/2011 19:02:19
tankslappa

@rikkertkoppes Very nicely done... and I love spaghetti :-)

Posted 19/01/2011 04:28:23
rikkertkoppes

This is my first proposal for the 230 part of the circuit, in the form of a plug. Based on Mqrius' triac part and the transformerless circuit I posted earlier. With the addition of a fuse.
(edit: flaw in the sketch, photoshopped it a bit)
Any comments?

Posted 19/01/2011 07:52:14
mqrius

The transformerless PS is fine, but your connection from the optocoupler to the JeeNode is wrong. You reversed the IO connection and the ground connection; this way no current will flow through the internal diode of the optocoupler. Just reversing them will do the trick.
Have you done any calculations about the power dissipated in the capacitors? You're using capacitors, so you'll get maximum current when the voltage is zero, in short:
Voltage: sqrt(2)•230•sin(pi•100•t)
Current: sqrt(2)•X•cos(pi•100•t)
X here is the current you pull through the capacitors.
Multiplying and integrating:
2•230•X•(1/2)•sin(2•pi•100•t)
=230•X•sin(pi•200•t)
P = int 230•X•sin(pi•200•t)/(1/50) from t=0 to t=1/50
P = 230•X•2/pi
Let's say you draw 50 mA (that's some 30 for the JeeNode, and some for the optocoupler, and of course you need enough to maintain 3V at all times), that means
P = 230•(50/1000)•2/pi = 7 W
That's quite a lot. Divided over your two capacitors it's 3.5 W per capacitor.
PS: Your prototype looks pretty slick :)

Posted 19/01/2011 09:50:00
rikkertkoppes

I understand your calculation but not the reason for it, what does it imply for the caps I need to buy?
From the site I mentioned the following list is used for the power supply:
Resistors: 1M Capacitors: 1uF, 250V ppc Elco: 100uF, 25V Zener: 5V, 1W Diode: 1N4007
I recon a fuse of about 1A will be just fine, no need to go any higher for just a few lights
This higher power stuff is a bit out of my league, so really I am googling and guessing a bit. I really want to be sure before building the part, that's where I need you guys :)

Posted 19/01/2011 10:22:08
mqrius

Well, it basically means your capacitors will get warm. Do you know how standard resistors can only dissipate like 0.5W? Capacitors have a limit too, although it's harder to find. A physically larger capacitor can probably dissipate more power/heat, but if you can find a datasheet or something that mentions that, that's even better.
I don't have a lot of experience with the high power stuff either, nor with transformerless PSs. I guess the only way to know for sure is to try building it, and seeing how much current you can source out of it without losing too much voltage. If it's not enough, you could double the capacitors or something.
The easiest way is of course to grab a small transformer somewhere, but that's expensive and takes up space.

Posted 19/01/2011 10:43:28
bgreene

After looking at your circuit, my experience is that ssrs (triacs with snubber components) can have enough leakage current to cause problems with low power consumption items which have switched mode power supplies in them (such as low wattage hi-eff lamp bulbs or power supplies for led lights). in the relay off state, the items tend to flicker as the smps keeps rebooting. Also, if the load is an old-style filament bulb, the short-circuit which occurs when the bulb blows can take out the ssr. So while i'd very much like to have ssrs everywhere, i've elected to use good old fashioned mechanical relays.

Posted 19/01/2011 14:42:39
rikkertkoppes

Capacitors that dissipate? I always thought that was negligible, since their impedance is nearly imaginary. I'm a bit more of a theorist, but it can surely be that in real life there is some real part to it.
What I thought was that the resistor / capacitor bit would have an impedance of
1/Z = 1/R + i omega C, and hence
Z = R / (1 + i omega RC) = R /(1+omega^2 R^2C^2)]
which calculates to (RC=1) ] (1-i 2pi50)
of which the real part is 10^6 / (1+2^2 pi^2 50^2) ~ 400 ohm

Posted 19/01/2011 14:48:00
mqrius

Oh damn, this is a bit embarrassing. You're right, ideal capacitors don't dissipate, but just store and release energy. Uhm, yeah, ignore my last 2 posts then. Well, that takes care of that problem.
bgreene: Do you know what a typical current is for this to happen? My triacs are rated at a maximum of 1mA leakage current... The other problem might be solved by a fast fuse, I guess. Not sure if that would be an optimal solution, or if it would work reliably though...

Posted 19/01/2011 17:05:24
rikkertkoppes

I currently added a port and attached a relay plug to it. I still need new screw connectors, because the one I got from conrad didn't fit a 2.5mm^2 wire. In between is a bit of foam board for insulation and support
This bit will probably be replaced with a round custom board with either a triac solution or relays lying flat, which creates some more room. The whole thing fits nicely in a standard wall box.
Right now working a bit on the software.

Posted 19/01/2011 21:23:38
bgreene

the problem is caused by the fact that the switchmode regulator is designed to sleep until a bootstrap capacitor has charged up sufficiently for the regulator to run. the triac might be only 1ma but the rc snubber will add to it. try it and see with say a 10w high eff lightbulb. of course, you could always omit the snubber if you don't plan on using any inductive loads. but you still need to watch out for gotchas such as when you switch 220v onto say, the control unit on your boiler which uses an opto-coupler on the input and will operate with very low current or solenoid valves which are highly inductive etc I've just changed the office lights to leds and had to replace the ssr with a mechanical relay 8=(

Posted 19/01/2011 21:28:13
tankslappa

@bgreene, that's annoying. Relays are all well and good, but if you want to do dimming, or avoid nasty RF, it's far easier to do it with triacs.
I know exactly what you mean about the low energy and strip light start up circuits, I have 2 switched in my house (top and bottom of the stairs) which have a neon to illuminate the switch when it is off. The neon lets through enough current for one of the energy saver bulbs to try to start... If you watch it in the dark, when your eyes are used to the dark, you see it pulsing very dimly. Needless to say it keeps killing energy saving bulbs, so standard bulb went in.
Unfortunately the green wash from the UK government has now dictated that energy saving bulbs are all you are allowed now, and filament bulbs are not supposed to be available any more! Gonna be some strange old smuggling going on into this country soon!

Posted 20/01/2011 00:21:27
rikkertkoppes

Got my order from conrad today. Turns out the caps needed are really huge. They take up the majority of the footprint available for the 230 to 5V and 5 to 230V part, which is a 5cm diameter circle to be able to fit in a standard wall box and leave room for wiring. My original (prototype) idea to use the relay plug as is and put the remaining components around it will not work.
Hence, I'll squeeze everything on a new board. Pictured is the setup for the biggest components. The two caps are evident. Input and output connectors are on the right, connector to the jee node on the front and the elco in between. Not added are some resistors, diodes and a Zener. On the back, next to the caps, there is just enough room for one relais, which I'll take off an assembled relay plug, since the relays I ordered are still pending (just as jc's I suppose)
I don't think I'll be able to place an smd relay driver, so this will be replaced with a resistor, transistor and a diode in this prototype.

Posted 25/01/2011 18:31:54
bgreene

Can't see the voltage rating on the caps but make sure they are 400volt rated if you want to be legit. Also, the discharge resistor should really be 2 in series because if it fails you'll be fried. if one side of your circuit is not earthed then a person touching any part of the jeenode will be exposed to 110v. People always think, ah its just 220v, but you have to allow for all possible failures due to components or surges (lightning in the area, whatever). Years ago I would have done it your way but bitter industrial experience says consider doing it the easy way with a psu like 68-626-79 from elv.de. long term, its safer (vde approved)and better (6watts stabilised).

Posted 25/01/2011 21:50:16
rikkertkoppes

They are 400V indeed, that's why they're that big i suppose. What about those two resistors in series? I don't really get that. If you've got one and it fries, it leaves an open circuit, if you've got two and one fries, you've got an open circuit and one resistor, what is the difference?
The whole 230v part will be encased by the plumbing end cap on the right, and it will be in the wall anyway. Furthermore, there will be a fuse on the live wire as well
If I could find a transformer and rectifying circuit small enough to fit in that space, I'd use that as well. But I think, considering the two capacitors, fuse and encasing that the whole bit will be safe enough to feed the jee node (with the info I have now. I could very well be mistaken, but that's one of the reason of posting it here anyway)
Lastly, I will do some testing with the circuit before it is used. That's what this prototype is for. For the real thing, I will definitely have a custom pcb made. Trusting a breadboard like this seems foolish.
For the testing part I still need a separating transformer (bathroom transformer) and a fuse or something. I don't want to get fried while testing :)
Anyway, any comment considering this bit is welcome. It really should be well thought through, best way to do that is with all of us

Posted 25/01/2011 22:11:08
mqrius

Looks nice, if a bit crowded :P
I don't really see why trusting a breadboard like this would be foolish. You can have soldering errors on a pcb as well, and the added structural integrity really isn't all that relevant I think...
Were there no smaller usable caps available? They should list the size somewhere, or perhaps in the datasheet... With a size like that a small transformer might be viable as well, if you can find a suitable one.

Posted 25/01/2011 23:15:34
tankslappa

Woo.. Big caps! :-)
About the two resistors, you want two in series. Why? Well resistors have a breakdown voltage. Normally we don't even notice this as the voltage is usually way higher than we ever need with resistors, but when you start playing with mains you could easily exceed it and the resistor could track and go closed circuit!
Unless you know otherwise, assume the random resistors in your pot are okay to 200v only... And AC mains is >320 peak.

Posted 26/01/2011 07:51:49
rikkertkoppes

Ah, that makes sense. I got these resistors: Widerstaende0207TK50.pdf>, see page 2, its the MBB0207 type
Noted are: Umax: 300V Zulaessige Spannung gegen umgebung (not really sure what that means) - 1 min: 500V - dauernd: 75V
Does that mean these can only withstand an enduring voltage of 75V or (umax) 300V?
In that case I need to find better resistors :)

Posted 26/01/2011 08:30:46
tankslappa

Thanks to google... "Umax: 300V Permissible voltage against ambient - 1 min: 500V - Constant: 75V"
Although I'm still not sure! Yours will be subject to 320v (European mains AC x 1.414 for peak) 50 times a second (half wave rectified), so not constant, but that's > 300V max against ambient :-/
Errrr... Maybe someone who speaks fluent German might be able to help with those particular specifications as I'm not certain enough to be sure, and when playing with mains AC it's best to be certain!

Posted 26/01/2011 08:50:06
bgreene

umax or working voltage is the max voltage between the 2 wires. insulation voltage is what the coating can withstand ie max voltage between the metal filament or carbon block and the environment ie the component beside it or the earth plane below it or your finger on top of it. insulation drops with time (it is an erosion process so to say) and applied voltage.

Posted 26/01/2011 10:42:12
rikkertkoppes

A question with an eye to the future: is anyone interested in pcb's of my project? It will be
1 pcb, square, approximately 7cm sides to fit a standard wall box, with standard mounting holes in the middle of the sides, 6cm apart, containing: - atmega, resonator. - rfm12b. - power stuff (voltage regulator, caps, elco). - ftdi connector. - port 1. - four switches with LEDs.
This basically is a jee node, although in another form factor and stripped down to the bare parts. The ftdi port is facing outward, port 1 is facing inward, unless you mount them otherwise. You need to add a bezel yourself, but presumably most blind plates will fit
1 pcb, round with a 5cm diameter, designed to fit inside the wall box. Because of its dimensions, a standard plumbing end cap will fit over it. This can be used as a case, although holes for the 230 connectors need to be made. It will contain: - a relay, connected to dio of port 1, basically half a relay plug. - a capacitive power supply. - fuse. - 230v connectors.
The two pcb's will plug together. As a standoff and isolation, you can saw a ring from a piece of 5cm plumbing pipe. Presumably, the round pcb will have to be sawn by hand
For now, I only have to finish the power supply and do some extensive testing at 230v. If everyone works fine, I will also prototype the casing and bezel. After that, I am very likely to produce (at least) 5 of these for myself. If anyone is interested I'm happy to produce some more.

Posted 27/01/2011 20:42:23
tankslappa

Sounds very nice, and I would be interested except over here in the UK we have a completely different "standard" wall box... Or should I say series of standards! Some of which are very shallow.

Posted 27/01/2011 21:43:37
jcw

Hey, I'd be interested in one!

Posted 28/01/2011 14:26:58
rikkertkoppes

Finished the 230 v part, hooked it up to mains, after reassuring all solder joints were sound. Everything seems fine. There seems to be a 4.7 v output. No idea how stable it is, cause I don't have a scope or other time resolution measurement device. I expect it a bit saw-toothy though.
If I'm correct, the Zener will create a more or less 50 hz block of 5 v, the parallel elco will smooth that out a bit. Is there anyone willing to build the circuit and make a plot?
From what I have measured, I feel quite confident it won't destroy a jee node.
Fuhermore, I originally (see sketch) designed a fuse in power supply, however, this is a little useless. If I want to prevent people from getting fried, a 1A fuse won't do, since something like 20mA is enough. I'd rather depend on proper casing and the difference switch of the house. Lastly, the circuit wouldn't draw that much current anyway, unless something shorts out, but then I have the central fuses
However, the relay used does have a maximum rating (5A), which, if crossed, may cause problems, sparks, fire etc even without tripping the central fuse. Therefore, I think I will use the fuse in that part of the circuit
Any warnings or thoughts before hooking up the jee node to it?
Next step will be designing pcb's I'm relatively new to eagle, so I hope I can manage. Will post it anyway

Posted 01/02/2011 18:12:22
gjniewenhuijse

very very interested.

Posted 02/02/2011 06:48:56
tankslappa

Looks very neat. Do you have the final power supply circuit you went with and I will see if I have the parts laying about to make one and stick it on the scope.
My PSU design is slightly different. I've opted for full wave rectification before the zener so my cap only has to fill in a tiny gap of the cycle and fill in the missing 180 degrees. Overkill I'm sure, but it does make the required capacitor much smaller.
I really must get round to getting this design off paper (and out of the box of bits I ordered months ago) and onto a bit of board!
BTW, even small fuses aren't much good for protecting people from electrocution. They're too slow. By the time that bit of wire heats up and drips into the glass tube you're already skipping round the room cursing in your chosen language!

Posted 02/02/2011 07:40:50
joop
Here is a simulation of your circuit with some waveforms (zener 5.1v).
```
      green = ripple R(var) 1k and 10k
      blue  = 50 hz block of 5 v (net z,d,r,c) 
```
mainsremote.wmf> mainsremote.wmf>
Posted 02/02/2011 12:34:41
rikkertkoppes

Thanks Joop, that is more or less the block wave and sawtooth I expected. What software did you use for that? Looks very useful!
@tankslappa: the circuit is exactly what is in the second link from Joop. The fuse I described isn't meant to prevent electrocution, since, indeed, it is useless. Proper casing will take care of that.
By the way, I plugged in a jeenode this morning, but no cigar. I have to investigate it a bit this evening, but either the voltage is too low (although 4.7V should be just enough I think), or the whole thing is unable to provide enough power (current).
I'll get back on that one

Posted 02/02/2011 15:56:25
tankslappa

@Joop, that's a superbly useful bit of software, so put me down for wanting to know what it's called. I'm not 100% convinced by its sign wave, but that's a minor point. Just out of out interest, what happens if you put in components outside of their operating range? Is there a nice animation if I stick a 16v cap across the 230v AC? :-D
@rikkertkoppes, the 4.7v should be more than enough to bring up the JeeNode. Joop, could you up the load on the output. If I'm reading the info correctly, the lowest resistance it tries is 1K, which would only be a 5mA load, a bare JeeNode with RF is in the 20-30mA region.

Posted 02/02/2011 23:10:30
joop

Rikkert , maybe there is to much ripple ,see the load with 100-400 ohm like Steve requested, maybe you need a bridge or a bigger elco. mainsremote.wmf>
Steve , the circuit exploded when I tried your suggestion :)... The wave input is a sinus wave top 115 v above zero and top 115 v below zero. not 100% convinced by its sign wave
The software tool = LTspiceIV. from Linear Technology no restrictions it's total free.

Posted 03/02/2011 11:36:29
tankslappa

Wooooo squiggly lines! Quite pretty actually, although maybe not too useful as a PSU!
Thanks for the link Joop, I'll have a play - I want to see the capacitor explode ;-)

Posted 03/02/2011 12:15:29
baustromverteiler

Hej, I'm also interested in some of thosee pcbs . . . Also a single sided pcb design would really be great, so I could etch my own pcbs and slowly expand the system. I really appreciate the great effort you put into this project!

Posted 04/02/2011 11:18:23
rikkertkoppes

I finally did some measurements. My current meter turned out to be broken, so I bought a new one today.
It turned out the system draws up to 50mA with all LEDs and relay on. At 5V, that is an equivalent of a 100? load, which, as Joop showed, is a bit too much for the supply circuit.
So, some debugging this weekend. I might try a bigger elco or a higher voltage zener. The simulation software mentioned above might come in handy to tweak some parameters
Oh, and full wave rectification might also be an option indeed

Posted 05/02/2011 12:38:26
rikkertkoppes

@Joop could you share your simulation file? I am struggling in LTspice to get everything right. Something is wrong with my diode and zener and I can't seem to reproduce your results

Posted 05/02/2011 17:57:11
joop

Rikkert here remote.asc> is the simulation file .
Maybe the diode and zener can't be found ,you can test it with other types.
Joop

Posted 05/02/2011 18:30:17
joop

Here are the two models.
You can insert them between your other models like this: Go to this directory C:\$01Backup\My Documents\LTspice\lib\cmp Just open the D standard.dio file with LTspice ,if LTspice is installed it will open a sort of text-file with all the diode models you can insert these diodes in any position in this file. Save this file and your models are updated.
1N4007
.model 1N4007 D(IS=3.872n RS=1.66E-02 N=1.776 XTI=3.0 EG=1.110 CJO=1.519E-11 M=0.3554 VJ=0.5928 FC=0.5 ISR=1.356E-09 NR=2.152 BV=1000.0 IBV=1.0E-03 Tt=4u)
1N4733
.model 1N4733 D(Is=1.214f Rs=1.078 Ikf=0 N=1 Xti=3 Eg=1.11 Cjo=185p M=.3509 Vj=.75 Fc=.5 Isr=2.601n Nr=2 Bv=5.1 Ibv=.70507 Nbv=.74348 Ibvl=4.8274m Nbvl=6.7393 Tbv1=176.471u)

Posted 05/02/2011 18:52:26
rikkertkoppes

I did some simulations with a bridge recitfier, bigger elco and higher voltage Zener, but it just doesn't work. What does help is lowering the resistors to about 1k, which results in a 150mA amplitude at 150V amplitude, that's 16W rms, which is a bit too much for ordinary resistors.
In the meantime, I also found a very small transformer (22.7 x 32.3 x 21.8 mm, 710287 at conrad), which is just about the same dimensions as the two 1?F caps. On the other hand, I'm not quite sure what is going to happen if I place a relay right next to that transformer. We'll see
While waiting for those, time for some simulations with other transformerless power supply designs. That LTSpice software just came in very handy!
@Joop: I think the voltage source needs to be a 325V amplitude by the way

Posted 06/02/2011 12:20:14
joop

Yes,thanks 3251/V2 RMS.

Posted 06/02/2011 12:54:07
mqrius

@Rikkertkoppes (and others): About the transformerless power supply: Using a bigger low voltage cap isn't going to help under a constant load. It might help if the leds only blink now and then for example.
Under a constant load, however, the maximum output is limited by the high voltage caps. Double their value, and you instantly double the output current. You can try it right now, by just replacing one of them with a wire (right now there's 2 in series, making the effective capacitance 0.5µF.) Or of course you could try to find 2µF caps, if you have the space.
(Using 1k resistors moves the current limiting function from the caps to the resistors. The problem with that is that resistors are dissipative, meaning you suddenly need to lose ~10W of heat.)
Combining this with bridge rectifier gives enough current to sustain ~45 mA in simulation. Perhaps a larger low voltage cap can be used to power the times 50mA is needed.
Here's the simulation: http://forecast.student.utwente.nl/Sockets/45mA%20transformerless.asc (both setups have equivalent behavior.)
I'd be really interested if someone tried this…
Edit: another option to squeeze a little more juice out of the circuit is to use a higher voltage zener diode. The Jeenode can run with up to 13V power supply, so using a 12V zener (or two 5.1V in series) will boost the output a bit (to 60 mA in the simulation). The Jeenode's voltage regulator will get a bit warmer, but it shouldn't cause any troubles.

Posted 06/02/2011 14:12:56
kami

Hey guys,
i just try to catch some signals with the new OOK 443 Plug from Jean-Claude.
I am using this remote device:
http://jeelabs.org/2010/04/23/not-so-home-easy/
I can see with ookrelay2.pde this output:
[ookRelay2] HEZ 85 85 205 204 212 0 HEZ 85 85 205 204 170 0 HEZ 85 213 212 204 212 0 HEZ 85 213 212 204 170 0 HEZ 85 213 76 205 212 0 HEZ 85 213 204 204 170 0 HEZ 85 213 204 212 212 0 HEZ 85 213 204 212 212 0 HEZ 85 213 204 212 170 0
Is it possible to send this commands to control the power switches with a jeenode?
Can someone help me with this stuff?
Thanks a lot.
Cu kami

Posted 06/02/2011 19:56:23
rikkertkoppes

Built it. And yes, it delivers a nice 4.9V under load. You can see the LEDs on the buttons light up. The jee node boots up
This might be a very nice solution. It saves me one of these bulky caps, however, as mentioned in the transformerless power supply article I posted earlier, you might end up with your circuit under mains potential if hooked up the wrong way.
The 2? version seems a bit safer (but too bulky to fit), however, everything will be encased and things have to be connected the right way anyway, otherwise the relay will not be connected to the hot wire.
Edit: it is also capable of driving the relay, it draws 47mA at 4.8V
For the caps: I used 459852 from Conrad, the 2.2?F version is 459877

Posted 06/02/2011 19:58:33
mqrius

Nice :D An important question: Is this solution still cheaper than a tiny transformer?

Posted 06/02/2011 21:48:08
tankslappa

Great work rikkertkoppes!
Mqruis, probably not lol, but that's not the point! It's far more 1337!

Posted 07/02/2011 02:15:35
rikkertkoppes

Here is a parts list (numbers from Conrad). You'd need a rectifier and an elco anyway:
446095 Elco 100?. €0.14. 162272 1n4007. €0.10 = €0.40
For the transformerless circuit: 180572. Zener €0.28. 418617. Resistor €0.10. 459852. Cap €1.31
And with a transformer: 710287. Trafo €4.62
There are some more parts on the print, like a fuse, connectors and the relay circuit, but the difference in price is this €2.93. not much indeed. You might want to spend this on a more reliable supply.
On the other hand, still have to find out whether a relay next to a transformer works anyway.

Posted 07/02/2011 06:28:22
mqrius

@tankslappa: It's not really more 1337; a transformer is probably more reliable...
@rikkertkoppes: Well, the power supply has been reduced from an expensive component (€5) to 'the pheripherals' (<€2), which is pretty nice. Too bad we can't easily cut back on the JeeNode, that's going to remain the most expensive component I guess…

Posted 07/02/2011 08:41:23
rikkertkoppes

I'll build the power supply on a breadboard tonight and probably build a transformer based version when that comes in. Then we can see if that even works and start in the PCB's
As said, I'm planning on 2 PCB's. One for the high voltage part and one for the jee node and buttons. For the UK residents among us, one or both may need another form factor.
Speaking of which: is it really very different? Maybe we can make a version that'll fit both "standards"
As for costs, you also need to take into account these pcb's. Maybe JCW is willing to ship jeenode kits without the standard pcb?

Posted 07/02/2011 09:18:39
mqrius

What's wrong with hole boards cut to proper shape though? Should reduce the cost considerably, no?
Also, when I first started working on this, my plan was to use a cheap PIC that can only receive info, to cut the costs. The JeeNode and the 2-way communication make things easier and more versatile, but also a lot more expensive...

Posted 07/02/2011 10:36:59
rikkertkoppes

Hole boards should work (my prototype is on one), but i'll definitely go for pcb's, since I am planning to build quite a few and all the hours spent on creating spaghetti on a hole board is also worth something.

Posted 07/02/2011 10:43:41
mqrius

True, but in my experience, custom PCBs are pretty expensive. Do we have an estimate on the price of such a thing yet? (I guess it depends on the amount of people buying, but some ballpark estimate would be nice :))

Posted 07/02/2011 12:52:36
rikkertkoppes

I soldered down a 8.2V Zener, just before realizing that that might fry up the relay, which is designed for 4-6V

Posted 07/02/2011 18:01:42
rikkertkoppes

Here is the updated prototype, placed on a spacer ring made out of a 5cm plumbing tube. The cap will fit right over it.

Posted 07/02/2011 19:38:30
tankslappa

Very nice.
To answer your question about how different can a UK socket/switch back box be... Very is the answer!
Have a look at page 35
http://content.yudu.com/Library/A1qipl/RexelSenateElectrica/resources/index.htm?referrerUrl=http%3A%2F%2Fwww.rexelsenate.co.uk%2Fdoc%2F39%2FSector%2520Plus%2520Catalogues.html
"Steel knock out boxes" and "dry lining boxes".
So there is more space to play with width and height wise, but depth wise you would be in trouble. For surface mounted stuff you can get them shallower than 25mm too!
What fun! :-D

Posted 07/02/2011 19:53:54
mqrius

@rikkertkoppes: "I soldered down a 8.2V Zener, just before realizing that that might fry up the relay, which is designed for 4-6V" I believe relays are mainly dependent on current, so just increasing the series resistance a bit should prevent problems.

Posted 07/02/2011 22:10:44
rikkertkoppes

@Tankslappa: I see you've got 47 and 35 mm depth (and even 25), for the rest of it, it looks like there is a little more room sideways. What is the distance between the holes?
To make it (a little more) compatible, we just have to make sure the holes are in the right place and everything is as shallow as possible. What is the most common depth?

Posted 08/02/2011 07:29:56
joop

@rikkert nice work, maybe you should try to test you'r cicuit a few days or a week in an actual real situation on the power-lines, before you make a print layout etc, and check that the Atmel is still running ok.
The Atmel could lock-up or behaves strange if there is to much noise from the powerlines (refrigerators etc)or noise from the environment(sparks etc),it could run for a few days ok and sudden strange things can happen.

Posted 08/02/2011 13:41:22
tankslappa

The shallower ones tend to be used for light switches (typical!), the bigger ones for 13 Amp sockets due to the depth of the back of the socket faceplate (the earth pins on our plugs protrude ~23mm) and the 3.5mm ring main twin and earth cable... Plus of course the switch.
I shall ask my brother, who is an electrician, he will be able to tell me if there is a real defined standard about which box goes where.
The hole centres for the screws on a single box are a nice standard 2 3/8 inches apart ;-)
Oh, okay, I won't be that rotten, that's a smidge under 60.4 mm. I will ask my brother for a few boxes next time I see him and I'll take some measurements and make some cardboard templates I can drop in to find what is practical.
What are you planning to do for the face plate?

Posted 08/02/2011 16:50:46
rikkertkoppes

Well, whaddayaknow. The Dutch (and Belgian for as far as I know) hole distance is exactly the same. The thing is , our boxes are round, leaving a maximum inner radius of about 54mm, which happens to match my prototype. As for the depth, our standard depth is around 41mm, there are deeper ones for central boxes in the ceiling. There are also miniature sized wall boxes, but these are only used to connect some wires for a wall lamp or something
The current depth of my prototype is 38.5mm, due to the distance between the two boards and the height of the cap. We may be able to cut that down a bit.
As for the face plate, I am planning to cut up a commercial blind plate. That way, it stays in style with the rest of my switches. Furthermore, this is customizable by everyone individually, for as far as I know, the placement of the switches will fit inside the size of the inner part of most blind plates I have seen
Well, anyway you should be able to at least mount it, it may stick out a bit though :)
By the way, I hooked up some Christmas lights. It is quite strangely satisfying to be able to switch them through a serial interface via jee link, as wel as the soft buttons on the front
@Joop: that's good thinking. I never thought about being able to reset these things when they are in the wall (other than cutting mains and waiting for the caps to drain). A paperclip style reset button might be a good addition

Posted 08/02/2011 17:43:52
tankslappa

Hehe, nice to know we managed to sneak a nice imperial measurement standard past the EU regulators, now keep quiet! Seriously though, I bet yours are exactly 60mm, where as ours are very definitely the imperial measurement of 2 3/8" (60.325mm).
Generally we don't have boxes in the ceiling.
Oh sht! I just realised that there is a huge problem with implementing your light switch in the UK! It won't work! We don't have a neutral behind the switch to power the thing!
I can tell you're raising an eyebrow right now... Basically after the war we had a shortage of housing (lots of it was a lot shorter than it was and the doors wouldn't open!), anyway cue mass rebuilding. Unfortunately there was also a shortage of copper for wiring, probably due to the demand from the rest of Europe! So the building regulators decided that new buildings should have ring main for lighting and sockets as this requires less copper than the star type arrangement. This wiring methodology is still used today. This means most houses have 2 fuses for the 13A sockets (usually 30A), one ring upstairs and one ring downstairs. Same for lighting, two rings, fused at 5A... Then maybe one for the electric cooker and another for a water heater.
Now here comes the problem... The lighting ring goes from ceiling rose to ceiling rose providing the live, neutral and earth. The ceiling rose has 4 connector blocks, a big earth one, a triple for live, a triple for neutral, and a twin for switched live. The bulb connects between switched and neutral... A twin and earth cable then runs down from the ceiling. This is connected to the live, the switched and the earth point. NO neutral!
Actually, here's a picture!
(Bit of a mix of standards there! Modern ring main cable use brown and blue colours not red and black!).
I got so focused on the transformer-less power supply (which will apply to my 4 way 13A strip) that I forgot your application is going in a light switch!
So as you were, don't worry about trying to make it fit in 25mm depth!
Bugger!

Posted 10/02/2011 21:25:28
mqrius

Ah, interesting… I suppose you could try to add a ground line to the switch, although that does make things harder, indeed…

Posted 10/02/2011 23:54:40
tankslappa

The problem is there is no neutral, so although the switch box has a live, and switched live wire to turn the light on and off, it doesn't have the neutral required to get the power to run the transformer-less power supply and the ATmega. I know I could be really naughty and use the earth, but the leakage trip will put a stop to that very quickly!
The only solution is to pull through some triple and earth cable from the ceiling rose down to the switch, then I can have a live, neutral, switched and earth... Although there is a problem with that... I already have triple and earth in the hall/landing, as it has two way switching, so they're all used, and still no neutral!
On the plus side of things, it does mean I have everything I need up in the ceiling to put the remotes up there without having to do anything to the switch, so no cosmetics to worry about and plenty of room to tuck it into the ceiling space... I just have to sort out the controller. Another mystery battery powered box on the coffee table coming up!

Posted 11/02/2011 00:49:03
rikkertkoppes

A, here it is a bit the same. Ceiling roses are all connected together and wall switches and sockets all run down from the central rose in a room. Our rose box is deeper, with plenty of room to tie everything together (there are no preinstalled triple blocks, so any number of connections can be made as long as it fits)
Normally, for a single switch, no neutral runs down, however, it is fairly easy to add a wire. Also, when you've got switch and socket combinations, a neutral just has to be there
Oh, and it's definitely around 60.4mm here. Kinda strange indeed, but I measured it with calipers on several boxes. Pat yourself on the back, some imperial weirdness indeed came through

Posted 11/02/2011 06:31:54
rikkertkoppes

Now I think about it...
Do we actually need a neutral?
Can't we just draw power from the loop we have? On fact an incandescent light is just another resistor in series. Drawing a tiny bit of current we need for the jeenode won't light it. Not quite sure whether that will work with led or other lights with electronics inside

Posted 11/02/2011 06:50:25
jcw

Pulling current through the lamp is possible, but it leaves a dangerous high voltage when "switched off". Sounds dangerous to me, and probably strictly forbidden.

Posted 11/02/2011 08:59:44
zooer

Power harvesting via lamp/load is used in some commercial dimmers so it shouldn't be illegal. Doing your own dimmers, however, could be open to various interpretations.

Posted 11/02/2011 10:51:31
mqrius

Are you sure those dimmers also draw power when the light's off? Because I kind of doubt that…

Posted 11/02/2011 11:58:59
tankslappa

A combined switch and power socket? Oooh, never seen one of those! They sound kinda fun.
I guess you could harvest a bit of power from the "open" circuit loop, but what happens when you turn the light on? You lose the potential difference between the live and switched live.

Posted 11/02/2011 12:50:10
rikkertkoppes

what about this thing:
this situation is when the light is off. This is basically the circuit we have thus far. There is a very small voltage across the bulb, around 4V, presumably much too low to light it up.
When the bulb needs to be switched on, a relay can short the capacitor and parallel resistor. The bulb now has a resistance of about 880ohm and ensures the voltage across the rectifier circuit is not too big.
Here are the LTSpice simulations:

Posted 12/02/2011 16:00:40
mqrius

I don't quite understand why the resistance changes when the light is on. Because it is heating up? Anyway, even if the resistance doesn't change, I think that circuit would actually work. It's very unsuitable for anything that can't handle being off with a small current still remaining (switching power supplies etc, not sure about the office bar light type thingeys), but if you're already sure you're going to use it with a bulb, then I don't forsee many problems.

Posted 12/02/2011 18:23:02
rikkertkoppes

An incandescent light has about 15 times higher resistance when on compared to off. I did the math based on a 60W bulb (230^2 / 60 = 881 when on, divided by 15 is 58)
This might be a solution for people who are not able to pull an extra neutral wire to the wall socket. I think I just stick to our original circuit, since I foresee switching to led lighting anyway, for which this circuit probably will not work.

Posted 12/02/2011 18:42:23
mqrius

Ah, okay, makes sense.
Yeah, the original design is better, but this could be used in very specific situations.

Posted 12/02/2011 19:05:03
rikkertkoppes

Finally got my tiny transformers (really tiny). About to make a power supply board with those ones and see how the relay behaves right next to it.

Posted 14/02/2011 10:21:50
tankslappa

/me wonders why people always use random objects he's never seen to give scale, and not something like a ruler :-p
What exactly is that? A wine gum with 20€ written on it?

Posted 14/02/2011 16:07:35
rikkertkoppes

A ruler might be interpreted as inches by islanders. That said, this particularly dutch candy (known as drop) might indeed be equally confusing. It is about 23 x 23 mm, smaller than those to caps I had in the transformerless version

Posted 14/02/2011 16:20:18
tankslappa

Nah, it's easy to tell the difference between inches and that new fangled metric stuff... Metric stuff will have sub divisions of 1/10th (and 1/20th)... Inches have 1/2, 1/4, 1/8, 1/16, 1/32, 1/64... I kid you not! (It is the true ancestor of binary!). It's only when you start playing with micrometers and feeler gauges that you'll bump into 1/1000th.

Posted 14/02/2011 16:35:14
rikkertkoppes

@tankslappa the transformer based version will likely end up at 32 mm or less. You would need a neutral for this solution though.
For me, I'll probably go for a transformer and relay solution in the end. It is a little more expensive, but only in the order of a few euros per unit. It will bring a smaller package and a beefier power supply.
Maybe there is a little room for a ir receiver as well.

Posted 15/02/2011 06:57:36
adamfred

Just catching up on this thread. Working my way backwards and saw the comment on running without a neutral.
I have some of these units http://x10-hk.com/store/index.php?cPath=2129 some of them do not require a neutral connection. They worked very well with incandescent lighting but I blew a unit after a week when using compact fluorescent bulb.
So I have opened the broken unit up and I see it has a triac, big cap, caps, +inductors+, diodes, fat resistors, 5v regulator and a tiny riser board with a PIC16F in it.
If you google x10 PIC16F you will find this appnote for using a PIC in an X10 node design with a transformerless power supply and triac output. It has some good guidelines and schematics I think.
http://ww1.microchip.com/downloads/en/AppNotes/00236a.pdf
Just something to check out if you haven't already, I will keep reading back in this thread now :) Also linked from that doc is Appnote AN954 also interesting.

Posted 15/02/2011 13:06:13
tankslappa

I'm not sure we want to base a circuit on one that blew up... Just my own personal thought mind you! ;-)

Posted 16/02/2011 12:35:05
rikkertkoppes

Does anyone have an idea why my tiny block transformer vb0.35/1/6 seems to output 9.35V ac, and rectified measures around 15.1V dc? The transformer is expected to output 6V ac. I have a feeling I am missing something obvious here.
By the way, this is my circuit
it is just the transformer, bridge rectifier and a 100? elco. The board on the left is part of the new power supply board. The relay bit still needs to be added, but my new relays have not been shipped yet and I broke off a pin of the remaining one I got from jee labs

Posted 16/02/2011 20:59:28
janschenk

First, the transformer gives 6 V AC when there is a load.
Because it is a little transformer, the difference in voltage between no load and full load is great, much greater than a big transformer. That's why you measure 9.35V AC. I think you will measure 6V AC at full load.
Second. the voltage (6V AC) is the effective voltage (Ueff) the maximum voltage is sqrt(2)Ueff http://en.wikipedia.org/wiki/Alternatingcurrent#MathematicsofACvoltages
when there is no load, the DC voltage will be Umax. The voltage will drop to about 6V DC at full load. http://en.wikipedia.org/wiki/Reservoircapacitor

Posted 17/02/2011 00:03:29
rikkertkoppes

D'oh, I should have realized the sqrt 2 bit. I didn't know that the transformer gives a higher voltage output without load. Why exactly is that?
Hooked up a jee node and everything seems nice now. The transformer doesn't affect the relay at all, all flux is nicely kept inside, as it should for good transformers.
Bottom line: I'll go for a transformer based power supply and pull some extra neutral wires to the wall boxes
Now it's time for a bit of thought about the board design and visual appeal. There are some construction questions still.

Posted 17/02/2011 06:31:13
tankslappa

I'm sure I know the answer to this rikker, but it's been a while since I've done transformer theory! I usually stick to how they behave in normal use, not abnormal (unloaded). So you'll have to forgive me if the following is completely wrong! I'm going from old memory here!
IIRC without the secondary loaded there is very little current in the primary, this means the volts drop across the primary winding is negligible (think of it as a resistance), causing the core to saturate to higher than normal levels. This skews the turns per volt of the transformer, giving you more volts for less turns in the secondary, so up goes the voltage.
As soon as you load up the secondary, the load on the primary increases, causing a volts drop across the winding, the core saturation drops and your turns per volt returns to the design level.
As I said, that might be complete rubbish, but it looks right to me (at 7am!). Someone else might come past in a minute and correct me!

Posted 17/02/2011 07:12:39
janschenk

also the impedance is a factor. small transformers have a high impedance/internal resistance impedance>
resistance> this also applies to transformers
With small transformers the output voltage varies a lot with different loads because of the high internal resistance/impedance

Posted 17/02/2011 11:04:42
rikkertkoppes

This is the new power supply mounted on tha back of my jee node pcb. Right next to it is the old transformerless power supply. The only difference actually is the cap, resistor and Zener being replaced by a small transformer. The transformer is rated to deliver 0.35W, which boils down to 0.58A at 6V. This is actually about the same as the transformerless power supply was able to deliver.
The new board is a bit lower, a total depth of 31mm vs 38mm, it is also bait more expensive, €2.92 to be exact. However, it just feels a little safer. The powerless supply would be possible to run without a neutral wire, as simulated a few posts earlier.
The new board would fit in the uk 35mm depth boxes, holes in the square board match the Dutch, Belgian and UK holes. You really need to have a neutral though
Anyway, onwards with the enclosure
Oh, the grid on the mat is in cm ;)

Posted 17/02/2011 18:59:33
tankslappa

Very nice work Mr Rikkertkoppes!
My only criticism would be that there seems to be a big metal bolt rather close to the fuse, which I presume has mains running through it?
Apart from that it's superb!
Here's a weird one for you...
Farnell sell the transformer... Here it is
http://uk.farnell.com/block/vb-0-35-1-6/transformer-0-35-va-6v/dp/1711391 - £3.79
But... If you opt for the identically specced and size, but dual 6v output (and connect the two secondaries together to get the 0.35VA)
http://uk.farnell.com/block/avb0-35-2-6/transformer-0-35va-2-x-6v/dp/1131455 - £3.25
So it might be worth checking your supplier to see if you can save the Euro equiv of 50pence... Which is probably 55c
I just noticed it also has split primary, which means those colonials with their lawyer friendly low voltage mains could use it too.
Oooh, just found the 2x6v output with single 0-230 primary...
http://uk.farnell.com/block/vb0-35-2-6/transformer-0-35va-2-x-6v/dp/1131608 - £2.82
So have a look at your suppliers, you might be able to claw some of those Euro things back by getting a twin 6v secondary.

Posted 17/02/2011 20:03:36
rikkertkoppes

That bolt is indeed not in the best place probably, but that's what you do prototyping for. In the meantime, I have been testing it for quite some time continuously. It is running fine all right. However, the transformer is getting a bit warm. Not too much, you can still touch the thing, but I think I need to take into account a bit when making an enclosure. I'll probably drill some vent holes in it.
For series production however, another problem has cone up. The relays I planned (the same ones JC is unable to get for quite some time for his relay plug), won't be deliverable until the end of April. A normally sized relay may be an option, but at the moment I don't see how I'd fit that on the board.
Oh well, let's concentrate on board design, software and aesthetics. I'd like to button functions to be set via wireless. Since there is only one relay per jee node board, I'll probably use the jee node identifier letter to address the relay. Then, for every board, I need to maintain a list of what every button should do.
I can think of a couple of actions for buttons. - toggle one or more relay (if off, turn it on, if on, turn it off). - turn off one or more relay. - turn on one or more relay. - turn off one or more relay and undo on second press within a certain amount of time.
The last action would be useful for "all off" buttons I'd like near doors. It is useful to be able to turn all lights in the room, but it would't make much sense to turn all lights on when pressed again (and make it behave like a toggle switch). Instead, it seems more useful to just restore the previous state if pressed again.
Then, for all four buttons on every board, I need a setting like: [button nr] [behavior] [relay letter]+
These setting should be stored in EEPROM memory. Furthermore, I think the relay and button led state should be stored as well, in case some glitch in the mains voltage reboots the jee node.
Lastly, I am still not sure what to do with the button LEDs. It is pretty useless to have them indicate whether a light is on, if that light is in the same room. For lights in other rooms, it could make sense. On the other hand, making the LEDs behave differently for different buttons might be confusing. I might have it indicate that some lights are on for the "all off" buttons. That might also visually discern those "special" buttons from the rest.
The LEDs may also serve some special purpose, like indicating menu options or something. I am not quite sure whether I want to be able to change the settings with only buttons and LEDs as in and output. You might press two buttons simultaneously to get into programming mode, then press a button you want to program and then cycle through the lights you want to control with it. Not quite sure about all that still. But ey, it is software, it can be changed (there is an ftdi interface on the board)
Anyway, any ideas about it all?

Posted 19/02/2011 19:32:41
rikkertkoppes

@jcw I am currently in the process of board design, would you be so kind share your eagle library? That would save me a lot of searching for specific components

Posted 23/02/2011 18:05:40
tankslappa

Rikker, I'm not really familiar with Eagle, so forgive me if I'm talking rubbish - you should be used to it by now ;-) but aren't eagle files the .sch and .brd at the bottom of this page? http://jeelabs.net/projects/hardware/wiki/JeeNode

Posted 23/02/2011 22:41:20
herve

@rikkertkoppes
I was working on the same power supply type, but with the 2x115V-2x6V (2x0.175VA) BLOCK transformer. I'm having issue with the amount of power, not having more than 100mA in DC output (with a 100uF/10V filter). Do you know if there is a "special" way to connect such a transformer. I've put both 115V in serial and 6V in parallel (I need 6V not 12V). Any Idea? Did you check how much DC power you was able to get with the one you got?
Thanks!

Posted 27/02/2011 12:12:40
rikkertkoppes

@tankslappa all the files of all boards are on the wiki indeed, but jcw uses a nice library in which he cleverly gathered useful and custom components. I'd really like to use it too, would save me a lot of time (JC hope your watching)
@herve. These small transformers are indeed rated for very small currents. I believe mine is up to 60 mA or something. I don't think you could get much more out of it, it's just a size problem in the end
In the meantime, the relays I ordered finally arrived. These are the same ones jcw uses for his relay plug. Now I am sure I got these, I can go on with the board design.
Fir the rest, I have been working a bit on the firmware. Nothing definitive yet though

Posted 09/03/2011 18:42:38