Three Switches, Three Lights, and Some Relays

"they can be switched off by any switch"

If any switch remains in the "off" position, what happens then? Does it pre-empt the other switches? Say one or more switches is/are in the "off" position and you attempt to turn on some lights using another switch. Which switch has priority? The "on" one or the "off" one? And say you don't want to turn on all the lights, just some of them. In this scenario one or more switches will be in the off position, shutting off all the lights. That's why this problem looks complicated. Using conventional toggle switches is at the root of the problem.

If you use DPST momentary switches with center off in place of the toggles, your problem would be a lot simpler. Push the switch one way to turn on some lights (you can easily rig DPDT relays to latch), and push the switch the other way to turn them all off.

You are dead right, of course I mean that the latest switch that changes position takes priority, momentary switches would make my life much easier, but this is not an option for this project unfortunally.

What is your budget? Relays are expensive and the design will require a number of them.

How comfortable are you with digital electronics?

Three possibilities:

1) The digital solution requires 4 one-shots (555s), three R-S latches, seven diodes, three DC-operated power relays and a DC power supply. You can replace the power relays with solid-state relays if desired, as they interface with logic levels.

2) The all-relay version requires four time-delay relays (NO with contacts closing for a few seconds when coil is energized), three latching relays (pulse on/pulse off), seven diodes, three DC-operated power relays and a DC power supply. Much more expensive than digital version.

3) If you know how to use and program a PIC microcontroller, it gets cheaper yet, but you will still need three DC-operated power relays (solid-state highly preferred) and a DC power supply.

Wow, thanks for your quick reply. Your solution no.2 have given me some ideas, I will start drawing something now. (still not sure if I ll get there)

fred w.

I'm working up a sketch. Send me a PM with an email address and I'll mail it to that. Uploaded images on CR4 are fairly lo-rez regardless of the original resolution and I'm afraid the details won't come through.

If you're familiar with logic "Truth Table's," can you construct one? If not, then you'll need a processor to constantly monitor the states of all lights and switches. Whether the processor is hard-wired (so to speak) or programmable is neither here nor there, but the process you describe is certainly doable.

One of the things you don't mention is whether or not the switches must handle the power to the lights. Another thing not mentioned is whether or not the switches you do mention are of the ordinary variety used in electrical construction, even though you do say that "momentary switches' are not an option.

A four way switch is what you need. Placed between two three ways to turn lights off in at location. The four way swaps the travelers.

I can make the third three way switch into a 4 way switch by using it to power a relay with two NO and two NC contacts, but it still wouldn't fit my bill as I need Switch A to turn on light A and B. Swith B to turn on light B and C and Switch C to turn on lights A, B and C. Each Switch must also be able to turn off All lights in a LTP fasion (latest takes priority) I.e. if lights A and B is lit, next Switch that is turned is going to turn them off. If this Switch is turned again, it will only light up the assigned lights for that switch and so on.

Did you make a Karnaugh map out of it?

http://en.wikipedia.org/wiki/Karnaugh_map

Makes the problem more surveyable

The difficulty is that the circuit must maintain state and so is not purely combinatorial.

Btw, the Quine-McCluskey algorithm - a tabular approach - is much easier to use than Karnaugh Maps and is less error prone. Practical Karnaugh maps are good to four terms, but doable to six with considerable difficulty. Beyond six terms they're completely unmanageable. QM can handle arbitrarily large numbers of terms, but the map doubles in size with each additional term and such exponential growth limits its practical size to thirty terms or so. QM's methodology is well suited for computer-aided optimization, fortunately.

The method of choice today for medium-to-very-large-scale FPGAs and such is the so-called "Espresso" heuristic logic minimizer which can handle an arbitrarily large number of terms with ease and only slight growth with each additional term. Everything has tradeoffs, however, and this approach does too: it does not guarantee global minimization although it comes pretty close.

The switches are double pole triple throw. Either the power or ground opened in the center position.Simple.

Posts #1 and #2 consider that option. Unfortunately, it's not an option.

You can't have three three way switches in one lighting control. What you are describing is a master switch in series with two three way switches. You will have to have a separate switch to turn on C.

You would probably be better off using an addressable system of switches and relays such as an X-10 type device.

I was trying to avoid this, but what the heck, I don't think much these days . . .

One way to solve this problem (without using tried and true techniques) is to start with the output states you desire.

In other words, you want any switch to turn all lights OFF, regardless of their present state (or so it seems). Therefore, you must have SET/RESET latches feeding each light, and all RESET inputs to the latches tied together.

Any rising edge of any switch must turn on either A and B, B and C, or A, B, and C.

Any trailing edge of any switch must turn off either A and B, B and C, or A, B, and C.

Any change in state of any switch turns all lights OFF (if I have interpreted your original post correctly).

Dual monostable multivibrators (kind of the delay relay mentioned in someone else's post) must be connected to each switch, and the switches must (should) be wired such that when they are ON, they are pulled up to a high voltage (And I don't certainly necessarily mean 120VAC, so please, everyone, don't jump my case. Describing what needs to be done is kinda like talking someone down in a 747 who has never flown anything bigger than a Cessna), and when they're OFF, they're grounded.

Alright, we have inputs and outputs. Inputs produce a pulse, and we can tell which way they were flipped last by remembering whether or not the rising edge triggered an output or a falling edge triggered an output. This can also be accomplished by using SET/RESET flip flops.

Now we have memories of states of switches. It now is just a matter of combinatory logic to determine what to do with the knowledge of the states of the switches, and I think I'll let someone else explain that part to you.

By the way, all logic and transitional functions I have mentioned above can be done with 120 volt stuff.

Hi,

Hope this sketch helps.

The dotted line shows the positive supply route.
The wiring to the lamps is omitted for simplicity.
The switches need to be positioned "normal" to give the
requested selection. (ab only, bc only, abc-all on or off.)
The first switch (left) being a master of the supply; where
the "off" position ensures all circuits are safe. (no supply)

There are probably better circuits but this is "off the cuff."

jt.

Freedom of the mind is the only true freedom.

As one of your requirements for this installation is to use existing 3-way wall switches, I've simulated the action of a momentary, center-off SPDT switch by means of two "one shot" time-delay relays per switch. Each time a switch is toggled one way or the other a pulse is sent to the associated latching relay. The "on" position sets its relay, and the "off" position of any switch resets all three relays. Even when one or more of the toggle switches remains in the "off" position, any of the three switches can turn on its associated lights. I believe this is the behavior you desired.

The latching relays connect a DC voltage to the power relays by means of a diode network. The network implements the truth table that determines which lights are lit, depending on the new state of the last switch toggled. The whole thing is a form of state machine, implemented using relay logic, as indicated in your posts.

I don't suppose this is a "hallway switching" application - if so Multiway switching may solve it for you.

Hi Kyzine,

The OP's client's switching requirements could not be solved by means of multiway switching, unfortunately. It certainly would have been a lot simpler - and cheaper - if the new scheme could have been implemented with 3- and 4-way switches, but the client specified that his rather peculiar switching requirements must be implemented using the 3-way switches already in place.

Hi Eu,

Agreed, but it's my nature to wonder if a clients requirements are indeed what the client actually means it wants. "peculiar" being the operative observation, and us not knowing if all the on's and off's are indeed 'necessary' in the practical manifestation.

just exploring.

It could be done with much simpler components and logic if the switches were 2 pole (DPST).

If not, the latching relays would work, but it would be far less expensive and less complex to just use a Siemens LOGO! smart relay. They cost about $150US with software, then you can program it on your PC to do whatever you want and no mixing of AC and DC components etc..

_{^{JRaef, a 3-way switch is a DPST switch by any other name.}}

Siemens makes nice stuff. In 2000 I wrote a compiler (in C++) which generated Step-7 code for Siemens PLCs.

No, a 3-way switch is a SPDT (Single Pole, Double Throw), a.k.a. "changeover switch, Form C switch etc.

I said Double Pole, Single Throw. On-Off, but 2 poles side by side. The 2 poles would both switch at the same time, i.e. you would use one if switching 2 phases or both sides of a 240V single phase circuit. By having 2 poles on the same switch, you can isolate the circuits so that one pole of each switch operates a relay coil and the other pole on each switch operates a master circuit which enables any other combination of relay contacts connected to the lights.

Never mind about this however. I tried to draw the logic and realized it can only allow a switch to turn off the lights it turned on. No way for a binary (2 position; On-Off) device to have 3 logic states without adding pulse and latching relays as previously described, or using it as in input to a smart device.

Here was my logic, flawed as it is.

You're right, of course. My bad. Look, it took me 30 years to get this CRT Tan, okay?

_{^{psst: why are all those capacitors labeled as resistors? That ain't gonna work.}}

"_{^{psst: why are all those capacitors labeled as resistors? That ain't gonna work.}}"

I once hired an "Engineer" who asked me that exact thing... turned out she failed t tell me tat her Electrical Engineering degree was in EE / Computer Science. She also considered 120V to be "high voltage".

Just an update to my initial post, I ended up using a mixture of europium's excelent diagrams for relay's and contactors pluss some PLC programing and a Crozet PLC unit. I hadn't planed to go down the PLC route, however this made sence economically and also gave me the oportunity to later implement this system to a C-bus or DALI network. I added some more functions such as LED indicators, burglar alarm/ world timer and "dusk to dawn" inputs and the system has now been running smoothly since November. The reason that I couldn't change the switches themself was that they where particularly specified by the client.

Thanks for all the comments and suggestions.

Fred.