Using lighted doorbell buttons with 3.3V logic

[madsci1016]

I haven't been able to think of any ways to do this so I'm seeing if anyone else might have an idea.

I want to be able to use a handful of regular old lighted doorbell buttons in a display I am building for my local science museum. The button will be read by a 3.3V logic multiplexer.

But in order for the button to light up, it needs to float at 16V and draws about 40mA. The light draws power from the button contacts when the button is open. Protecting the logic from 16V is easy, a zener and a resistor. It's pulling the line up to 16V and letting the button draw 40mA in such a way that the button can be pressed without damaging the 16V supply that is stumping me. Also, I'd like to be able to light 8 buttons off the same 16V supply.

A 100ohm pull up resistor would have to dissipate 2.5W while the button is closed and would cause a 4V drop when the button is opened. Neither is practical.

Any ideas on how I could do this?

[Soeren]

Hi,

But in order for the button to light up, it needs to float at 16V and draws about 40mA. The light draws power from the button contacts when the button is open.

I assume this means it doesn't light up when the switch is closed?

Protecting the logic from 16V is easy, a zener and a resistor.

If the 16V supply is even modestly regulated, a resistive divider will be just as good and cheaper.

It's pulling the line up to 16V and letting the button draw 40mA in such a way that the button can be pressed without damaging the 16V supply that is stumping me. Also, I'd like to be able to light 8 buttons off the same 16V supply.

Not sure how the buttons are wired, but I assume they just short out the lamp.
When you series connect a doorbell (oldfashion type at least) with a 40mA lamp (16V/40mA = 400 Ohm, the lamp will see most of the voltage, as the doorbell has got a much lower impedance and hence the lamp will light. Shorting out the lamp, the doorbell sees the full voltage and makes a racket.

A 100ohm pull up resistor would have to dissipate 2.5W while the button is closed and would cause a 4V drop when the button is opened. Neither is practical.

Any ideas on how I could do this?

You might find it overkill, but a CCG (2 transistors and two resistors) for each of the lamps would solve it. It would need a slightly higher voltage, as the CCG will drop a volt or so.
When a button is pressed, the respective CCG must be able to handle around 0.7A for as long as the button is pressed.

[madsci1016]

Thanks for the reply Soeren, but further conversations with museum folks have lead to the decision not to use doorbell buttons, they don't believe they are kid resistant enough.

But thinking about the design does bring up another puzzler. Same idea, but the button has a Red/Green bi-directional LED across the NO contacts. I would want to be able to control what color the button is (red or green) but still be able to read it, but only need it to be read while it is green. I'm going to think about it tomorrow, but figured I'd ask as well.

[Admin]

I know it's irrelevant now, but a quick comment about the opening post . . . a resistive divider, dividing by 8, might work. That gives you a 2.0V and 1.5V difference between high and low (from 16V to 12V). It depends on what voltage logic the 3.3V mcu you are using cares about.

As for your second problem . . . multiplexers have internal resistance measured in the kohms so you can't power the LED directly from it. You'll need to add transistors to control the LEDs with the multiplexer. It would be easier to control the LEDs directly from the mcu I/O if that's an option (and current is low). Reading the two-way LED is pretty easy, in that one of the pins will either be high or low depending on which way the current is flowing.

[Soeren]

Hi,

[...] they don't believe they are kid resistant enough.

They may be right on the money there.
On Danish train stations, there used to be emergency phones with 5mm stainless steel panels and buttons of the same material, less than 0.5mm protruding - I've seen a few of them ruined.

But thinking about the design does bring up another puzzler. Same idea, but the button has a Red/Green bi-directional LED across the NO contacts. I would want to be able to control what color the button is (red or green) but still be able to read it, but only need it to be read while it is green. I'm going to think about it tomorrow, but figured I'd ask as well.

OK.
Does it have to be bidirectional LEDs (2 pin) or can it be 3-pin LED's (which could reduce the circuitry from 3 to 1 transistor(s) per button)?
Is it a must to have the button physically shunt the LED?

[Soeren]

Hi,

Reading the two-way LED is pretty easy, in that one of the pins will either be high or low depending on which way the current is flowing.

There will always be a voltage drop depending on direction, but the "other end" won't be ground, so how would you read it "pretty easy"?
Even worse if the button have to physically short the LED.

I don't say it's hard, but "pretty easy" sounds to me like something you can do while on the phone or such 

[madsci1016]

The reason I posed the design like that is for ease of wiring. This will be for an installation in an exhibit that will involve several "Push to learn about X" stations. The director liked feedback at each station in the form off a red/green led, so it was easy to see what station was playing, or if it was ready for another button press.

It would be easy to run a single cat5 between all the stations and to the control box and I'm willing to trade more parts in the control box for this, if it's reasonable.

[madsci1016]

But really if it was just a green LED that would either be steady lit at every station when the system was idle, and only blinking at a station that was playing while the rest were dark; I bet the museum would still like it. And that would be easy to do.

[Soeren]

Hi,

Perhaps the attached circuit...

Edit: D1 is redundant.

[madsci1016]

Won't tying the button to Q2's base mean the highest the PB_OUT will float is 0.7V? Not easily read by the controller.

The only thing I don't like is that it requires two lines per station and a common vcc and GND rail. So one cat-5 cable could only carry 3 stations.

If you tie color and PB_OUT together and remove Q2, then you could use one line to read the button and set color, with the obvious effect that when the controller pulls all button lines low to make the LEDs red, the button could no longer be read, but that is ok. Button presses when audio is playing will be ignored anyway.

But I'd still like to also keep the ability for turning the LED off (also don't need the read the button during this time), the original function of Q2. If the controller's logic was tri-state, could you use the High-z state to turn the LED off? Hmm, what about the third state being the controller pulling the button line past the VCC voltage at the controllers, like to 5V.

I think this just turning into a mental challenge if anything. The more i think about it, the more it makes more sense in simplicity just using green LEDS at the stations parallel with the buttons, with switchable pull-ups on the controller to turn them on or off.  One custom board to make instead of one for each station.

[Soeren]

Hi,

Won't tying the button to Q2's base mean the highest the PB_OUT will float is 0.7V? Not easily read by the controller.

Yes, you're right. I originally had a base resistor on Q2. Then, when I went on to finish it after some shopping with my GF, and as an afterthought thought that R2 would keep the base current in check, forgetting the output in the process - my bad.

The only thing I don't like is that it requires two lines per station and a common vcc and GND rail. So one cat-5 cable could only carry 3 stations.

If you tie color and PB_OUT together and remove Q2, then you could use one line to read the button and set color, with the obvious effect that when the controller pulls all button lines low to make the LEDs red, the button could no longer be read, but that is ok. Button presses when audio is playing will be ignored anyway.

But I'd still like to also keep the ability for turning the LED off (also don't need the read the button during this time), the original function of Q2. If the controller's logic was tri-state, could you use the High-z state to turn the LED off? Hmm, what about the third state being the controller pulling the button line past the VCC voltage at the controllers, like to 5V.

It all depends on how much it can deviate from your original spec's and what you need minimum.

I think this just turning into a mental challenge if anything. The more i think about it, the more it makes more sense in simplicity just using green LEDS at the stations parallel with the buttons, with switchable pull-ups on the controller to turn them on or off.  One custom board to make instead of one for each station.

While I don't see any need for the light to go out while the button is held down(?), i does make sense to keep things like that as simple as possible and the red/green indication would be more needed if you were using incandescent lamps, as LED's are very hard to kill (when driven and assembled well), so if it's out, it's probably for a valid reason - and I think the average public are getting accustomed to the longevity of LEDs as well.

[madsci1016]

While I don't see any need for the light to go out while the button is held down(?)

Not a need but a side effect of the need to keep things simple. IE, I don't care if it does go out while the button is pressed, but I do care if I have to spend more money/time designing and constructing this system.

LEDS aren't out completely, just getting fancy with Red/Green LEDs is out. I think just a green LED will suffice.

[Admin]

Reading the two-way LED is pretty easy, in that one of the pins will either be high or low depending on which way the current is flowing.

There will always be a voltage drop depending on direction, but the "other end" won't be ground, so how would you read it "pretty easy"?
Even worse if the button have to physically short the LED.

I don't say it's hard, but "pretty easy" sounds to me like something you can do while on the phone or such 

Since it's 3.3V logic, the typical voltage drop of yellow and green LEDs would give you binary logic. But it'd probably be better to measure the button voltage, then control the LEDs separately. I'm not sure how he's wiring it, but if it was just attached directly to the mcu I/O without transistors, you'd have one end connected to ground (the I/O set to low).

I don't like the transistors method simply because it's a lot of wiring (I read multiplexer . . . how many buttons?). But I'm not sure of madsci1016's design limitations or desired LED current draw so he might not have any choice . . .