Thermostat control: relays vs TRIACs?

[gxti]

So the thermostat my new place is a decent early digital type but it's too close to the nice, toasty "wiring closet" so the rest of the place is colder and subject to fluctuations throughout the day. So I've got it in my head to make a thermostat with remote sensors :-)

Now the question is, what's the best way to drive those 24VAC control lines? Relays would be the first thing everyone recommends, and they can handle more situations (millivolt systems, which this one is not), but if I'm designing for my specific situation, would TRIACs work well? Specifically, I've got a 24VAC system with hot, common, heat, fan, and cool -- pretty much a standard setup.

Here's my first pass at a TRIAC design: http://partiallystapled.com/~gxti/circuits/2011/11/06-tstat-triacs.png -- not simulated or breadboarded but in my uneducated and inexperienced opinion it looks pretty good. VCC is 3.3 volts, so those resistors should be putting 15mA through the TRIAC, 5mA more than the datasheet calls for. Does the "gate voltage" work line a forward voltage on a diode, where I need to do (Vcc - Vgt) / Igt to get my resistor value? In that case I'd want 120 there.

[Uncle Vernon]

So the thermostat my new place is a decent early digital type but it's too close to the nice, toasty "wiring closet" so the rest of the place is colder and subject to fluctuations throughout the day.

What this says, is that there is nothing wrong with the thermostat you have other than incorrect placement.

So I've got it in my head to make a thermostat with remote sensors :-)

That may be be a solution but and of the day yourequipment is still binary in operation.On or Off, sensing that one room is colder/warmer than another, wont equalise the conditioned space. What do you intend,manual selection of sensor authority, averaging, hi select, lo select?

Now the question is, what's the best way to drive those 24VAC control lines? Relays would be the first thing everyone recommends, and they can handle more situations (millivolt systems, which this one is not), but if I'm designing for my specific situation, would TRIACs work well? Specifically, I've got a 24VAC system with hot, common, heat, fan, and cool -- pretty much a standard setup.

How sure are you that Triacs will work well? What is the actual load, while it may beturning on a furnace or compressor, odds on those control lines switch a small glass relay or feed into a control PCB. You may well find you need electrical additional burdon to ensure the triacs latch reliably. Your milage may vary but as a universal solution most designers would opt for relay or transformer switching.
The other point I'd note is that most commercial HVAC equipment would use high side switching so you you would be better tying your triacs the the 24VAC rather than the neutral.

[gxti]

What this says, is that there is nothing wrong with the thermostat you have other than incorrect placement.

I can't move the thermostat because the place is rented and I'm not interested in fishing stiff wires through the walls in any case. As for the sensors I'll be either using an average, or perhaps have it switch from one sensor to the other on a timer.

How sure are you that Triacs will work well? What is the actual load, while it may beturning on a furnace or compressor, odds on those control lines switch a small glass relay or feed into a control PCB. You may well find you need electrical additional burdon to ensure the triacs latch reliably. Your milage may vary but as a universal solution most designers would opt for relay or transformer switching.

I've looked at the internal wiring so I know that the control lines feed into 120VAC relays. I don't think holding current is a concern though as long as the gate is held high, but this might be a misconception on my part. In addition the hold current for these TRIACs is only 10mA which is less than the coil current of even most of the tiny relays I was looking at.

The other point I'd note is that most commercial HVAC equipment would use high side switching so you you would be better tying your triacs the the 24VAC rather than the neutral.

That's what I was aiming for. MT1 on the TRIACs is connected to the hot side, and I also used that as circuit ground so that I can easily hold the gate above that reference point. One benefit relays (or opto-TRIACs) have over this configuration is that I wouldn't have to worry about reference points, but if this configuration works then I'll give it a shot.

[Uncle Vernon]

What this says, is that there is nothing wrong with the thermostat you have other than incorrect placement.

I can't move the thermostat because the place is rented and I'm not interested in fishing stiff wires through the walls in any case. As for the sensors I'll be either using an average, or perhaps have it switch from one sensor to the other on a timer.

Doesn't leave a lot of scope for running additional sensor cables either. You could go RF but expense vs return ratio isn't looking real flash.

How sure are you that Triacs will work well? What is the actual load, while it may beturning on a furnace or compressor, odds on those control lines switch a small glass relay or feed into a control PCB. You may well find you need electrical additional burdon to ensure the triacs latch reliably. Your milage may vary but as a universal solution most designers would opt for relay or transformer switching.

I've looked at the internal wiring so I know that the control lines feed into 120VAC relays. I don't think holding current is a concern though as long as the gate is held high, but this might be a misconception on my part. In addition the hold current for these TRIACs is only 10mA which is less than the coil current of even most of the tiny relays I was looking at.

TRIACS may not turn on reliably with non-resistive loads, because due to the phase shift the holding current may not be achievable at trigger time.

The other point I'd note is that most commercial HVAC equipment would use high side switching so you you would be better tying your triacs the the 24VAC rather than the neutral.

That's what I was aiming for. MT1 on the TRIACs is connected to the hot side, and I also used that as circuit ground so that I can easily hold the gate above that reference point. One benefit relays (or opto-TRIACs) have over this configuration is that I wouldn't have to worry about reference points, but if this configuration works then I'll give it a shot.

It is a common not a ground. Your drawing shows the hot lead at the same potenial as your circuit ground, not a good thing. Opto triacs such as MOC3020 and MOC3021 would be your safest path.  http://www.jaycar.com.au/images_uploaded/optocoup.pdf

[gxti]

I'm going to have to get some actual parts and figure this out the old-fashioned way, there's too many conflicting conclusions bouncing around in my head and I won't know for sure unless I try. I don't see any reason that a straight TRIAC shouldn't be able to switch a small inductive load if an opto-TRIAC can.

As for the common/ground/hot/whatever thing, I'm not shorting anything anywhere. I have simply selected the red wire to be common on my PCB, instead of the black one. It's AC so after rectification it's all the same. And since, as you said, the red wire needs to be the one switched into the output, this makes it easy to reference the gate signal to its MT1 terminal. Defining red == 0VDC means that 0VDC turns the TRIAC off and 3.3VDC turns it on.

[Uncle Vernon]

I don't see any reason that a straight TRIAC shouldn't be able to switch a small inductive load if an opto-TRIAC can.

There isn't! I suggested Opto to ensure isolation, try it by all means it may work perfectly in your situation but switching small inductive, or just Small loads via triac is often problematic.

As for the common/ground/hot/whatever thing, I'm not shorting anything anywhere.

Your diagram show your 0V control logic as ground and also shows the mains control line grounded. Do you intend your controls to be at mains potential?

I have simply selected the red wire to be common on my PCB, instead of the black one.

which is shown at the same potential as your controls, this puts your entire circuit at mains potential. Are all those remote sensors going to be suitably cabled and insulated?

[gxti]

No mains here, boss. Just 24VAC. Sorry if the second post was misleading, the mains is all safely contained inside the furnace. I'm going to go the TRIAC route, but maybe see if I can get a second set of footprints for relays in case it doesn't work out.

[Uncle Vernon]

No mains here, boss. Just 24VAC. Sorry if the second post was misleading, the mains is all safely contained inside the furnace. I'm going to go the TRIAC route, but maybe see if I can get a second set of footprints for relays in case it doesn't work out.

Yes reading back you did say 24VAC, but you did also mention feeding 120VAC relays.  Regardless I'd still be looking to isolate the furnace control voltage from that of my own controls, but that is your design decision. Only takes one phase reversal or secondary ground change to let the smoke out.

Cheers

[jackbob]

Triacs will work reliably if you use them right. I have a topic called small projects and have a great example board in there of triac switching. They will work fine with inductive loads, even read the datasheets. Also keep in mind that solid state relays exist too. If you decide to go the triac route oversize the triacs by to about double the continuous load draw to be safe. Use triac drivers such as moc3041's like my example and you should be ok.

[Zero999]

Another way to isolate the control voltage from the current being switched by a TRIAC is to couple the gate drive via a small transformer. You just need to ensure a pulse is sent to the transformer every time the AC voltage being switched passes through 0V.

A totally different solution altogether is to use two MOSFETs back-to-back with the gates driven by a small transformer and a rectifier or across high voltage capacitors and a rectifier. The capacitors should be Y-rated to provide adequate isolation from the AC circuit being controlled. The AC control voltage needs to be a high frequency as to keep the MOSFET's gates charged and allow for low value capacitors to be used to minimise leakage. The AC control can be derived from an astable circuit such as a couple of logic gates in the capacitive coupled solution or a blocking oscillator with the transformer.

[ivan747]

Sorry for not bringing any contribution, but, does anybody have a link to some webpage where I can find information on 24V systems. They seem to be very interesting.

-----
Just a comment: About a week ago I saw a little digital thermostat called the Nest. It is supposed to be smart and know your schedules, ad it also adjusts to your preferences.
-----

You may want to use TRIACs since they are smaller and they don't wear off, but your circuit must be referenced to the voltage on one your TRIAC's anode. For interfacing TRIACs with mains, I found this application note, which may be handy for you:
http://ww1.microchip.com/downloads/en/AppNotes/00958A.pdf

Now that I think about it, didn't 24V systems actually remotely control relay coils on electric heaters, valves on gas heaters and something else on A/Cs? Are you trying to replace these things?

[tecman]

You have one big problem.  Triacs require bi-polar triggering to operate on AC.  Your circuit seems to just use a 5 volt unipolar trigger signal.  You will as a result only turn on for 1/2 of the sine wave.  Best solution is to use an opto with a triac output to drive your main triacs.  Something like a MOC3061.  This will provide bi-polar triggering and full AC output.

paul

[Zero999]

You have one big problem.  Triacs require bi-polar triggering to operate on AC.  Your circuit seems to just use a 5 volt unipolar trigger signal.  You will as a result only turn on for 1/2 of the sine wave.

That's not true, TRIACs will trigger on all four quadrants, they're just more sensitive on some than others, see the Microchip application note for more information.

Here are some more application notes on TRIACS.
http://www.datasheetcatalog.org/datasheet/SGSThomsonMicroelectronics/mXyzyuz.pdf
http://www.datasheetcatalog.org/datasheet/SGSThomsonMicroelectronics/mXyzyvq.pdf

[tecman]

You have one big problem.  Triacs require bi-polar triggering to operate on AC.  Your circuit seems to just use a 5 volt unipolar trigger signal.  You will as a result only turn on for 1/2 of the sine wave.

That's not true, TRIACs will trigger on all four quadrants, they're just more sensitive on some than others, see the Microchip application note for more information.

Here are some more application notes on TRIACS.
http://www.datasheetcatalog.org/datasheet/SGSThomsonMicroelectronics/mXyzyuz.pdf
http://www.datasheetcatalog.org/datasheet/SGSThomsonMicroelectronics/mXyzyvq.pdf

If you look at your referenced docs, you will see that they define the 2 quadrants of operation when triggered by a unipolar source.
Look at:  http://en.wikipedia.org/wiki/TRIAC  In that they show the trigger polarity required for each quadrant of operation, and you will note that both polarities are required.  With a single polarity trigger, it is basically acting like an SCR.

paul

[Zero999]

Look at:  http://en.wikipedia.org/wiki/TRIAC  In that they show the trigger polarity required for each quadrant of operation, and you will note that both polarities are required.  With a single polarity trigger, it is basically acting like an SCR.

No.

If the gate trigger is always positive with respect to MT1/A1 and AC is flowing through MT1 and MT2, the TRIAC will trigger alternately in quadrants 1 & 4.

Triggering in Quadrant I
Quadrant I operation occurs when the gate and A2/MT2 are positive with respect to A1/MT1.

Triggering in Quadrant IV
Quadrant IV operation occurs when the gate is positive and A2/MT2 is negative with respect to MT1.

The TRIAC will operate as a TRIAC regardless of the polarity of the gate vs anode current. It will only go into SCR mode if it's only triggered on positive or negative cycles or the anode current is DC.

[gxti]

I've shelved this project for a while but these are some good ideas. Transformer drive is overkill (the goal was reduce part cost ) but I like the idea of driving through a capacitor. The downside is that with only 3.3Vpp the capacitor is going to cut that in half again since it would be re-centered on the MT1 reference, thus 1.6 volts above and below but I can probably find a way to drive it with the rectified 24VAC I'd have feeding in to power the controller.

Another possibility would be to use a single step-down transformer in the power supply (I was originally going to use a linear one big enough to handle the 20+ volt drop) so that I could set VCC = MT1 and use a simple pull-down to trigger, e.g. with a MOSFET or darlington array. An added benefit is that TRIACs are easier to drive in the negative-gate quadrants. I've been meaning to learn how to hand-wind transformers and this would be an easy exercise. Hmm... maybe time to take this one off the shelf.

Or I could just use relays! But as great as relays are, that wouldn't be any fun.

[daedalus]

If it were me I would bite the bullet, spend the $140, and buy a wireless thermostat/receiver off of ebay. Its then a 5 min job to fit, rather then days of research, design and production, testing for a one off board that you wont see a dime back for.

[IanB]

Sorry for not bringing any contribution, but, does anybody have a link to some webpage where I can find information on 24V systems. They seem to be very interesting.

It's the standard thermostat wiring system for domestic central heating/cooling systems in the USA. From my somewhat sketchy memory, several wires are brought out to the thermostat. There is a 24 V AC supply, a ground, and various control wires for heating, cooling and circulation fan. I think the thermostat controls each item by switching the 24 V onto the relevant control wire to energize it. Google should turn up lots of info.

(Oddly and annoyingly for my thermostat it doesn't derive power for the electronics and backlight from the 24 V power wire. Rather it uses a separate and independent battery supply. There may be a conservative design reason for this, but it is frustrating having to replace batteries when there is a permanent power feed available inside the thermostat.)

[cybergibbons]

Just a warning for anyone in the UK - our controls are most often at 240V, whereas a lot of the rest of the world sticks to ELV.

[ejeffrey]

(Oddly and annoyingly for my thermostat it doesn't derive power for the electronics and backlight from the 24 V power wire. Rather it uses a separate and independent battery supply. There may be a conservative design reason for this, but it is frustrating having to replace batteries when there is a permanent power feed available inside the thermostat.)

I think some systems don't have dedicated power and ground wires: just a pair of contacts that need to be open or shorted to complete a circuit to energize a relay.  This makes it annoying although not impossible to draw power from the wires.

[ivan747]

It's the standard thermostat wiring system for domestic central heating/cooling systems in the USA. From my somewhat sketchy memory, several wires are brought out to the thermostat. There is a 24 V AC supply, a ground, and various control wires for heating, cooling and circulation fan. I think the thermostat controls each item by switching the 24 V onto the relevant control wire to energize it. Google should turn up lots of info.

Only switching? You mean switching on and off slowly like a laundry iron? I have a TRIAC laying around, so I will try to do something with it. Lastly I have found I am starting to need a function generator, an iPod Touch and a frequency generator app are not enough.
I will try to find more information on Google, but I asked here because I had a hard time finding anything useful, like an application note.

[IanB]

Only switching? You mean switching on and off slowly like a laundry iron?

Yes, that is completely standard for home heating systems. The thermostat switches the heating on when the house (room) is too cold, and switches it off when it is too hot. Since a typical furnace or boiler can only run at full power this is the only option. Area heating (or cooling with central A/C) typically cycles with a 2 degree dead band. I find the cycling somewhat annoying in my house as it happens, but there is no better option using the installed heating system I have.

I am surprised at your question, actually, as home heating systems the world over work the same way using a thermostat like this. Is there something else behind your question? You can't use a TRIAC to do anything clever here, you can only turn the system on and off the way it was designed. If you try to replace a proper thermostat with a device of your own construction you run the risk of breaking something expensive. I would not advise it.

[Uncle Vernon]

It's the standard thermostat wiring system for domestic central heating/cooling systems in the USA. From my somewhat sketchy memory, several wires are brought out to the thermostat. There is a 24 V AC supply, a ground, and various control wires for heating, cooling and circulation fan. I think the thermostat controls each item by switching the 24 V onto the relevant control wire to energize it. Google should turn up lots of info.

If there is such a thing as standard wiring! I think you will find it will be mostly vendor standards. Most older basic (ie: domestic) units will have mains rated controls. Why add a controls transformer when you don't have pay for one. But you will find 24VAC, 24VDC 32VAC and all mannor of other proprietary setups.

Cost is the ruling factor in domestic setups. So wiring was Common + Heat and Cool usually what else do you need for a bimetallic or capillary stat.  Some may have a fan wire  some extra stages.

Some of the older Honeywell stats had a heat anticipator, so you may find a 24V or 240V Neutral wired to these. Sensors with a control switch and indicator lamps may also have a neutral for the lamps.

Modern stuff anthing goes, with all manner of proprietary wiring and bussed comms. Interfacing to mass produced low end splits is a nightmare unless you are sure of specific models

That annoying battery was just the lowest cost way of offering a universal controller. And also a way of ensuring Clocks etc could run when the unit shuts down.  The common active in a great many cases will be switched via an indoor fan relay and will only have juice on it when the fan is operating.

There are a lot of accessory devices on the market that can utilise the existing cable core and bus more signals across existing wiring. But you are going to have to modify wiring at the Stat location and within the unit.

Anything can be done, nothing complex about it, but within a rental the OP is going to be on a hiding to nothing trying to realise any payback or energy saving.
 

[Uncle Vernon]

If you try to replace a proper thermostat with a device of your own construction you run the risk of breaking something expensive. I would not advise it.

As too many have found out the hard way. Safety controls in anything but brand new equipment is usually rudimentary.  When commercial automation was introduced and older pneumatic and electro mechanical controls were replaced the Digital Control, (or more correctly bad application of digital control) was responsible for a lot of damage.

Boilers, chillers etc don't like to be started and stopped too often. With a primitive bimetallic thermostat that didn't happen. Replace that stat with an Arduino or other good idea can cause a mess of trouble when it starts and stops 10 times a minute.
I've seen 500 Ton chillers destroyed by rapid switch. I've seen heating relay fused and contactors blown off the wall by too many starts/hour.  The thermostat is a primitive device, selecting the correct one isn't always child's play.  I'll back IanB's warning 100%!!

[IanB]

If there is such a thing as standard wiring! I think you will find it will be mostly vendor standards.

In the USA I am really sure it is a standard system. If you go into the hardware/home improvement store the shelf is full of thermostats from all sorts of vendors, Honeywell, Invensys, etc., and they are all drop-in replaceable with each other. There is a standard wiring scheme with colour codes on the wires.

The first thing I did when I moved into my house was purchase an electronic day/week timer thermostat to replace the mechanical mercury switch thermostat the previous owners had left behind. It just dropped right in, connecting the right wires to the right terminals, and it controls the heating and cooling fine.

It is the existence of the standard wiring scheme for domestic central A/C systems that allows the State of California to consider offering a special wireless controlled thermostat that can turn off your air con by remote command in times of high demand. How's that for a tough sell?