Triac not shutting off (24v AC sprinkler valve control)

Question

Hi All,

I'm working on replacing my aging and broken sprinkler controller, and found some open source software (OpenSprinkler) to run on a Raspberry Pi that handles the scheduling very nicely (plus it has an app for my phone).

With that software I'm able to control a couple of shift registers to turn LEDs on and off, simulating the sprinkling zones.

I'm not very familiar with TRIACs (read: not at all familiar, never used them before ), but recently I learned that the sprinkler valves use 24v AC and that TRIACs would be a good way to switch them, with relays generally being rather big once you get 16 of them lined up, and regular power transistors only being good for DC.

I bought some BRT12H "non-zero crossing" opto-isolating TRIACs (datasheet: http://www.vishay.com/docs/83689/83689.pdf), because they seemed to meet the desirements of opto-isolating and AC for 250mA.

I hooked them up (with 220ohm resistors from shift registers to TRIACs), and it all looked promising when I turned the first zone on (sprinklers worked, yay!), but then it wouldn't shut off (sprinklers still working, not yay!). It only shuts off if I remove the 24v power momentarily.

I've done some Googling around but have not yet found any solution, and to be honest it's not making a whole lot of sense to me (yet... hopefully yet).

I did find out that TRIACs (in general?) need to cross zero (volts?) to turn off. I'm not sure what this "non-zero crossing" bit means but I have a nagging suspicion that it's a key part of the issue.

Should I have bought "zero crossing" TRIACs instead?

Can anyone help shed some light? ...and hopefully also shed a solution

ps, I did see that I can buy a pre-built board for not too expensive (about $80 after shipping and taxes), but it's a cool DIY project and it would be far more satisfying (and educational) to build it myself.

edit: While doing more searching for answers I noticed that SSRs (solid state relays) do exist that are tiny chips, like this one, DIP-8: http://www.sharpsma.com/webfm_send/335 - very cool.

Thanks!

-Nico

mcb1 · Accepted Answer

In time gone by we used to make a board that controlled Stage lighting. The pinspots were a large 24v transformer and a 6v lattern lamp, so were heavily inductive. Most designs take the phase and feed it through a resistor and opto, then into the gate. We found this caused lots of problems with inductive loads. We copied some piece of commercial gear that connected the opto/resistor between the gate and the load. Fixed all our issues ..... Hopefully this will fix your issues. Mark

jw0752 · Answer

Hi Nico, AC solenoids will convert to DC solenoids quite easily if you are willing to reduce the voltage. What will remain constant for proper operation is the wattage of the solenoid. For example your solenoids appear to be about 6 Watts. 24 V AC * 250 mA. As you have stated the coil resistance is 32 ohms so the DC voltage at which 6 watts will be produced in this resistance is V = SQRT(32 Ohms * 6 Watts) = SQRT( 192) = 13.9 Volts. In reality your solenoids will likely energize at 9-10 Volts. You can experiment with this and of course the new lower voltage across the 32 Ohms will produce a current of about 430 mA. You can experiment with your solenoids to see how this works out. I would probably try 12 volts and see how well the solenoids respond. Be certain to test them under pressure as the plunger has to be pulled, not only against the force of the spring but also against the force created by the pressure of the water over the area of the opening in the valve seat. DC solenoids do not convert to AC solenoids as easily. The plunger of an AC solenoid is especially designed not to react to the fluctuating magnetic field. The plunger of a DC solenoid on the other hand will react and you will hear a buzz that will be the plunger chattering against its stop. Over time this produces metal wear and failure. It is also quite annoying to listen to. You can relate this to the way AC and DC relays act when subjected to voltage for which they are not designed. For the fun of it I ran an experiment with a 120 VAC solenoid I had here in the shop and I was able to convert it to proper operation using 40 Volts DC at 190mA where it originally was 120 Volts @ 70 mA. John

D_Hersey · Answer

You are using LV, so it probably won't matter as much, but for line potentials, only certain types of Cs are a good choice in this rough service, metallized polyester being one.

jc2048 · Answer

"And what is the best way to figure out the values?"

Iteration. Increase the capacitor value and decrease the resistor value a bit and try again. But when you get down to a resistor value of 100 ohms and up to a capacitance value of 100nF, stop and have a think about whether the problem might be something else.

What's going on is quite complicated and involves the coil, the snubber components and the triac. Modelling that lot is complicated. Changing a couple of values on a breadboard is easy.

Could you tell us what the inductance and resistance of the solenoid are, as measured by your meter?

jc2048 · Answer

The simulator is TI-TINA. It's a cut-down version of a commercial offering which is given away for free by Texas Instruments. http://www.ti.com/tool/tina-ti The schematic capture is irritating - you draw the nets on, so it's very easy to get nets that don't join or bits that sit on top of each other. The simulation warns you if there are dangling connections, but you still have to go back and repair something the software should just get right. There is a reasonable range of library parts built in, which is good. It's good because importing component models is a bit of a pain [that's the case with any simulator - it isn't a deficiency of this particlar one]. You could also try Linear Technology's offering http://www.linear.com/designtools/software/ I tried this a long time back and it was very biassed to Linear's own devices [understandably], but perfectly useable. I really ought to give it another go because I'm sure they'll have developed it further in the meantime. [TI's offering is also biassed, but since their catalogue includes so many generic parts you don't really notice.] Since they are free, you could install both and see which you like using. I was being stupid about the solenoid; of course the inductance changes as the plunger moves because the magnetic circuit changes. Don't even know where you'd start to model that, though.

D_Hersey · Answer

A dual-SCR output SSR would be a little bit better than a TRIAC-output SSR for retriggering, but we often have to snub or anti-spike SSR's as well. You might use quencharcs.

D_Hersey · Answer

http://www.onsemi.com/pub_link/Collateral/AN1048-D.PDF

D_Hersey · Answer

I am, as usual, in consonance with MK, I think you would be well served to slow down and do some more reading, the information is of fairly general relevance. The snubber works equally well placed across the switch or placed across the load. But it is dissipative. If your duty cycle is <50%, as in your case, you want to shunt the load, for sake of efficiency. Sometimes, shunting your snubber with a MOV can let you use a higher-Z snubber, it is useful to experiment.

D_Hersey · Answer

Sometimes, snubbers are needed with relays as well. The inductive spike can pit or weld the contacts.

D_Hersey · Answer

The only relays that work on an arbitrarily inductive load are the mercury-wetted types. Thyratrons can handle fairly large inductances.

mcb1 · Answer

Should I try with a secondary triac? Most of the applications I used for 10A max, so the triac was essential. I did find this in the datasheet. Personally I'd add one. Mark

mcb1 · Answer

Personally I don't think the snubber is going to resolve your issues.

The load size is not enough to generate large spikes that will overwhelm or damage the switching components.

It is also not likely to cause EMC, as you're using solid state devices, rather than contacts.

The frequency of switching is not high so again unlikely to cause EMC.

One of our members of another club has been killing solenoids, and it seems it is due to DC.

However he's switching AC so I'm not 100% on how he's getting a DC component in the solenoids.

Mark

shabaz · Answer

Hi Nico, I hope you're well! I saw this topic late, but just wondering, it could be easier to control the low-voltage sprinkler from a DC supply than AC (i.e. diode bridge and capacitor, and then a MOSFET circuit or similar if you're looking for electronic control). I did look at the URL regarding sprinklers and it goes on about inductance, but I think they're over-thinking it. With an AC supply the valve will use a certain amount of real power. Provided the DC supply provides that too, then there is no issue with overheating or reduced life, etc. If you have access to a variable/bench DC supply, an option could be to increase the voltage until it functions reliably, and doesn't get hot. I'm not knowledgeable on valves though, so maybe I've missed something. But could be worth a try with DC just to confirm operation.

mcb1 · Answer

I was wondering what the purpose is of the two 1n4148 diodes? You aren't alone in that. It would stop voltage less than 0.6v from passing, and since it is AC it requires back to back, but other than that I'm not sure. Since it worked in a piece of commercial equipment, we simply copied it and I've used it ever since. Mark

michaelkellett · Answer

That's a good app note from Motorola/On-semi - by the time you've read and digested it you'll know lots of good stuff and you'll feel better for it !! I had an issue recently with a current source that was fine with resistive loads but didn't like inductors - fortunately the inductive load was always the same so I put a series R and C in //, not just to snub but to cancel out the inductance completely. (Google "Zobel network"). MK

jw0752 · Answer

Hi Nico, I tend towards the practical and the numbers you have suggested sound very close to what I have seen used in similar circumstances. It can't hurt anything to pop the R - C snubber in the circuit and see if it solves the problem. John

jc2048 · Answer

Now I'm confused.

Here's a very simple simulation. Voltage source (24Vac) driving a coil with the values you just gave.

For 24Vac rms, the peak voltage is 33.94V. The simulation gives the current as 798mA peak (which is 564mA rms) - it's a bit less than you would expect from the dc resistance because the inductance is already having an effect, even at 60Hz. So where did the figure of 250mA come from? Am I missing something? Or have I got it wrong? [Might have done - it is the end of a long day.]

Can you do an experiment?

Connect the solenoid to the transformer. Measure the voltage across it. Then put the meter in series with it and measure the current. We can then see how well it matches the simulation. If it does then your 300mA triac probably isn't the part to be using.