Triac blowing

[okw]

I have a circuit that auto-fills a water tank to around 50% level (+/- a hysteresis band) based on an immersion sensor and some logic (not important for the question).
When auto-filling (low water level detected), a solenoid valve opens (energized) to direct the water into the tank, while the pump is running. When level is restored, both pump and valve is de-energized.
The user can also manually power the pump via a switch to dispense water to an outlet. In this case the solenoid valve is closed so the water flow takes direction to the outlet (not the tank).
Where the water goes, doesn't really matter to the question.

I've produced the PCB and tested it many times (2 PCBs have been on for a long long time with extensive manual pump engaging), without problems. But the other day, a triac on one of the PCBs exploded. It happened when the manual pump was already engaged, then the auto-fill triac mechanism kicked in and also powered the pump.
The question is, can it pose a problem, that the triac can engage the pump while the user has it manually engaged?

I've attached a the circuit (no logic, just the 250V stuff).

[Phil1977]

As far as I read your schematics, you completely short-circuit the mains voltage part of your triac-circuit when the manual pump switch is engaged.

Did you accidentely connect the switch and the triac to different phases? As long as "250V-AC-L" is the same on both sides of the circuit I can really see no way to blow the triac when the switch is closed.

[bingo600]

I'm not an EE , and no Triac guru

But could it be there is a slight delay in the Triac, that would create a phase difference vs the "Real mains phase".
Then when opening it via the "Auto" , the Triac output tries to "fight/adjust" the Mains Sine comming from the manual contact ??

Ohh this is my 2K post 

[schmitt trigger]

When the manual switch pump is closed, there must be exactly zero volts across the Triac.
This got me thinking, could it be that the switch was “chattering”? This could have produced a large inductive kickback.
I ignore what type of switch you are using, but most simple switches are not designed to switch motors on and off, and can become flaky.
What I would do:
-Make sure the switch is actually rated for a motor load.
-Add a 264V (withstanding) MOV across the Triac.
-The cherry on the top is to also add a snubber to the photoTriac. Data sheets show you how this should be done properly.
-Add a fuse. If its I2t is properly coordinated with the motor’s startup current and triac’s rating, it can also protect the triac.

This last recommendation cured issues for a Triac based transformer tap changer.

[okw]

Did you accidentely connect the switch and the triac to different phases? As long as "250V-AC-L" is the same on both sides of the circuit I can really see no way to blow the triac when the switch is closed.

I'm quite sure I didn't as all the modules (solenoid, pump and heating element) all are connected permanently to neutral, and needs line for power. If I switched N/L on the PCB, it wouldn't switch on any of the three.

[okw]

I'm not an EE , and no Triac guru

But could it be there is a slight delay in the Triac, that would create a phase difference vs the "Real mains phase".
Then when opening it via the "Auto" , the Triac output tries to "fight/adjust" the Mains Sine comming from the manual contact ??

Ohh this is my 2K post 

Uh, congratulations!

Yes, that's what I was thinking too.

[okw]

-Make sure the switch is actually rated for a motor load.

It's rated correctly for this motor.

-Add a 275V (withstanding) MOV across the Triac.

Will do

-The cherry on the top is to also add a snubber to the photoTriac. Data sheets show you how this should be done properly.

I already have a snubber, R5 (39R) / CX1 (10nF).
Or did you have anything else in mind?

[Phil1977]

Did you accidentely connect the switch and the triac to different phases? As long as "250V-AC-L" is the same on both sides of the circuit I can really see no way to blow the triac when the switch is closed.

I'm quite sure I didn't as all the modules (solenoid, pump and heating element) all are connected permanently to neutral, and needs line for power. If I switched N/L on the PCB, it wouldn't switch on any of the three.

And is there only one "L" supplying the whole system?

The MOV is definitely a good idea.

[okw]

Yes, L and N is the same for all.

[schmitt trigger]

That snubber is for the Triac. Check figure 6 on Texas Instruments’ MOC3020…MOC3023 data sheet to see how it is done.

Speaking of data sheets, I downloaded the Z0103N data sheet. It is a very small triac! Replace it with something beefier, at least 5 amps.

[okw]

That snubber is for the Triac. Check figure 6 on Texas Instruments’ MOC3020…MOC3023 data sheet to see how it is done.

Oh yes, I'll add that. Do i divide R4 in two (360R => 2x 180R)? And what capacitor value? And is X2 required?

Speaking of data sheets, I downloaded the Z0103N data sheet. It is a very small triac! Replace it with something beefier, at least 5 amps.

It's switching a 0.2A pump (48W 250V). Wouldn't 1A be more than good enough?

[schmitt trigger]

The 0.2 amp is running. What about starting? This would be several times the steady state value. Give yourself overload margin.
Plus, if you add a fuse, the Triac itself must have a larger I2t than the fuse. The device you have has only 0.35  amp2s
The Renesas BCR3AN-14 has a steady state current of 3 amps but 3.7 amps2s

[temperance]

-R6 connecting to the base of OK1 is not correct.
-The mechanical switch has extremely short opening and closing times and as such you will have very high spikes when operating inductive loads. The snubber circuit might not be correct to fully dampen the over voltage spike.
-R6 might as well be broken because it will probably not withstand any over voltage or even just regular spikes on the mains grid. This requires wire wound or pulse rated resistors. But this simple circuit is not very robust. (no protection for over voltage for the opto, triggering problems,...)

A proper circuit is shown on page 6 with some practical values shown on page 22. For R1 you will need a low wattage wire wound resistor or some other type able to withstand surges.

https://www.st.com/resource/en/application_note/dm00451014-controlling-a-triac-with-a-phototriac-stmicroelectronics.pdf

[moffy]

Put a snubber across the motor, when the Triac switches off it naturally turns off at zero current, but the manual switch when it turns of can still be carrying a substantial current through the motor, all that inductive energy will find somewhere to go, in your case blowing the Triac. You need to provide a safe alternative path for that energy to discharge.

P.S. You could use a switch to 0V through its own 560R resistor to the LED Pin 2 of the MOC3083 to switch the mains, that way you would avoid the issue, not needing the mains switch, as long as that wouldn't interfere with the drive from 'AUTO_FILL_TANK'.

[okw]

Put a snubber across the motor, when the Triac switches off it naturally turns off at zero current, but the manual switch when it turns of can still be carrying a substantial current through the motor, all that inductive energy will find somewhere to go, in your case blowing the Triac. You need to provide a safe alternative path for that energy to discharge.

How do I size the C and R?
And would it be best practice to place the snubber locally on the PCB (easier for manufacturing and assembly), or near the motor itself?

P.S. You could use a switch to 0V through its own 560R resistor to the LED Pin 2 of the MOC3083 to switch the mains, that way you would avoid the issue, not needing the mains switch, as long as that wouldn't interfere with the drive from 'AUTO_FILL_TANK'.

I was thinking about it, but it would require a lot of re-wiring. I planned to use this as a drop in replacement. But if I en up doing it, I could just put an opamp buffer between the switch and AUTO_FILL_TANK signal to not interfere.

[wraper]

It's switching a 0.2A pump (48W 250V). Wouldn't 1A be more than good enough?

It's not a resistor, did you check an actual inrush current? Not to say just an RC snubber circuit is probably inadequate for an inductive load like this.

[moffy]

Put a snubber across the motor, when the Triac switches off it naturally turns off at zero current, but the manual switch when it turns of can still be carrying a substantial current through the motor, all that inductive energy will find somewhere to go, in your case blowing the Triac. You need to provide a safe alternative path for that energy to discharge.

How do I size the C and R?
And would it be best practice to place the snubber locally on the PCB (easier for manufacturing and assembly), or near the motor itself?

P.S. You could use a switch to 0V through its own 560R resistor to the LED Pin 2 of the MOC3083 to switch the mains, that way you would avoid the issue, not needing the mains switch, as long as that wouldn't interfere with the drive from 'AUTO_FILL_TANK'.

I was thinking about it, but it would require a lot of re-wiring. I planned to use this as a drop in replacement. But if I en up doing it, I could just put an opamp buffer between the switch and AUTO_FILL_TANK signal to not interfere.

I cannot say what the RC values should be as I have no idea of the worst case energy stored. It really makes much more sense to use a triac as the switching element.

[okw]

It's not a resistor, did you check an actual inrush current? Not to say just an RC snubber circuit is probably inadequate for an inductive load like this.

I did not, I'll do it now.
It's an Ulka vibration pump.

[wraper]

Also R4/R3 resistance is way too low for a such high sensitivity triac.

[okw]

Also R4/R3 resistance is way too low for a such high sensitivity triac.

I plan to replace the triac with a BT134-800D, but I don't have it in my stock at the moment. Sensitivity goes from 3 / 5mA (I-II-III / IV) to 5 / 10mA.
Would an even less sensitive triac (e.g. BT134-800E, 10 / 25mA) be beneficial in this kind of circuit?

And how do I calculate the R3/R4?

[NiHaoMike]

Would an even less sensitive triac (e.g. BT134-800E, 10 / 25mA) be beneficial in this kind of circuit?

Only by being less likely to get triggered by EMI. Granted, that's unlikely to be a problem with reasonable care taken during layout.

[okw]

Finally got it sorted. The two pump systems that had been working for weeks was made by another company. Control boxes are almost identical, but apparently not. A small rewiring, and the triac doesn't blow.

As a precaution I added a 100nF/100R snubber over the pump (in addition to the snubber already over the triac).
Could this cause problems, when I don't have numbers for the pump and just ballparking the RC?

And for measuring the inrush current, and the induced dump, would a Fluke 183 set to fast min/max (1ms peak, 5000 counts) do the job? Or could the spikes be more narrow?

[schmitt trigger]

The inrush current period on a small motor driving a low inertia load might be in the vicinity of 200 ms. Should be plenty for most meters with a MAX function to capture it.

Of course, larger motors driving large inertia loads may take several seconds.

[floobydust]

There is no motor as I understand it    OP says it's a vibrating solenoid pump, made by Ulka. Common in espresso machines etc.
These need a diode or two (series and across coil) i.e. 1N4007 which can be inside the coil or external. They get 1/2 wave AC and max. duty cycle 1min on/1min off they do heat up.
I have no idea how they behave electrically as the pintle moves. Current should be impedance limited?

OP ensure that you have a small fuse. The AUTO/MAN switch can be on the low voltage side, turning on the opto/triac. I think R3 triac gate is too low, it's usually around 300-1k ohms.
The Z0103 triac is pretty small and won't take any drama, they usually use BT136 TO-220 parts in coffee machines.
A solenoid valve is also an inductive load and will spark across relay contacts, can cause an MCU to crash, so I would have an RC snubber for that.