Triac with capacitive loads

[whiskeyjack]

I have a circuit where I am using triacs to switch home appliances like fan, tube light etc. This is the circuit I am using:


(TRIAC1, TRIAC2, TRIAC3 come from arduino GPIO pins)

First load is a fan with a capacitor based regulator. (Sorry for showing fan and tubelight as a resistor). Speed is controlled by rotating a knob which changes the series capacitance. Lower capacitance means lower fan speed.

As seen from the circuit, I didn't use a snubber because I thought using a snubberless triac will do the job, however the following incident made me rethink about the situation:

1) When Bulb and tubelight are off and fan is ON, and I try to regulate the fan speed by rotating the knob, sometimes there is a flicker in the tubelight and bulb. Flicker is very short lived but still visible. Flicker doesn't happen at higher speed. It usually happens at lower speed levels. For ex - There are these levels: OFF, 1, 2, 3, 4, 5 (5 means direct connection to Live). When I switch speed from 2 to 1 (or vice versa), I see a short lived flicker in the tubelight.

2) If the fan is running at low speeds - 1 and 2, and I try to turn the fan OFF by pulling the GPIO (TRIAC1) LOW, the fan won't turn OFF. Instead it produces humming sound and continues to rotate at a lower speed which indicates partial turn ON of triac.

Here is the fan regulator I am talking about:


Putting a RC snubber on triac 1 didn't solve the issue. Moreover I am using snubberless triac so I think it's not the inductive load that is giving me issues. The capacitor based regulator is the culprit here. Maybe LC circuit is going in resonance when regulator is set at low speeds and that resonance is giving me these issues.

If anyone could suggest something to get rid of this issue, I'll be very thankful.

[anurag102]

Hi WJ,

I was facing the same problem today... was wondering if you found a way out ??

[oldway]

Indeed, the BTB16-600BW is a snubberless 3 quadrants triac but the driver triac of the MOC3021 is not "snubberless".

This is the reason of yours problems.

The triac of the MOC3021 must be protected against high dV/dt by a snubber as described in the datasheet of the MOC3021. (see fig 7 inductive load circuit).

http://www.ti.com/lit/ds/symlink/moc3020.pdf

[Benta]

Three different Triacs controlling three different loads, and one load influences the other two?
It's got nothing to do with snubbers, that's for certain.

My suspicion is a bad Neutral in your house wiring.

[oldway]

Some explanations

There are 4 ways to make a triac conduct:
1) gate pulses
2) over voltage (here greater than 600V peak)
3) high dV/dt (the snubberless triacs can withstand at dV/dt higher than 500V/µs)
4) overheating, at very high temperature, the junction loose control.
The only possibility seems to be the first one.
And the gate signal is coming from the auxiliary triac of the MOC3021 which is not "snubberless"
The solution is then to add a snubber to this auxiliary triac.

[Benta]

@oldway:
try to read to original post. Thank You.

[oldway]

@oldway:
try to read to original post. Thank You.

Would you mind to post something more constructive ?
Do you think I can't read or that I am too lazy to answer without reading the original post ?
What's wrong with my answer? 

[SeanB]

Correct to add the snubbers to both the triacs and the gate driver. the trigger voltages are coming from the mains wiring having some intrinsic inductance, capacitance and this causing a voltage drop on triggering, causing the other ones to switch on ( remember dVdt is only rate of change, and turning on the one triac causes a pretty big pulse as the inductance of the wiring in the supply makes this voltage drop), so you do need the snubbers. Adding some overvoltage protection as well is also a good idea, and for the fan motor a series inductor to reduce the current spikes as well. The worst loads to switch with a triac are pure inductance and pure capacitance, as the turn off is very difficult without snubbing, and using high dVdt triacs as well.

As to the OP, the fan motor does not use a capacitor to control speed, it uses a tapped transformer, to switch in what basically is a lower supply voltage as you want to drop speed, so the motor slip increases and thus the lower speed. Better fans are wired differently, the speed changes are done by having multiple poles, and the windings are arranged so that you change the number of poles of the motor. high speed might be 4 poles, medium 6 poles and low 8 poles, all using the same rotor and stator windings, just having the windings of the coils spread differently on there.

Adds to cost a lot, as you need to have room for all 3 sets of windings, and a compromise on the start winding for efficiency, but the slip is constant, the losses are low ( because of the large amount of iron in the core) in the core and copper losses dominate. However most are wound with a constant pole pitch, or only 2 pitches, so there are more losses, but there is less copper wire ( or in cheap ones copper coated aluminium now) and iron core needed, and it will be running a lot hotter.

[oldway]

Did you checked with an oscilloscope you don't have spurious out of control pulses coming from your Arduino when such interferences occurs ?
This could happen if you have bad sincronisation signal or bad power supply of your Arduino.

[wraper]

First of all add snubbers and get rid of MOC302x. They are not well suited for 220-240V (only rated for 400V peak), especially for loads which are anything but resistive. Seen quiet a few devices which had troubles because of them. Even if they work fine at the beginning, they tend to degrade over time and start false triggering. Use MOC3051- MOC3053 instead. http://www.farnell.com/datasheets/301912.pdf
220R resistors are a bit too low as well (have seen too low resistance a lot of devices, which caused a lot of optocoupler failures). Peak current through the MOC302x must never exceed 1.2A, but with 220 ohm it can be up to about 1.4A if switched during the peak mains voltage.

[wraper]

@oldway:
try to read to original post. Thank You.

Try to not write useless posts, thank you. oldway explains the issue completely correctly.

[whiskeyjack]

I have an update and I apologize for posting it so late. The issue was majorly due to the false triggering of my optocoupler (MOC3021). Due to PCB size constraints, I was unable to use any sort of snubber. Finally I found a better triac driver - FOD420 which seems to solve the issue. However in 1% of the cases, I still see a small problem where if I turn on a highly inductive load like a mixer grinder or some fans, the lights still flicker.

[floobydust]

I think you're missing a gate resistor, 330R-1k like in the MOC3041.  I find it reduces false triggering due to dV/dt.

[Zero999]

I have an update and I apologize for posting it so late. The issue was majorly due to the false triggering of my optocoupler (MOC3021). Due to PCB size constraints, I was unable to use any sort of snubber. Finally I found a better triac driver - FOD420 which seems to solve the issue. However in 1% of the cases, I still see a small problem where if I turn on a highly inductive load like a mixer grinder or some fans, the lights still flicker.

I hope you've allowed adequate creapage and clearances between the AC mains and DC logic sides. If not, it's potentially very dangerous.