LED Dimmer fires when connecting load

Hello,
I'm trying to design light dimmer for driving the LED bulbs.
Initially design was based on standard triac circuit with optocoupler. Unfortunately that solution wasn't especially good for driving low power LED bulbs at small PWM values.
I have searched internet and I found how to make alternative circuit with galvanic isolation provided by transformer. Circuit from the attachment has been built and tested. Every thing works ok, starting from very low PWM values to 100%.
Unfortunately now I have to face up different problem. From time to time, during connecting the load, I can observe that triac occasionally fires and blink could be observed.
I have tried triac which are not EMI sensitive like BTA312-800C, but it does not change anything.
Any ideas what can I improve and why triac fires?



Best regards
Krzysztof

The triac firing by itself IS NOT the FAULT.

The problem is the snubber. The snubber has an impedance which allows leakage current.

Said leakage current slowly charges the Led ballast's bulk electrolytic capacitor. When its voltage reaches the threshold voltage required to fire the led driver, it will start.
But the leakage current is insufficient to maintain the driver powered up and the voltage collapses.
The cycle repeats itself again and again.

In essence, you have created a relaxation oscillator.

Fortunately, the cure is simple: remove the snubber.

schmitt trigger said:

The triac firing by itself IS NOT the FAULT.

The problem is the snubber. The snubber has an impedance which allows leakage current.

Said leakage current slowly charges the Led ballast's bulk electrolytic capacitor. When its voltage reaches the threshold voltage required to fire the led driver, it will start.
But the leakage current is insufficient to maintain the driver powered up and the voltage collapses.
The cycle repeats itself again and again.

In essence, you have created a relaxation oscillator.

Fortunately, the cure is simple: remove the snubber.

Click to expand...

Thanks for your reply.
Unfortunately your suspicion cant be true.
1) Bulb blink only ones when is connected (There are no oscillations)
2) The same (but less often) can be observed on the traditional incandescent lamp.

My suspicion are now around dv/dt problems. When load is disconnected triac has high impedance but much lower in compare to the open circuit on the load connector. When load is connected then situation is opposite: Triac has small leakage current and load has very low impedance that why triac voltage rises from 0V to mains peak value in very short time.
What I do not understand is that the BTA312-800C should be robust for high dV/dt values especially with additional snubber circuit.
I will make measurements and post some results today.

It seems a bit unusual that you didn't provide any series resistance from transformer to the TRIAC. This sounds like making the thyristor more prone to trigger in presence of the smallest stimulus, from wherever.

You know, Andre is actually correct. The diode should not be there, as it effectively leaves the gate floating.

Besides, Triacs have bilateral triggering. The diode is not necessary.

Remove the diode and put a 47 ohm resistor instead.

schmitt trigger said:

The diode should not be there.

Click to expand...

In fact, in having no return path at the secundary side, seems like the tronsformer core would keep permanently a remanescent magnetism with the same polarity, quickly pushing it to the saturation region.

After in depth investigation, the RC snubber circuit has been found as one of cause that triac is fires sporadically when load is connected.
On load connection the contacts in the socket are “Bouncing” what in some simplification we can consider as switching load “on” and “off”. When it’s “on” the snubber circuit is charging and remains charged when load is disconnected. When load is reconnected triac is polarised with snubber voltage which need to be reverted by mains voltage. When disconnection time is long and mains voltage swings to opposite voltage, triac gets voltage shock and dV/dt can reach 800V/us (according to my observation.)
After removing snubber, triac fires less often but still it happens…

Besides, Triacs have bilateral triggering. The diode is not necessary.
Remove the diode and put a 47 ohm resistor instead.

Click to expand...

In the document "TRIAC CONTROL BY PULSE TRANSFORMER" AN436 provided by ST, diode is listed as protection against false triggering due to voltage which is present when triac is conducting.

In fact, in having no return path at the secundary side, seems like the tronsformer core would keep permanently a remanescent magnetism with the same polarity, quickly pushing it to the saturation region.

Click to expand...

I can't agree because transformer works here in fly-back topology.

Summarising: after removing snubber circuit problem is less visible .. but is still there and now I have feeling (after observations with oscilloscope) that is not connected with dV/dt any more.