I have built a circuit according to Dimming 230V AC with Arduino, with two output channels. The basic principle is:

  • An ATtiny does zero detection via an optocoupler; when AC crosses 0 V, it triggers an interrupt.
  • After a certain delay (e.g. pi/2), the ATtiny fires the TRIAC for phase cutting.

The Triac is a BT136-600E with an IRMS = 4 A and Ipeak = 25 A.

We have connected four (not seven, as originally written) 35 W halogen lamps to each output. This is about 500 mA when they are all on.

When testing it with one lamp, the circuit worked as desired. With all 4 lamps, one Triac simply broke apart. Why/how can this happen?


The measurements below are always with a single 28 W halogen bulb.

After the comments, I have built a new board with a BTA316-600D which stands 16 A rms. It is now attached to an Aavid 5297 heatsink which can dissipate 14 W of heat at ambient temperature of 45 °C, which is the maximum power dissipation of the BTA316 at 12 A.

The halogen lamps have a cold filament resistance of 136 Ω (measured with the Ohm meter), which would result in 1.7 A (rms) flowing through the filament of each lamp while it is cold, as opposed to 120 mA when it is hot.

To measure how much current is really flowing and when, I started with a 0.5 Ω 3 W shunt in series to the halogen lamp. The current flowing through the halogen lamp would then be calculated by the voltage drop across the shunt resistor.

Trying to measure the voltage drop with a Picoscope failed because, as I then learned, the ground reference of the scope probe is connected to the case ground and thus also to the USB wire ground and my Laptop ground and finally the power plug ground, which triggered the RCD fast enough to prevent additional harm.

The next attempt was to use one probe on each side of the resitor, this time with a Rigol DS1054Z which can measure 230 V AC. However, the measurements turned out to be too imprecise and too low-res to measure mV differences at the 230 V range. The delay of the voltage difference is a bit odd too.

Triac measurements

I had the opportunity to get some measurements with a current clamp (yellow) and a differencial probe (green) connected to a Keysight 4000 X. Here, the measurements while the dimmer was active were as expected. The current stayed below 1 A while ramping up. However, when just turning on AC power (with a power cable with built-in switch, as used for turning lamps on and off), we measured peaks up to 10 A during < 200 ns.

Triac measurements

While this gives some insight into what is happening, it does not answer the question for me. The current flowing through the Triac while ramping up is relatively low. The peak when turning on AC is weird (why does it occur?), but very short as well.

This is the schema:

Dimmer schema


Tricac split in two parts

A close-up of the triac, the two parts bent apart for better visibility.

enter image description here

  • 3
    \$\begingroup\$ that breakage is because it exploded. Did you take into account that to achieve that 4A RMS, you need to supply sufficient cooling? What does the data sheet say about thermal resistance (in °C/W)? Estimate how much power is converted to heat within the TRIAC and thus how much hotter the device gets. \$\endgroup\$ Commented Apr 21, 2019 at 9:42
  • 6
    \$\begingroup\$ Your 1 A "all on" current measurement is for hot filaments. Measure the cold filament resistance of a single lamp and add it into your question along with the turn-on current calculation. It might answer the question. \$\endgroup\$
    – Transistor
    Commented Apr 21, 2019 at 9:42
  • 1
    \$\begingroup\$ Can you ramp up the fireing angle slowly? \$\endgroup\$
    – winny
    Commented Apr 21, 2019 at 12:28
  • \$\begingroup\$ @Transistor I will do some measurements as soon as the board is repaired, shunt has arrived already. \$\endgroup\$ Commented Apr 24, 2019 at 16:02
  • 1
    \$\begingroup\$ I meant that you should measure the filament resistance on a cold lamp using your multimeter. Then use that to calculate the switch-on current for one lamp and for seven and see how that matches the triac specifications. \$\endgroup\$
    – Transistor
    Commented Apr 24, 2019 at 16:22


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.