I think my triac batch is defective

Thread Starter

mariosergio

Joined Feb 23, 2018
23
Hello everyone

I've designed a circuit for dimming lamps with a triac (home automation). Everything was great and I also manage to produce a small batch, that got sold and installed at some houses. I always knew my product would cause some lamps to flicker, when operating on very specific conditions. It was not a huge issue because increasing or decreasing the dimm just a little, makes the flicker goes away, and also it only happens with some of the LED lamps, not all of them.
Right now I'm planning a production of a second batch, and I decided to get rid of that problem for once. I've been troubleshooting the circuit for a week, and I'm getting closer..
I notice that the problem only occurs when using triacs of a specific batch. If I switch to another batch of the same product, everything works just fine.

The triac I'm using is the ST BTA16-600BW.

The 'OK' product reads:
BTA16-600BW
GK01P VU
CHN 331

The 'Falt' product reads:
BTA16-600BW
GK04R VU
CHN 940

1) Datasheet doesn't say a thing about the markings below the product code. Does anyone know more about it?
2) Is it possible that the 'OK' product was just made 'stronger', so it resists to some mistake I made in my circuit?
3) Is there a place to check if a batch is known to be 'bad'? Like a recall for electronic components?



Thanks a lot for any help!!
 

Ian0

Joined Aug 7, 2020
10,253
Running LED luminaires from phase-fired dimmers is always going to be problematic.
There is no standard for how an LED luminaire is to perform for it to be classes as “dimmable”, and the phase-fired dimmer predates the LED luminaire by many decades. if there is nothing weird about your dimmer circuit, then I suspect that the problem lies with the LED luminaires. Either people are trying to dim luminaires that are not dimmable, or those that are labelled “dimmable” are perhaps not quite so dimmable as they should be.
I’ve also never known SGS-Thomson make a bad triac: the BW types being particularly robust.
Post your triac circuit, so we can see if there is anything weird about it.
 

Ian0

Joined Aug 7, 2020
10,253
Some lamps are made specifically to allow them to be dimmed with triac dimmers. Such lamps are called “dimmable”.
The problem is - to misquote George Orwell - some are more dimmable than others.
If the output waveform from the dimmer has no jitter on the leading edges, then the fault is in the luminaire.
My guess is that the customer has bought named brand luminaires in fancy boxes, and therefore he believes that they must be perfect and the small company that supplied the control gear must be at fault.
It’s also far easier to go after the small company, because the small company generally gives a damn about whether their product works.
 

Thread Starter

mariosergio

Joined Feb 23, 2018
23
Hi folks, and thanks for the reply!

What bothers me about this, is that I have the circuit on a breadboard, so I can easily change the triac. Any triac from the 'fault' batch makes the lamp flicker, and any other triac works fine. It takes me 1 second to swap the triacs, so absolutely nothing else changes.

The 'fault' triacs work fine for a lot of other light bulbs, so as the 'ok' batch.

Below is the circuit I'm using. The MCU does some other stuff, like receiving commands from serial, but I remove that from the circuit, so we can focus on the 'power' part.

I appreciate any suggestions or recommendations! I'm soon making a new batch (like 50 or so), and any changes must be made now!

Thanks a lot!

Triac.png
 

Ian0

Joined Aug 7, 2020
10,253
It looks OK to me.
What does the output waveform look like?
At what point do you switch off the drive from RC1?
I would recommend that you switch it off no later than 9ms after zero-crossing for a 50Hz supply. Keeping it switched on after that leads to the possibility of the triac remaining switched on into the next cycle.
 

MisterBill2

Joined Jan 23, 2018
19,377
The second line, below the type number, is a production run/date code/manufacturer code. It appears that the production lot is different, and so some of the characteristics may be different. Device properties not described in the data sheet are usually not guaranteed , and so they will probably differ some from batch to batch.
Soyour design may need to be adjusted a bit to make the triacs in that lot work correctly.
 

Thread Starter

mariosergio

Joined Feb 23, 2018
23
There is clearly some difference from one batch to the other.

Below are pictures of the waveform. Red line is the mains AC (60Hz), and Yellow line shows the voltage at the lamp.
The first picture is the 'ok' triac and the second picture is the 'fault' triac.

Arrow labeled 'A' shows a point where the yellow line never attaches to the red line, indicating trick is turning off too early.
Arrow labeled 'B' is the 'flickering' part. Sometimes the yellow line overlaps the red one, and sometimes it looks like in the picture.

I believe that the Triac is turning off too early, but that is all I can conclude from the pictures...

At what point do you switch off the drive from RC1?
I'm controlling the triac using 70uS pulses. So at the right moment I turn on the MOC, and then I turn if off after 70uS. In theory, after turning 'on' it should remain 'on' until the voltage crosses zero again, right?

OK.pngFault.png
 
Last edited:

Ian0

Joined Aug 7, 2020
10,253
I'm controlling the triac using 70uS pulses. So at the right moment I turn on the MOC, and then I turn if off after 70uS. In theory, after turning 'on' it should remain 'on' until the voltage crosses zero again, right?
Yes - but only if the MT2 to MT1 current exceeds the holding current.
You should keep the triac on until 9ms after zero crossing. That unfortunately means that your control circuitry will consume more power.
The spec for the holding current is a maximum. It is guaranteed to stay on if MT2 current exceeds Ih, but below that, it might and it might not.
Triac sensitivity for both trigger current and holding current varies as much as transistor Hfe and also vary with temperature.
If Ih=50mA.
If Ih=50mA, to get it to latch on until after 9ms, the minimum load would be about 30W, and that assumes a resistive load. What he weird and wonderful circuitry in a "dimmable" LED luminaire may be doing at that point is anyone's guess.
 

Thread Starter

mariosergio

Joined Feb 23, 2018
23
Yes - but only if the MT2 to MT1 current exceeds the holding current.
You are absolutely right! So the difference is that the 'OK' batch can keep the triac 'on' even with a low current flowing.

So is not that I got a faulty batch, it is just that I got lucky with the second one!

To test that I connected a resistor in parallel with the lamp, (to keep the current flowing) and the flicker goes away!
Well, I can't have a 10W resistor inside my product, so I must work around this.

You say I should keep the triac on until 9ms after zero crossing. That is for 50Hz, right?
If I got it, your idea is to turn it off right before the end of the cycle, and for 60Hz it is something around 8.33ms.

Thank you very much for solving this 'mystery'!!
 

Ian0

Joined Aug 7, 2020
10,253
I turn it off between 90% and 95% of the cycle - 9ms to 9.5ms on a 50Hz supply.
Don't wait until the very last moment. If your trigger extends even a fraction of a millisecond into the next cycle, the triac will be ON for the entire cycle. Triacs turn off very slowly (the sensitive gate types* even more so), so give them as much time as you can, without letting the holding current turn them off too early!

*BTA08-600TW etc.
 

MisterBill2

Joined Jan 23, 2018
19,377
OK, so it was not a parameter missing from the spec sheet, it was just one of those usually not an issue. And changing the time is hopefully a small change in a program variable rather than a mask change in a custom device.
 

Ian0

Joined Aug 7, 2020
10,253
OK, so it was not a parameter missing from the spec sheet, it was just one of those usually not an issue. And changing the time is hopefully a small change in a program variable rather than a mask change in a custom device.
It was much the same problem as assuming the “typical” value of Hfe in a transistor circuit. After all, the holding current is a function of the gain of the two hypothetical transistors that make up an SCR.

Now that the TS’s control circuit is working, it won‘t be long before he is faced with the “some dimmable luminaires and more dimmable than others” problem!
 

MisterBill2

Joined Jan 23, 2018
19,377
The fact is, that usually "All mass production units are equal, but some aremore equal."
Of course, "equal" means within acceptable product limits. And usually those are a secret that is closely guarded.
 
Top