So then it's ok to just pulse the triac on and not all the way to near the next crossing?
Multiple conflicting ideas here.
And about not triggering below 4.5ms (in the 1st quadrant), I can say it didn't matter with the triacs I used, it worked just fine and no blown fuses.
The toroid I tried was only a 200VA, but still, if it wasn't recommended to trigger within the 1st quadrant, then I would've seen something there.
Multiple conflicting ideas here.
And about not triggering below 4.5ms (in the 1st quadrant), I can say it didn't matter with the triacs I used, it worked just fine and no blown fuses.
The toroid I tried was only a 200VA, but still, if it wasn't recommended to trigger within the 1st quadrant, then I would've seen something there.
That is the result of using a short, single triggering pulse: in principle, you could go under 500µs safely, but if the triac fails to hold after it has been fired, it will go into an erratic behaviour because the transformer has not been reset properly.
The safest method to be able to use a 10° to 170° firing range, indifferent to the load type, is to apply a constant trigger for the whole off the conduction period
The ZCD I am using explains how the pulse is created. For 50Hz it lasts 1ms, 1/2ms before ZC and another 1/2ms after ZC. I used a timer resolution of about 125ns to be able to really control timing to the last possible bit.
Looking at a window of 4ms, and knowing that you want the ramp to be slow, the 125ns resolution ensures this condition is met.
Using the falling-edge of the ZCD pulse, adding 475us delay before turning the Triac ON for 275us, then turning the Triac OFF.
As long as it is dangerous going down to 4.5ms, I think I can get away with 4.3ms, what do you think? 2ms will ensure I am in the safe side of this madness
Thank you for mentioning these points, excellent!
Can you please explain more.
I already have both ZCD and Triac circuits assembled/tested, just need to order the MCU to get started:
Last edited:
Can I get away with 500us pulse, without having to use a PWM to trigger the Triac during the period I need it to stay ON?
Hi,
For does that still have some doubts about the principle if you look at metal simulation you can see that the trigger pulses in blue are moving from 90 degree toward the 45 degree. Lets analyze the first pulse that will be 8ms. At the 8ms the triac will turn on from 8ms to 9ms. Since the simulation is for the power turn ON means that every time you fire the triac the triac will be conducting until the AC voltage goes to zero. Once the voltage goes to zero the triac will turn off itself and stay off until the next pulse. During the period between the pulse that turn on the triac and the voltage goes to zero the transformer during this period will be providing voltage to charge the capacitor. Both firing of the pulses to turn the triac will be done for the positive/negative cycles of the AC in sequence. The firing cycle will be repeat from 9ms to 4.5ms for both positive/negative cycles. In other words you are ramping the AC voltage like a dimmer. The control of the firing pulses to trigger the triac is the rheostat that control the dimmer. Simple.
For does that still have some doubts about the principle if you look at metal simulation you can see that the trigger pulses in blue are moving from 90 degree toward the 45 degree. Lets analyze the first pulse that will be 8ms. At the 8ms the triac will turn on from 8ms to 9ms. Since the simulation is for the power turn ON means that every time you fire the triac the triac will be conducting until the AC voltage goes to zero. Once the voltage goes to zero the triac will turn off itself and stay off until the next pulse. During the period between the pulse that turn on the triac and the voltage goes to zero the transformer during this period will be providing voltage to charge the capacitor. Both firing of the pulses to turn the triac will be done for the positive/negative cycles of the AC in sequence. The firing cycle will be repeat from 9ms to 4.5ms for both positive/negative cycles. In other words you are ramping the AC voltage like a dimmer. The control of the firing pulses to trigger the triac is the rheostat that control the dimmer. Simple.
Instead of using a single pulse at the start of the conduction, use a pulse lasting the whole conduction time, almost to the next zero Xing, with just a guard time of several hundreds microseconds.
This makes sure that the triac doesn't misfire because of the inductive load, and will create the right initial conditions for the next half-cycle
If 500µs is enough to reach the holding current of the triac with the transformer's magnetizing inductance, yes.
Alternatively, a large snubber can keep the triac conducting if the transformer's current has not risen to a sufficient value
How large should the snubber be Elvee, what are the dependencies here when selecting the snubber values.
It depends mainly on the magnetizing inductance of the transformer: if it small, the current will rise quickly, reaching the hold current in a short time.
The snubber just has to supply enough current to keep the triac conducting until the current in the load becomes sufficient.
If the control pulse is longer, it eases the constraints.
The problem of inductive load control with triacs is discussed in a number of documents, for example:
https://prom-electric.ru/media/an307_triacs.pdf
The snubber just has to supply enough current to keep the triac conducting until the current in the load becomes sufficient.
If the control pulse is longer, it eases the constraints.
The problem of inductive load control with triacs is discussed in a number of documents, for example:
https://prom-electric.ru/media/an307_triacs.pdf
Elvee, the pulse train sounds like a good idea, do you know at what frequency and duty cycle this should be done?
What I'm aiming for now, is doing the pulse train using the same 250us wide pulse but make it multiple instead of just one. And I think a 50% duty cycle should be fine.
Once the sequence is fully done, we go to full turn on anyway, and so it's like a 100% duty cycle.
The key is to stop the pulse train well before the next crossing. The datasheets, at least those mentioning it, advise keeping at least some 200us of non triggering before the next crossing.
Thinking about the small error delays induced by various things, like the response time from the zcd, the software execution time and whatever propagation, we know that pulsing 250us from 9ms after crossing leaves plenty of time left before next crossing, much more than the recommended 200us. That little buffer of extra time allows handling whatever error delays may be incurred.
So I plan to just pulse 250us as before for the first step with the 9ms initial phase delay, and then when the delay decreases with each step, a pulse train with 250us @ 50% duty cycle, ending at about the same time before next crossing.
This should work fine and take care of whatever misfiring issues.
Once the sequence is fully done, we go to full turn on anyway, and so it's like a 100% duty cycle.
The key is to stop the pulse train well before the next crossing. The datasheets, at least those mentioning it, advise keeping at least some 200us of non triggering before the next crossing.
Thinking about the small error delays induced by various things, like the response time from the zcd, the software execution time and whatever propagation, we know that pulsing 250us from 9ms after crossing leaves plenty of time left before next crossing, much more than the recommended 200us. That little buffer of extra time allows handling whatever error delays may be incurred.
So I plan to just pulse 250us as before for the first step with the 9ms initial phase delay, and then when the delay decreases with each step, a pulse train with 250us @ 50% duty cycle, ending at about the same time before next crossing.
This should work fine and take care of whatever misfiring issues.
Where will you stop? I looked at P. 6 "FIRING BY PHASE SWEEP" https://prom-electric.ru/media/an307_triacs.pdf, they say you can go beyond 4.5ms, hence you can keep swapping till you reach the start of the half-wave "firing decreases until total conduction":
This is interesting, what I understood from tauro that I should not switch the TRIAC between 0~4.5ms 45°, specifically before 4.5ms of the half-wave, now this is becoming confusing concerning phase sweep...
This area before 45degrees is the 1st quadrant, and I've been firing triacs in that 1st quadrant just fine. I don't see anything wrong with firing it there, it works.
I used a phototriac to fire the triac, and there is no question about positive or negative pulse there.
The triacs work just fine when fired in that 1st quadrant. Not certain if this has an impact on that, but I've been using snubberless types.
I used a phototriac to fire the triac, and there is no question about positive or negative pulse there.
The triacs work just fine when fired in that 1st quadrant. Not certain if this has an impact on that, but I've been using snubberless types.
Exactly! There is no mention of stopping the decrease of the phase angle at the first quadrant, and it's explicite about "firing until total conduction". That means, no stopping the decrease when 1st quadrant is reached (at 4.5ms).
Works fine, and used all over the place.
Plus it makes for a more gradual sweep with much less of a brunt when switching fully on.
My sequence was made to last about 1.5sec total, in phase decrease steps of 500us, from 9ms and down from there, until 1ms, then full turn on.
The only thing I was doing that I will change is to fire the triac with a single 250us pulse and counting on it to "stick" from there on.
I will just add a train of 250us @ 50% duty cycle instead of just one pulse. Should work just the same, but should prevent any spurious misfiring, if it ever wanted to happen...
The only thing I was doing that I will change is to fire the triac with a single 250us pulse and counting on it to "stick" from there on.
I will just add a train of 250us @ 50% duty cycle instead of just one pulse. Should work just the same, but should prevent any spurious misfiring, if it ever wanted to happen...
I am not questioning the +/-ve half-waves, I understand now that I should do both, we are on the same page concerning this part 🙂 add to it that I also simmed your ZCD circuit, and it appears to be much better, hats off to you 😉
I don't know much about power electronics (triacs) and their quadrant non-sense, I tried to read a lot about it, but my tiny brain doesn't seem to be able to understand it, so this area isn't my playground. I will follow what members recommend for this part.
However, I still don't understand if the way TRIAC is connected plays a role or not, looking at this datasheet for Figures 10, 11 and 12, the load location appears to be different, A2, A2 then A1, respectively. I don't understand why this is done this way and what difference it makes. Bryston appears to be having the load on A1, as in Fig. 12 from the datasheet. I am also wondering if a DC blocker is required or not as Bryston also use a DC blocker, what do you think?
I forgot to mention that you can use a PIC with PWM, this will come handy in case you decided to pulse the TRIAC as Elvee suggested.
I don't know much about power electronics (triacs) and their quadrant non-sense, I tried to read a lot about it, but my tiny brain doesn't seem to be able to understand it, so this area isn't my playground. I will follow what members recommend for this part.
However, I still don't understand if the way TRIAC is connected plays a role or not, looking at this datasheet for Figures 10, 11 and 12, the load location appears to be different, A2, A2 then A1, respectively. I don't understand why this is done this way and what difference it makes. Bryston appears to be having the load on A1, as in Fig. 12 from the datasheet. I am also wondering if a DC blocker is required or not as Bryston also use a DC blocker, what do you think?
I forgot to mention that you can use a PIC with PWM, this will come handy in case you decided to pulse the TRIAC as Elvee suggested.
Last edited:
I certainly did not invent any of this. I just looked at how it's been done in various ways and I thought this was the best way to go about it.
The only thing I need to adapt a bit, is to trigger with a train of pulses instead of a single one, that's all.
I think it doesn't make much difference which side is the load connected to, but what is important is which side the triggering is referred to (A1/A2).
I definitely think the dc blocker is a must nowadays.
It seems this isn't due to whatever small dc we could be generating from our soft start scheme, but rather the incoming dc from the mains.
There are too many dc generating devices out there on the mains that can be encountered, and since the toroids don't like the dc very much, we might as well handle this and just include the dc blocker, for good measure.
I didn't include it in this design I'm posting, but I've been thinking about it.
I was just posting a design for a stand alone soft start, but it should be accompanied by other things such as a dc blocker.
Sure, it can be done. I was just doing it with a simpler PIC with the more beginner's knowledge about PIC assembly programming that I have. In time, perhaps I will go deeper into that and do it that way, later.
For now, I just wanted to make sure what I did works and it uses a simpler small PIC.
In general, the pulse width should be no shorter than 50µs, because triacs are relatively slow and need some time for a reliable triggering.
The pulse repetition frequency depends on the resolution required: for a soft start, it can be relatively coarse. Something like 4~5kHz should be OK.
One thing to remember: the transformer/triac/rectifier/filter combination can behave in unexpected ways, and thorough testing is required before you commit to a definitive hardware/PCB solution.