Ok, after some calculations and a few quick thermal sims, I figured the triac could remain without a heatsink for a load of up to around 300VA or so, keeping the junction below the 125C limit.
But the Triac would get quite hot.
The Triac having a junction-ambient thermal res of 60C/W, with that 1.5V drop, it can add up real quick.
I didn't think a Triac had that much Vdrop. I thought it was more like MosFETs, with their very low RdsOn..
So I suppose it's better to bolt on a small sink, even a small one with something like ~32C/W can make a difference.
But the Triac would get quite hot.
The Triac having a junction-ambient thermal res of 60C/W, with that 1.5V drop, it can add up real quick.
I didn't think a Triac had that much Vdrop. I thought it was more like MosFETs, with their very low RdsOn..
So I suppose it's better to bolt on a small sink, even a small one with something like ~32C/W can make a difference.
Going through the triac's datasheet over again, for the BTA/BTB24 types, the 25A devices, if I get it right, the equivalent to the MosFET's RdsOn would be Rd for the Triac, and for that type, it looks to be 16mohms.
And the voltage drop might be that VTM figure, which states max 1.55V.
The junction-ambient thermal res for the TO220, insulated or not, looks like 60C/W pretty much for any of them.
We're only dissipating a few watts when fully turned on, but with no heatsink, if the toroid is a little big, a small heatsink would be a good thing to use.
I think I should try to give a little bit more space to the triac on the pcb to allow slipping a small sink there.
And the voltage drop might be that VTM figure, which states max 1.55V.
The junction-ambient thermal res for the TO220, insulated or not, looks like 60C/W pretty much for any of them.
We're only dissipating a few watts when fully turned on, but with no heatsink, if the toroid is a little big, a small heatsink would be a good thing to use.
I think I should try to give a little bit more space to the triac on the pcb to allow slipping a small sink there.
Some observations: to reliably control an inductive component with a triac under arbitrary conduction conditions, it is necessary to apply a permanent triggering for the whole of the conduction time.
You might think that you can get away with just a single pulse, but you are always going to be caught somewhere.
An example: at the end of the soft-start period (conduction~=180°), your single pulse will be located just after the zero-crossing, but with a classical rectifier+filter cap as a secondary load, the high-current charging pulses only happen 90° later. Even with a substantial snubber, the triac might fall out of conduction before the current pulse arrives.
After the 90° current pulse, you could think that the conduction does not matter anymore, and you can let the triac do what it pleases.
However, if the triac doesn't pass the last part of the sine arch, the core of the transformer will not be preloaded with the correct initial conditions to cope with the next complete half-cycle, ledaing to saturation.
All of this may lead to an erratic conduction and DC component.
A permanent triggering does not necessarily require a constant DC current: short, energetic pulses at a high repetition rate also work quite well, at a reduced average consumption.
Note that you do not need to dissipate ~1W in zero-crossing detector: there are lower power, reliable solutions
You might think that you can get away with just a single pulse, but you are always going to be caught somewhere.
An example: at the end of the soft-start period (conduction~=180°), your single pulse will be located just after the zero-crossing, but with a classical rectifier+filter cap as a secondary load, the high-current charging pulses only happen 90° later. Even with a substantial snubber, the triac might fall out of conduction before the current pulse arrives.
After the 90° current pulse, you could think that the conduction does not matter anymore, and you can let the triac do what it pleases.
However, if the triac doesn't pass the last part of the sine arch, the core of the transformer will not be preloaded with the correct initial conditions to cope with the next complete half-cycle, ledaing to saturation.
All of this may lead to an erratic conduction and DC component.
A permanent triggering does not necessarily require a constant DC current: short, energetic pulses at a high repetition rate also work quite well, at a reduced average consumption.
Note that you do not need to dissipate ~1W in zero-crossing detector: there are lower power, reliable solutions
So you think a long pulse all the way to before the next crossing, or close to it, should be used instead of a single short one?
I haven't noticed any odd behavior when I tried it on a toroid. Granted, it was only a 200VA, but I did load it with a bridge and cap banks, 30mF worth. Shouldn't it have misbehaved then?
That is a proposition that is much more "involved". Handling a train of pulses that must be varied in length with each sequence step is quite a bit to do. Would be much easier to just widen the pulse, although it would still need to be of varying length.
Are you referring to the dissipation in those 2 resistors on each side of the opto's led?
What would you propose to be better?
?
Absolutely: in industrial controls, it is systematically done that way (or with a burst, for reduced consumption)
A single pulse triggering will not necessarily fail every time: it will depend on the triac characteristics, the load, the transformer's parameters, other factors and their combination.
Doing things the right way will generally cost nothing, and save many headaches and problems afterwards
I don't think so: in general, there is a simple way to use the counters timers to do that. I cannot give you specific clues for your PIC, but many members probably can.
If you have enough current available, you can just lengthen the pulse without modulation.
You can search the WWW for ideas: there are lots of them.
You can also keep the same circuit and increase the resistance level ~5 times: it will still operate reliably, at a much reduced power level.
An example of what can be found on the net:
https://forums.futura-sciences.com/...44994-petite-schematheque-de-futura-zxing.png
Other examples here:
Zero Cross Detector Electronic Circuits
Last edited:
😀😀😀 probably not in Florida 😀😀😀
It would be pushing the limits real hard for sure. Probably not a good idea.
I'm making an alternate version with a beefier triac in a TOP3 case, with more room around to heatsink it. Should help with bigger toroids.
The TOP3 version of the BTA/BTB24 is obsolete and there aren't many alternatives. The BTA41 does have the TOP3 case, and it's a 40A type.
Hi,
Attached it is for your consideration a picture showing how I did the ramp of the AC using the zero crossing. I did it in 2012 and still up to today it is working with not a single problem. For the traic used the BTA256OOBW because it s very robust, type T03 easy to mount and isolated. Also I would like to apologize to the owner of the thread if you do not mind to thanks ELVEE for his advice to fixed the problem encountered when reaching the 90 degree blowing the fuse. Attached it is the link of my thread explaining how did the ramp of the AC using a micro to prevent the inrush current when turning on an equipment with a high current transformer.
link:http://https://www.diyaudio.com/forums/power-supplies/217526-preventing-inrush-current-saturation-toroidal-ei-transformer-2.html"]http://https://www.diyaudio.com/forums/power-supplies/217526-preventing-inrush-current-saturation-toroidal-ei-transformer-2.html[/URL]
Attached it is for your consideration a picture showing how I did the ramp of the AC using the zero crossing. I did it in 2012 and still up to today it is working with not a single problem. For the traic used the BTA256OOBW because it s very robust, type T03 easy to mount and isolated. Also I would like to apologize to the owner of the thread if you do not mind to thanks ELVEE for his advice to fixed the problem encountered when reaching the 90 degree blowing the fuse. Attached it is the link of my thread explaining how did the ramp of the AC using a micro to prevent the inrush current when turning on an equipment with a high current transformer.
link:http://https://www.diyaudio.com/forums/power-supplies/217526-preventing-inrush-current-saturation-toroidal-ei-transformer-2.html"]http://https://www.diyaudio.com/forums/power-supplies/217526-preventing-inrush-current-saturation-toroidal-ei-transformer-2.html[/URL]
The link is wrong; the currect link is:
Preventing the inrush current saturation in a toroidal/EI transformer
Jan
Preventing the inrush current saturation in a toroidal/EI transformer
Jan
I see tauro, it has been done before, and successfully, never thought of such scheme to be honest. The idea was to use such delay scheme: 9ms, 8ms, 7ms, 6ms, 5ms, 4ms, 3ms, 2ms, 1ms, fully ON, for each half-wave for a couple of seconds.
In post #98 you posted the final scheme that you used for the production project, what do you mean by 500 times? Also are the actual delays like this for each-half wave: 9ms, 8ms, 7ms, 6ms, fully ON, and you are repeating each delay for like 125 times? So you have avoided the 5ms delay which is at 90°?
In post #98 you posted the final scheme that you used for the production project, what do you mean by 500 times? Also are the actual delays like this for each-half wave: 9ms, 8ms, 7ms, 6ms, fully ON, and you are repeating each delay for like 125 times? So you have avoided the 5ms delay which is at 90°?
On the phase shifting slow turnon: this is not really necessary, IF you make sure the transformer is turned on at exact the same point on the waveform as it was turned off.
The reason is the following. When turned off, the magnetization point will be *somewhere* on the BH curve. The remanent magnetism will keep the magnetic domains at that position.
If you then turn the xformer on again at the exact same point on the phase, there will be NO inrush current for the core (there will be inrush current for the rectifier/cap of course but that is much less). The worst case is when you turn on the transformer 180 degrees from the turn off point; at that point, the core induction pretty much collapses and the only thing limiting the core inrush is the resistance of the primary wire. The very first time you turn on equipment with this provision, you don't know the position of the core on the BH curve, so there may be large core inrush. But from then on you can have it under control without the need for phase shifted soft turnon.
One other datum point: the delay between turning on a triac and it actually conducting current can be several milliseconds, so that needs to be taken into account in the software.
Jan
The reason is the following. When turned off, the magnetization point will be *somewhere* on the BH curve. The remanent magnetism will keep the magnetic domains at that position.
If you then turn the xformer on again at the exact same point on the phase, there will be NO inrush current for the core (there will be inrush current for the rectifier/cap of course but that is much less). The worst case is when you turn on the transformer 180 degrees from the turn off point; at that point, the core induction pretty much collapses and the only thing limiting the core inrush is the resistance of the primary wire. The very first time you turn on equipment with this provision, you don't know the position of the core on the BH curve, so there may be large core inrush. But from then on you can have it under control without the need for phase shifted soft turnon.
One other datum point: the delay between turning on a triac and it actually conducting current can be several milliseconds, so that needs to be taken into account in the software.
Jan
Given that a half wave is "only" 10ms long, if a triac took several ms to conduct, that would make it quite hard to work within those 10ms and apply proper timing.
I've been reading whatever I came across to figure out what pulse duration would be good to use. I was using a single 250us pulse to turn the triac on, but if instead of a single, a train of pulses is to be used that lasts much longer, maybe each short pulse in the pulse train can be much shorter.
I've seen mentions in application notes about a 20us based pulse, and then repeat that.
As a compromise, perhaps a train of 100us pulses might be usable.
One thing I've seen in more than one app note is that triacs shouldn't be triggered too close to the next zero crossing, and to leave at least 200us of not triggering before that next zero cross event.
Anyway, none of this is new, except that I'm among those just getting into it now. It's been done, long ago, and that gradual phase shifting method works. I've tried it, although not with a big toroid and I had no measurement means.. I wish I had something useful to view the waveform resulting (by actual measurements, not sims).
I've been reading whatever I came across to figure out what pulse duration would be good to use. I was using a single 250us pulse to turn the triac on, but if instead of a single, a train of pulses is to be used that lasts much longer, maybe each short pulse in the pulse train can be much shorter.
I've seen mentions in application notes about a 20us based pulse, and then repeat that.
As a compromise, perhaps a train of 100us pulses might be usable.
One thing I've seen in more than one app note is that triacs shouldn't be triggered too close to the next zero crossing, and to leave at least 200us of not triggering before that next zero cross event.
Anyway, none of this is new, except that I'm among those just getting into it now. It's been done, long ago, and that gradual phase shifting method works. I've tried it, although not with a big toroid and I had no measurement means.. I wish I had something useful to view the waveform resulting (by actual measurements, not sims).
I facing same dilemma here, my oscilloscope is broken, and what ever code/method done needs to be tested anyway.
Concerning pulses train, do you need to keep that train of pulses for the period at which Triac is needed to be ON?
Concerning pulses train, do you need to keep that train of pulses for the period at which Triac is needed to be ON?
I wasn't doing anything more than a 250us pulse to turn on the triac, nothing more. But I guess we can go for a more "firm" method and do a pulse train for the longer duration, except that needs to stop at least 200us before the next crossing.
Relays work well but because they are mechanical do wear out eventually.
I see quite a few threads on DIYAUDIO where they are having trouble with protect/soft start circuits.
I have always used triacs for soft start or mosfets for DC protect circuits.
One problem with relays is the arcing when they drop out.
They drop out slowly due to fly wheel diode continuing to pass current through relay coil.
This can be fixed by using a Zener in series with the diode.
I see quite a few threads on DIYAUDIO where they are having trouble with protect/soft start circuits.
I have always used triacs for soft start or mosfets for DC protect circuits.
One problem with relays is the arcing when they drop out.
They drop out slowly due to fly wheel diode continuing to pass current through relay coil.
This can be fixed by using a Zener in series with the diode.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
If the relay is used to bridge a triac that is on, it doesn't really wear out, especially when it is opened and closed at zero crossing. But it does make sure that the 0.8V triac threshold is eliminated, eliminating any possibility of a DC shift in the mains.
Jan
Jan
A DC blocker works both ways, so it's a good thing to have anyway.
Going all solid state eliminates the need for electromechanical stuff, so what's the point if both are used then?
I keep the use of relays only to places where there are no other choices possible.
Going all solid state eliminates the need for electromechanical stuff, so what's the point if both are used then?
I keep the use of relays only to places where there are no other choices possible.
One thing that I have noticed in tests with a toroid, is that they're not too fond of those sharp edges from switching on off phase this way. So some of them may growl a tiny bit during the ramp up sequencing, but that only lasts as long as that ramp up sequence, and then they go quiet once fully turned on, so not really any big deal.