oh, and one more thing about triacs: I have not found any derating data in datasheets, so how do we infer that info?
I guess for the purpose of bridging the triac, a MOSFET with a very small Rds(on) could also be used, obviating the relay. The dissipation in the MOSFET would be very, very low; you can have MOSFETs with less than 10mohm Rds(on).
Jan
I think you can use the MOSFET SSR without the need for any Triac. Why would I use both? As long as you are able to control when to switch ON/OFF the SSR using delays, assuming the switch times for MOSTETs/Optocoupler is faster than Triac. The only advantage of Triac is it disconnects automatically at zero-crossing.
Looking again at post #98 it seems delays are as follows: 9ms, 8ms, 7ms, 6ms, 5ms then fully ON, that is On times = 1ms, 2ms, 3ms, 4ms and finally 5ms. I am wondering if it is necessary to go any further with delays like this: 9ms, 8ms, 7ms, 6ms, 5ms, 4ms, 3ms, 2ms and 1ms?
Looking again at post #98 it seems delays are as follows: 9ms, 8ms, 7ms, 6ms, 5ms then fully ON, that is On times = 1ms, 2ms, 3ms, 4ms and finally 5ms. I am wondering if it is necessary to go any further with delays like this: 9ms, 8ms, 7ms, 6ms, 5ms, 4ms, 3ms, 2ms and 1ms?
Last edited:
Using MosFETs across a Triac would require 2 of them, a full SSR.
Essentially doubling the number of SSRs, on Triac based, the other MosFET.
I agree a MosFET based SSR could be used in place of a Triac, and so only one SSR would be needed.
I don't see the advantage of using both types of SSRs together. It's either one or the other, and both seems like overkill complexity and part count.
If using a MosFET based SSR instead of Triac based, what would be the easiest most sensible triggering device to use from a PIC, with opto-coupling?
Still, a Triac based SSR is just one part, as opposed to 2 MosFETs to make a MosFET based SSR. The big MosFET advantage is definitely that very low Rdson to minimize dissipation and likely avoid any need for a heatsink.
The logic of turning on the MosFET based SSR would be a bit different from the Triac's, with no need to bother with not triggering too close to the next crossing, but have to willfully turn it off at the end of a half wave.
Essentially doubling the number of SSRs, on Triac based, the other MosFET.
I agree a MosFET based SSR could be used in place of a Triac, and so only one SSR would be needed.
I don't see the advantage of using both types of SSRs together. It's either one or the other, and both seems like overkill complexity and part count.
If using a MosFET based SSR instead of Triac based, what would be the easiest most sensible triggering device to use from a PIC, with opto-coupling?
Still, a Triac based SSR is just one part, as opposed to 2 MosFETs to make a MosFET based SSR. The big MosFET advantage is definitely that very low Rdson to minimize dissipation and likely avoid any need for a heatsink.
The logic of turning on the MosFET based SSR would be a bit different from the Triac's, with no need to bother with not triggering too close to the next crossing, but have to willfully turn it off at the end of a half wave.
Yes so it would suit the application. Ah well, things can be done in a complex and/or in a simple way. Since I am accustomed to power electronics I learned to like relays 🙂
Last edited:
Hi,
To: jan.didden thank you for fixing the link. I tried the review it worked but do not what happened.
To: metal Normally fired the triac 3 times at the same delay for each of the steps. For the turn ON start at 8.5 ms decrement it when reach 4.5 ms turn the triac on to full. Repeat the same procedure when turning the power off. The reason for firing the triac 3 times at the same delay it is to allow to slowly charging the capacitors. For the 500 in the thread #98 used it in a for loop to run it for 500 times ON/OFF with no single errors.
One advice it is to use an incandescent light bulb during the development because you can see if the firing of the triac it is in sync with the zero crossing If you see some light flickering means you are out of sync with the zero crossing. I was able to turn the light very slowly by firing the triac 3 or 6 times at the same time and with small decrement/increment in the firing delay.
To: jan.didden thank you for fixing the link. I tried the review it worked but do not what happened.
To: metal Normally fired the triac 3 times at the same delay for each of the steps. For the turn ON start at 8.5 ms decrement it when reach 4.5 ms turn the triac on to full. Repeat the same procedure when turning the power off. The reason for firing the triac 3 times at the same delay it is to allow to slowly charging the capacitors. For the 500 in the thread #98 used it in a for loop to run it for 500 times ON/OFF with no single errors.
One advice it is to use an incandescent light bulb during the development because you can see if the firing of the triac it is in sync with the zero crossing If you see some light flickering means you are out of sync with the zero crossing. I was able to turn the light very slowly by firing the triac 3 or 6 times at the same time and with small decrement/increment in the firing delay.
Derating what wrt. what? Triacs are ordinary semi's, and general rules apply (maybe I'm overlooking something, but I don't see anything peculiar).
Triacs belong to the same family as SCR's, and they also have a "plasma" conduction mode, meaning for example that they are able to blow a fuse before they melt themselves (warning: they are many restrictions to that principle, and triacs are much less robust than SCR's. Do not be fooled into thinking that an ordinary 10A fuse is going to protect a 10A triac, or even a 10A SCR).
By contrast, if you short the load of a hard-saturated BJT, MOSFET or IGBT, they will always meltdown.
I have tested the conduction symmetry of a number of common triacs (BT139/500 PH, T0812MH Tag, TXAL226B SSC, BTA21D RCA, TIC263D TI, TIC226M TI, BTA25 600 ST, MAC223 On).
The test was made at 500mA, and I left enough time to let the devices reach their equilibrium temperature.
The average voltage drop was 800mV, and the drop with A2 positive was generally (but not always) higher than for negative.
The voltage asymmetry is comprised between -4mV and +32mV.
The worst, 32mV value pertained to the BTA21 D (842mV+, 810mV-).
Not huge, but not negligible either.
The 32mV asymmetry would result in a 16mV average DC component at the load (the value is halved due to the averaging on the whole period).
This value could be compared to that caused by timing errors, in the zero-crossings detection for example.
If the errors amount to 100µs, they are 1/100th of a half mains cycle 10ms, 180°): 1.8°.The voltage reached after 100µs is 325*sin(1.8)~=10V.
Since the sine function is ~linear at small values, it is not necessary to resort to integration: the average voltage in the 100µs interval is thus 5V.
Averaged to the half cycle, it results in a 5*(0.1/10)=50mV DC average.
The values are of the same order of magnitude, meaning timing accuracy is as important as component symmetry for a minimal DC component
The test was made at 500mA, and I left enough time to let the devices reach their equilibrium temperature.
The average voltage drop was 800mV, and the drop with A2 positive was generally (but not always) higher than for negative.
The voltage asymmetry is comprised between -4mV and +32mV.
The worst, 32mV value pertained to the BTA21 D (842mV+, 810mV-).
Not huge, but not negligible either.
The 32mV asymmetry would result in a 16mV average DC component at the load (the value is halved due to the averaging on the whole period).
This value could be compared to that caused by timing errors, in the zero-crossings detection for example.
If the errors amount to 100µs, they are 1/100th of a half mains cycle 10ms, 180°): 1.8°.The voltage reached after 100µs is 325*sin(1.8)~=10V.
Since the sine function is ~linear at small values, it is not necessary to resort to integration: the average voltage in the 100µs interval is thus 5V.
Averaged to the half cycle, it results in a 5*(0.1/10)=50mV DC average.
The values are of the same order of magnitude, meaning timing accuracy is as important as component symmetry for a minimal DC component
Quite interesting and useful data.
How much DC could be handled by a toroid before it starts grouching a bit?
Would such dc also be sent out to the mains? That means it would be good to have a dc blocker at least. It would prevent any dc from going out, but not to the toroid.
How much DC could be handled by a toroid before it starts grouching a bit?
Would such dc also be sent out to the mains? That means it would be good to have a dc blocker at least. It would prevent any dc from going out, but not to the toroid.
Now this will trigger the real question on the ZCD circuit that yields an accurate pulse for both +/-ve cycles and be immune to noise. But is this asymmetry going to affect the inrush current to the extent that we will have to compensate it in the timer preload value.
I suspect your design will have two MOSFETs in series (double the Rds(on)) so that you can orient the drain-to-bulk PN diodes as you wish. I suspect you will orient them in such a way to make it possible to turn the whole thing OFF for both halves of an AC mains sinewave.
I also suspect you will choose MOSFETs with very high max VDS rating in order to work correctly in 240VACRMS applications.
When other folks have tried this (example: Soft as a Feather Pillow) they ended up selecting MOSFETs which cost USD 11 per piece (link) times 2 pieces.
An electromechanical relay is MUCH cheaper.
It very much depends on the transformer: for a 200VA type, <5mA should normally be safe enough (but there are certainly exceptions).
If the primary DC resistance is around 10 ohm, that would translate into 50mV.
These are conservatives figures but as I mentioned, very touchy exceptions certainly exist
A minuscule fraction would be sent to the mains, but it is normally a non-problem: the DC imbalance will be reduced by the voltage divider formed by the transformer's resistance and the mains resistance, which is normally tiny.
Problems occur when a hair-drier or similar appliance uses a diode to create a half-power setting, which is a much more brutal situation
Probably not, but there are still many around. A paint-stripper I bought somewhere between 2000 and 2010 still uses that principle.
Hi,
Yep, Metal great simulation just look like a copy cat as the way it is working in my ramp when looking it with the scope. Perfect simulation. A warning you can not let the trigger of the triac to go below 4.5 ms o 45 degree or you would see a big blown fuse. That it is my recommendation of using a light bulb at the beginning while testing the trigger firing of the triac synchronizing with the zero crossing at different angle when increment /decrement the AC ramp. My final trigger pulse it is 150 micro seconds duration. You can run a test by increasing/decrement the pulse duration until the triac turn on and make adjustment until find the ideal pulse time duration. Also remember the lowest the ramp is done the less inrush current occurred and the voltage charging the capacitors will be very slowly.
Yep, Metal great simulation just look like a copy cat as the way it is working in my ramp when looking it with the scope. Perfect simulation. A warning you can not let the trigger of the triac to go below 4.5 ms o 45 degree or you would see a big blown fuse. That it is my recommendation of using a light bulb at the beginning while testing the trigger firing of the triac synchronizing with the zero crossing at different angle when increment /decrement the AC ramp. My final trigger pulse it is 150 micro seconds duration. You can run a test by increasing/decrement the pulse duration until the triac turn on and make adjustment until find the ideal pulse time duration. Also remember the lowest the ramp is done the less inrush current occurred and the voltage charging the capacitors will be very slowly.