I wish to switch 24V AC from my HVAC with transistors. Here is one answer but I can't figure out what the hell "antiparallel" means, exactly, in respect to the SCR built from two transistors. Could someone answer this?
For what it's worth, I'd just buy a Triac (and probably an opto-coupled gate driver). Building your own SCRs (never mind TRIACs) out of transistors can be tricky: it can be really hard to control and match trigger characteristics over temperature. Look here perhaps for some ideas: triac for sale | eBay .
(You did mean the 24VAC control voltage to your HVAC, and not 240VAC right?)
It's kind-of silly to "build your own TRIAC." SCRs only work in one direction like transistors/MOSFETs, so you'd need two and you might need an inverter to drive the "reversed' one.
If you buy a solid state relay, make sure it's rated for 24VAC. Many 120V/240V solid-state relays don't work properly at low voltage.
The combination of low current and inductive loads is problematic for triac switching. There is zero reason to use a triac for HVAC switching applications and trying to re-invent the wheel is a pointless project killer.
Ask yourself: What is your goal? To build a thermostat or learn about reliable low current ac triac switching?
If you don’t own and know how to use an oscilloscope, you’d best stick with first topic, the second one is off the table, at least for the immediate future.
CurtCarpenter:
My guess is they mean something like this:
I hope not, that cannot work as the gates will pull massive currents and fry. The gate is constrained
by a forward biased pn junction to be close in potential to its cathode(MT1). You need two SCRs
of opposite polarity in parallel, not two of the same polarity in antiparallel.
Actual triacs are single devices and their action is rather more complicated than this it turns out
(wikipedia as usual has good info here).
Only one polarity of SCR is commonly available with a p-type gate, but complementary gate SCR
is possible.
SCRs built from a NPN and PNP device will be crude as the doping levels are wrong for optimal
SCR function. Design of bipolar devices are all about doping profiles.
I'm thinking overall that small relays may be best. I could altertnativly buy TRIACs. I have relays, but oddly am lacking diodes to protect the MCU from inductive flyback from relay. The current isn't great, as you are switching just another relay/contactor on heater and ac.
MarkT:
Only one polarity of SCR is commonly available with a p-type gate, but complementary gate SCR
is possible.
And if you'd bother to read the initial post here, you'll see that the issue wasn't how to connect two commercial SCRs in anti-parallel, but how to connect two SCRs made from discrete npn and pnp transistors that way.
mattlogue:
I'm thinking overall that small relays may be best. I could alternatively buy TRIACs. I have relays, but oddly am lacking diodes to protect the MCU from inductive flyback from relay. The current isn't great, as you are switching just another relay/contactor on heater and ac.
I have relays, but oddly am lacking diodes to protect the MCU from inductive flyback from relay.
At a pinch a BJT can be used as a diode - use the base-collector junction (the emitter-base
diode has a very low reverse breakdown voltage and usually isn't suitable). The current handling
for such a diode can be quite small though.
Alas LEDs would be great but they too have very low reverse breakdown voltages (typically -5V).
You can use a resistor across the relay winding to act as snubber, choose a value around R to 2R
in value where R is the resistance of the winding. You need to drive more current of course, but
it will limit the voltage spike perfectly well. Adding a capacitor in series will reduce the current
draw but it needs to be large enough to swallow the whole of the inductive spike.