AC Heater Control Design

[stumpja]

Hey guys. I'm looking to design a circuit to control an AC heater. Specs are 240V, 20A (closer to 15A continuous). I was planning on using a triac design, which, I'm of the opinion of, is a fairly common design path for this type of application. However, parts availability is a problem right now that I'm sure most are aware of, and I can't wait half a year or more for parts. I found some nice D2Pak parts from ST that are stocked, but at the power levels I'm looking to run at, I won't be able to dissipate over 20W easily in a D2Pak. This thread makes it painfully clear, getting 5W out of a D2Pak would take some careful planning and heatsinking.

So in light of part availability (no readily available TO-220's), I started wondering if it would make more sense to design a discrete SSR using FET's. This will be an ON-OFF design, so I will need to design a circuit to detect zero crossings, which shouldn't be an issue as there is already a micro in the design. However, after looking around for FET's, seems like this is going to get a lot more expensive. I was hoping to accomplish it without a heatsink, but to do that, I need FET's with really low RDSon, and they get pricey fast. I would be OK with paralleling multiple, it the cost wasn't ridiculous, especially if it meant getting rid of the HS, but that doesn't seems to be an option either.

So I'm back to using a triac, finding an available part, that I can heatsink well (I have less than an inch in height, so this will be something custom), and going this route. ST has an eval board that doesn't do what I need, it's closer to 10A, but it's ballpark.

I guess I'm on here wondering if anyone has any advice for this type of application, maybe something I missed?

[langwadt]

how many thousands are you building? digikey has plenty of through hole triacs

[schmitt trigger]

A little more complex way I have dealt in the past with excessive Triac dissipation is to use anti-parallel SCRs.

 Not only are you spreading the power dissipation across two devices, but many times for a given current and blocking voltage rating, SCRs have a slightly lower forward voltage drop.

[stumpja]

how many thousands are you building? digikey has plenty of through hole triacs

Good point. I guess I didn't mention that. The EAU per device would exceed 5k. In the current market, something with over 100 in stock is a good start. I could place an order for larger quantities with hopes that the lead time wouldn't be years. However, I will need enough stock now to prototype and validate in the coming months.

To your point, however, there is some stock on Digikey. The available options however, are expensive. compared to say this ST part, which is not stocked.

[stumpja]

A little more complex way I have dealt in the past with excessive Triac dissipation is to use anti-parallel SCRs.

 Not only are you spreading the power dissipation across two devices, but many times for a given current and blocking voltage rating, SCRs have a slightly lower forward voltage drop.

Thank you for this comment. I don't have experience with this, I'll do some research. If you happen to have any accessible reading, feel free to send me a link.

Thanks again!

[langwadt]

how many thousands are you building? digikey has plenty of through hole triacs

Good point. I guess I didn't mention that. The EAU per device would exceed 5k. In the current market, something with over 100 in stock is a good start. I could place an order for larger quantities with hopes that the lead time wouldn't be years. However, I will need enough stock now to prototype and validate in the coming months.

To your point, however, there is some stock on Digikey. The available options however, are expensive. compared to say this ST part, which is not stocked.

https://www.newark.com/stmicroelectronics/t3035h-6i/triac-600v-30a-to-220ab-3/dp/06X1069

[stumpja]

https://www.newark.com/stmicroelectronics/t3035h-6i/triac-600v-30a-to-220ab-3/dp/06X1069

Thanks! I was looking at the 800V part, but the 600V part should work fine as well for 240V operation as well.

I guess my issue now will be thermally heatsinking this part in something low profile.

Thanks again.

[NiHaoMike]

If the application only requires full power for a relatively short time but less than 50% power most of the time (e.g. an oven), you can use a mechanical relay for full power with a SCR across it for reduced power. A clever trick is to have the SCR ready to turn on when the relay is opened at just the right time so that the relay won't have to break the load and would last a very long time.

[floobydust]

I've used the triac+relay combination in heaters, the triac needs no heatsink as it's only on for for a couple cycles until the relay pulls in.
OP, do your thermal calculations as the triac would dissipate around 20W, which is an awful lot for heatsink size and cost, more so if your controller operates in high ambient temperatures. That might be a deal-breaker.
Otherwise, a relay is typically used at 15-20A, they only dissipate a few watts (coil+contact resistance).

[AccountRemovedPerUsersRequest]

I am in the middle of design process for very similar solution and I have been analyzing stock statuses of 25A triacs lately. I feel pretty confident that I can get TH triacs later this year before manufacturing starts. Looks like one specific model can be out of stock today but available next week. Anyway, there are plenty of equivalents to choose from.
Some other components are more critical in my design (in terms of availability).

A

[Siwastaja]

Is there a reason for dismissing pre-engineered SSR (solid state relay) module?

I mean, I like to engineer things from scratch as well, but SSRs are not too expensive, they have optoisolation, zero-crossing detection and EMI already figured out.

In the end, they just contain TRIAC as well. But the TRIAC is isolated from the heat spreader plate, so adding heatsinking is trivial, and driving the module is just like driving an LED (or even better, they often have an integrated constant-current circuit; usual is input spec like 3-30V).

With a heater, you usually have enough thermal mass so you don'tt want phase control anyway, full mains cycles is fine, with PWM frequency around 0.1 - 2Hz.