D2PAK triac max current

[AccountRemovedPerUsersRequest]

Hi. I noticed an existing thread about D2PAK cooling, but since this is not exactly that, I dare to create a new topic.

So, I am planning to use T25 triac in D2PAK. Am I missing something critical?

Tj max is 125C, so 25C ambient, the delta T is 100. Rth (j-a) is 45, so the max W is 2.2W.
The Rd is 16mR, so driving 4A it means 256mW. So... I should be more than safe using D2PAK with 1cm copper surface (=the default on the data sheet) on both PCB sides...

A

[Benta]

Are you mixing Triac and MOSFET here? I don't understand your reference to 16 mR.

[AccountRemovedPerUsersRequest]

Not mixing, but trying to calculate how safe I am to use D2PAK triac on 4A. Not sure if using the dynamic resistance in the formula is correct tho.


A

[floobydust]

A triac generates roughly 1W of heat per ampere. I use this for roughly estimating cooling requirements, assuming full conduction angle.
It's not that far off, example ST TN1215 12A triac with Fig. 1, 4A average current gives 4.5W dissipation, which is an awful lot for a D2PAK. You would need more PCB copper area, and would not design to 125°C that is too hot and what happens if the unit is in 30°C ambient.

[langwadt]

Not mixing, but trying to calculate how safe I am to use D2PAK triac on 4A. Not sure if using the dynamic resistance in the formula is correct tho.

a triac is mostly a fixed voltage of ~1-2V

[Siwastaja]

Datasheets always have catches for young players because really basic and important values may be hidden deeper down while weird secondary parameters which might be only used in special cases are in plain view, and confusing. (Vgs(th) in switching MOSFETs is one of such classics.)

TRIAC is like diode or IGBT; voltage drop in on-state is only logarithmically dependent on current (constant is an acceptable first-order approximation of a logarithm); not linearly like in a MOSFET.

For example, in the datasheet of part number posted by floobydust above, you need to look at Fig.9 to find voltage drop over device as a function of current. Then conduction loss power is simply P = U*I.

[AccountRemovedPerUsersRequest]

Yep, got it. I can see that the source of truth is the diagram of power dissipation. And, yes.... roughly 1W per 1A. So no way I can use such a small copper area.
And the On-state diagram leads to the very same answer.

OK, so I need to use another package with a heat sink.

thanks,

A

[asdf336]

You can heatsink a D2Pak on the back side with vias in pads (quite effective if you pack in like 100) or SMT heatsinks which will give you 2-4X more watts. 

But yeah that’s the size of it for D2pack.  A couple watts without additional effort.  TO220 is easier with the trade-offs that come with through hole. 

[AccountRemovedPerUsersRequest]

A heatsink on D2PAK is an option in deed, but it is easier (for me) to use TO220 heatsink at that size. If I did the math right, the min thermal resistance of the HS is around 15C/W. (4W@50C rise).

A

[Doctorandus_P]

You don't want your circuit to fail on a hot day so working with an ambient temperature of 25c usually is not realistic. If you put it in an enclosure then the inside of that box also easily get warmer.

On top of that, electronics above 100celcius are usually frowned upon.
It's a quite arbitrary limit, but easy to measure because it sizzles if you touch it with a wet finger.
It's already quite hot, and and this may reduce life expectancy.

[Siwastaja]

For "typical" easy office/home environments, calculate with Ta=50 degC. For more demanding, hot climates or any chance the product is used in direct sunlight, calculate with Ta=70 degC. Remember, if you use real ambient (room temperature) in the calculation, your thermal resistance model should include the product casing! If you don't have that, you need to somehow estimate local ambient temperature inside the case.

Then use no more than max Tj=110 degC. For parts rated for higher max Tj (some MOSFETs, SiC diodes, etc.), you can maybe go to 120-130 degC.

The advantage of D2PAK is that the footprint area is massive, allowing large number of thermal vias in pad. You can easily dissipate some 5W through the PCB coupled to a heatsink from the opposite side, using a thermal pad. Even more if you do the job well. If you have more than just one component to cool, this translates into simple and cheap assembly, compared to having to mount (with appropriate electrical isolation) each device to a heatsink separately.

If it is only this one TRIAC, a TO-220 case could be the easiest way out.

[AccountRemovedPerUsersRequest]

The PCB is going to be located in an environment, which will have 50C delta and still have some margins.

But yes, in practice, I am going to use a HS around 5-7C/W to make it safe in terms of environmental etc. variables.

A

[floobydust]

It's one reason appliance manufacturers use relays, they are cheaper and smaller than SSR, and dissipate much less heat.
Looking at 4A SSR SIP modules, i.e. Crydom MP240D4 it seems to heat up 50°C.