How much copper area to expose for soldering tab of TO220?

[HwAoRrDk]

I have a TO-220 package I am going to mount horizontally on my board, and I plan to solder the tab to an exposed copper area on the ground plane. I also have a bunch of vias for heat transfer to the bottom ground plane.

Trouble is, I don't know how much area to allow around the tab for good soldering. At the moment I have 0.025" margin. Not enough? The datasheet for my particular IC doesn't indicate that the tab on the package has chamfered corners or concave cut-outs on either side, so I can't rely on that for extra room for solder.

Also, if I increase the area, I have room for another line of vias around the top and sides. More is better, right? Or would there be negligible effectiveness for anything not right under the tab?

By the way, I couldn't figure out in DipTrace (if it's even possible) how to mask out an area of silk screen. At the moment, as can be seen, the silk outline for the TO-220 is over the exposed copper. From my research here, I think that virtually all PCB manufacturers will trim the silk screen layer according to pads, exposed copper, etc. and that part won't be printed on final boards. Is it safe to rely on this, or should I edit my component pattern/footprint to remove the overlapping lines?

[uncle_bob]

Hi

Ok, this is actually several questions. In no particular order:

For soldering, this is a surface mount part once you mount it horizontally on a board and solder it down. A D2 pack is a good example of a very similar "pure smd" part. You only need enough room for your solder paste and the alignment of the part. Do not bother trying to hand solder one, it's not gong to come out well. If it's hand mounted, screw it down instead.

For heat sinking, more layers of copper (4 is good, 8 is better ...) is always going to help. More vias also help. The amount of board area covered by "pure copper" could be anything from the size of the tab up to 60 mm or a side. It all depends on how much heat you have, what the max temperature is, and how much airflow is present.

Often on silk screen layers, you put in a negative object in the screen to make a hole. No idea of the settings in Diptrace to do that.

Bob

[HwAoRrDk]

For soldering, this is a surface mount part once you mount it horizontally on a board and solder it down. A D2 pack is a good example of a very similar "pure smd" part. You only need enough room for your solder paste and the alignment of the part. Do not bother trying to hand solder one, it's not gong to come out well. If it's hand mounted, screw it down instead.

Oh, that's exactly what I was planning to do - hand solder. I guessed it was doable if you could apply enough heat with the iron. I kind of need it to be soldered as I don't really have enough room for a mechanical fastener - there will be an object almost directly above the tab.

I'm not equipped to be doing SMD work, with paste, etc. Maybe I should look at getting into that.

For heat sinking, more layers of copper (4 is good, 8 is better ...) is always going to help. More vias also help. The amount of board area covered by "pure copper" could be anything from the size of the tab up to 60 mm or a side. It all depends on how much heat you have, what the max temperature is, and how much airflow is present.

Not entirely sure exactly how much heat to handle yet, but I do know there will be bugger-all airflow, so thinking I need to err on the side of excess. More vias it is then!

Often on silk screen layers, you put in a negative object in the screen to make a hole. No idea of the settings in Diptrace to do that.

That's what I was looking for, but Diptrace doesn't seem to cater for that on silkscreen layers - only for soldermask, routing keepout, etc.

[uncle_bob]

Hi

You are building a great big heatsink out of many layers of copper. It will (if designed correctly) be very good at moving heat around. If it's designed to handle 10W at an 80C rise, you will need 50W (delivered) to get it to 400C. If you want it to heat as fast as a normal solder joint, you will need to deliver 500W. Why 400C? In order to get all the solder to flow, that's likely what your contact point will be getting to.

So now, what happens to the junction on the transistor when it gets real hot? Nothing good.

What happens to the board material it's self when it gets real hot? Nothing good.

Do you have a 500W temperature controlled iron? Neither does anybody else.

How do you control the temperature on a 500W iron - variac. How well does that work? Not so well.

How do I know this .... yup ... been there /done that.

You can try to pre-heat the board (possibly with a torch, better with a hot plate) and take out part of the crazy stuff. Given how cheap a "toaster oven" reflow setup is, that's a much more civilized way to go. Switch from a TO-220 to a D-PAK and life is good.

Bob

[nctnico]

I'd create a large rectangular pad the size of the TO220 body and the tab. An 80W iron with a big tip set at 330 deg Celcius will be more than fine to solder it. BTW I have used a similar construction in some of my designs so I have some hands on experience. If the PCB turns out to suck too much heat from the soldering iron then I recommend using hot air.

[T3sl4co1l]

Hand soldering, look for at least 50 mils extra in one direction.

Tim

[HwAoRrDk]

Hmm, yes, I suppose when you put some figures to it, trying to solder something in an area that's designed to dissipate heat does drive home the fact it becomes much harder.

I think I'm going to give it a try anyway. I've, as suggested, increased to 50 thou margin around the tab and made the unmasked area a 'T' shape to meet more of the TO-220 body. And I'll order a couple of extra components in case I fry one with too much heat.

Other than that, I'll have to hope that I have a couple of millimetres clearance above to use a screw to fasten (yes, space is that tight).

My final measure without investing in a proper hot air gun would perhaps be a crazy attempt to reflow with a small butane torch - you know, the kind of small palm-sized ones mostly intended for heat-shrink.

[uncle_bob]

Hi

This really very much comes back to how much heat you need to dissipate in operation. For a heatsink design that will handle a half watt, yes, you can solder it down. For a design that will handle 10 watts ...it's going to be tough. The heatsink requirements are a very  significant driver to the board layout.

Bob

[KL27x]

Heat dissipation alone is just one factor. You also have to consider the max operating temp of the device to get the minimum output you need. For instance 10W  at 75C is going to require a lot more heatsinking than 10W at 150C, ambient being constant.  (Max temperature of surrounding components not withstanding.)

[rob77]

the tab of the TO220 package is NOT designed to be soldered... why are you not using a DPAK or D2PAK (bigger one) device ?

[Jeroen3]

Some to-220 are un-solderable. Beware.
Get a DPAK instead. That is still duplo sized SMD easy to do by hand. No paste required.

[HwAoRrDk]

Mainly because the equivalent IC (an LM2931) in D2PAK is literally twice the price. Plus the only D2PAK version available at my chosen vendor is from TI, which inexplicably and annoyingly has an inferior max. input voltage rating (26V versus 40V ).

[rob77]

LM2931 is 0.47 to 0.66 eur + VAT  in single piece @ TME. D2PAK, DPAK and TO220 - the price varies by manufacturer and actually the 2 most expensive ones 64 and 66 cents are TO220. cheapest 47 cents is a DPAK.

[T3sl4co1l]

Some to-220 are un-solderable. Beware.
Get a DPAK instead. That is still duplo sized SMD easy to do by hand. No paste required.

I've seen a few that are apparently nickel plated; soldering is difficult, but not impossible.  Most are tin plated, but it's not usually documented, so you're on your own in that regard.

Second the recommendation for D(2)PAK.

Tim