Reworking DPAK FETs from double sided FR4 PCBs with much thermal copper?

[ocset]

Hi,
Please help with removing and replacing DPAK FETs from an FR4 PCB with much thermal copper on/in it?

So, Today I went for a practical interview for a soldering job. It was required to remove 5 DPAK FETs from a PCB , and then replace them with different DPAK FETs. (The PCB was a 100W, 36V battery charger SMPS. This PCB was in production, but the FETs were said to be counterfeit, and thus in need of replacement) I failed  the assessment, and so didnt get the job.

I’ve now searched the web for “the way” to do this, but all the ways recommended wouldn’t have worked on this job.

The DPAK FETS were generally close together near the PCB perimeter. The PCB was double sided populated , so not generally conducive to placing on a hot plate…but the PCB area directly under the DPAK’s had no components , so therefore that “Bit” of the PCB could have been somehow placed on a “bit” of  the edge of a  hot plate. (however, no hot plate was available)

So, the PCB was……
8cm diameter round PCB,
It was 2mm thick and of FR4 (I did not have access to gerbers, schem or PCB layout  files)
Each DPAK FET had many 0805 resistors and capacitors in the close vicinity…ie, only about 1mm away.

There must have been a lot of internal layer  copper under the drain tab of each FET, since the 80W METCAL soldering iron was struggling to get enough heat into the DPAK tab. The tip was an approx. 1.5mm chisel tip of length ~1cm (there were no other tips available). Each DPAK drain pad had some 12 thermal vias in it.

I eventually managed to coax all the DPAK FETs off the PCB by using the metcal iron, and another soldering iron of 60W, in a kind of “chopsticks pincering” action….and using lots of solder to kind of help get heat into the drain tab and the gate and source legs. I was only allowed to use unleaded solder…..(99% Sn) Getting these 5 DPAKS off the PCB , without removing the surrounding 0805’s, took about 20 minutes.

There was a hot air gun available to me, but this only had one nozzle, and this nozzle was 3cm long and ~1.5mm in diameter…..it was totally inneffective, and could not get enough heat into any of the DPAKs, even on max temperature.

….***….***….***….***….***….***….***….***….***….***….***….***….***….***….**

Anyway…now for getting the replacement DPAKs on to the PCB……

…I managed to do this……..getting perfectly good electrical connection of all drain’s sources and gates…..however, the only way I could do it was to initially put a large “blob” of  (unleaded) solder onto the drain tab, and then kind of “splurge” the DPAK into this blob of solder..as follows….

...... I  initially tinned the drain tab of each DPAK…..i then melted up the “blob” of solder on the drain pad, and then quickly removed the solder tip from the drain pad, and then “splurged” the DPAK  FET into the molten solder blob on the drain pad….i then re-applied the soldering iron tip into the edge of the “blob”  of solder which was now sandwiched between the DPAK FET and the DPAK drain pad.

…This resulted in DPAK FETs which were not flat to the board…in fact, there was a “sandwich” of solder of some 1-2mm thick between the DPAKs and the DPAK pad. Even though this would not snag the enclosure, the interviewer told me it was messy, and so he failed me...to be honest, i needed the solder filled gap between DPAK tab and PCB drain pad, because then i could poke the solder iron into it and be sure the entire area of the DPAK tab was wetted and soldered to the entire area of the drain pad on the PCB

How should I have done this work? What tools should I  have used?
(I have searched google for “DPAK removal tools”, etc, but drew a blank)

[T3sl4co1l]

I'd have looked around for another nozzle for the hot air, or just went with it.  Or grabbed a general purpose heat gun, though that's tricky and I would make special note that I am using it with the utmost caution.

Also, wedge tip for the iron, if they have several around.

Patience is needed.  Wait for the board to heat up.  You aren't going anywhere until then.

When destructive removal is acceptable (seems to be the case here), the leads can be desoldered individually and bent out of the way with stout tweezers.  Then you don't need multi-iron trickery.

If they have flux appropriate for unleaded solder, and the irons go hot enough, there's no problem there, just takes a little longer to heat up.  Patience.

Once removed, wick excess solder off the pads.  Maybe clean up the area too, get rid of the crud you've put down.

Tack down the leads first, they're easiest.  Use enough solder to flux the joint and get a good fillet.  Use only enough solder on the tab to get a good fillet.  Again, preheat is key.


If the hot air is suitable, then just blast it with that for a while (again, warming the board primarily), lift the component, flux the area, and put down the new component; touch up with additional solder if necessary.

Tim

[pisoiu]

In addition to above arguments, from my experience, it is absolutely required to use hot air preheater for this kind of job, placed underside the component and raising surrounding area to at least 100-150 degrees Celsius.

https://www.alibaba.com/product-detail/New-Aoyue-853-Compact-Preheater-Hot_60464348486.html

Heating just locally, without preheating, especially with lead free solder, even if it does the job, you risk a lot, from layer delamination, vias ruptured, etc. Attachment beween copper pads and substrate is much weaker at high temperatures. Best case you destroy something and it is immediately visible on the test. Worst case, the damage remains hidden for a while and the device fails later at the customer.

[ocset]

it is absolutely required to use hot air preheater for this kind of job, placed underside the component and raising surrounding area to at least 100-150 degrees Celsius.

https://www.alibaba.com/product-detail/New-Aoyue-853-Compact-Preheater-Hot_60464348486.html

Thanks,  due to the components and connectors on the bottom of the board, i dont think we could have used that particular  hot air pre-heater to good enough  effect.

Tack down the leads first, they're easiest.  Use enough solder to flux the joint and get a good fillet.  Use only enough solder on the tab to get a good fillet.  Again, preheat is key.

..Thanks, this has given me a thought…because it has dawned on me that it could be possible to hand solder the DPAK on to the board, and end up with the majority of the tab of the DPAK not having wetted with solder, and thus not having  soldered properly over the whole DPAK  tab area…..this would mean insufficient thermal coupling of the DPAK to the PCB pad.

As such, it is making me think that  in fact, the best way to do this (other than actually putting the PCB through a reflow machine again to  reflow just  the DPAKs), would be to get a hotplate, and then somehow press just  that  underside part of the PCB that’s underneath  of the FETs, against  part of the hotplate….and then apply a  thin film of solder to the pad, and then wait until that thin film of solder goes  wet (melts), and at that point, then apply the new DPAK to the pad…(with this DPAK having a pre-tinned tab). ..This , I think, would be the only way to assure that the entire DPAK tab area was correctly solder bonded to the whole area of the DPAK pad on the PCB....

....Would you agree with this?

After all…some solder rework staff, might just solder down the gate and source legs…and then just “pretend” that the DPAK tab is fully soldered down…whereas they may just have soldered “round the outer edges” of the DPAK tab…and this would lead to insufficient thermal coupling of the DPAK to the PCB,  and thence potentially  lead to thermal failures.

So for soldering the new DPAK onto the PCB,  would you now say that a hot plate is essential for this…? (obviously going back through the reflow oven would be great but very expensive)

Also, i wonder if its acceptable in the EU to use leaded solder for rework cases like this?.....because the solder bonding of the DPAK to the pad would be far more reliable if leaded solder was used in such " hand rework" cases.

But then I suppose another way might be to put a thin fillet of solder on the DPAK  drain pad, and then some flux, and then  put the  DPAK on the pads, and then heat blast the hell (with a heat gun) out of the DPAK body until you see it shift and settle onto the pad, which would be an indication that the solder had reflowed under the entire DPAK drain  pad? ..also putting kapton tape over the surrounding 0805 components so they dont get blasted away by the "heat gun hurricane"?

[T3sl4co1l]

No, don't use a hotplate to reflow.

If you must use a hotplate for preheating, that's fine, of course a purpose-made one is preferable.

Hot air can be used the same way, perhaps with some insulation around (Nomex or fiberglass pads?) to reduce heat loss.  I've had to do that before, put insulation around a large object to make a sort of oven to get the thing up to temperature; but that was also a solid copper assembly, an exceptional case.

I've done DFNs on boards with 8 layers of planes, just takes a long time before the whole board gets warm enough the hot air can do its job.  In that case, I merely tin the pads, apply flux, and place the component.  In time, it settles down, and I can give it a nudge with tweezers to ensure it's seated, has the correct amount and distribution of solder, and likely doesn't have shorts under it.  If it has too much solder, I can push down and squeeze out the excess, wick it away and try again, perhaps with another dose of flux.  For DPAKs and such, where a toe or side fillet is visible, this last step is not necessary, you can see immediately how much solder is in the joint.


You clearly have little or no experience using hot air.  Get a machine and practice using it.  Set the temperature a bit above soldering temp, set the air flow as needed (for a board like this, you'll probably want max flow and a large nozzle), and wait.  If it's not melted in say 5 minutes, turn up the heat and wait some more.

The last thing you want to do is burn the shit out of it, with an uncontrolled heat gun, or a hot air machine set way too high.  That's why it's temp controlled, so you can't burn it up, at least not very quickly.

Chip components will not be blown away, at least not with a common machine.  If it's blowing that strong, turn it down, duh.  They also will not fall off the bottom, if the bottom side should get hot enough to melt.  Surface tension is quite strong compared to the size and mass of these components.

Tape is not advisable, as it will tend to lift and pull components if they happen to melt.

Tape and foil can be used as heat shields, say to keep hot air away from connectors, electrolytic capacitors, etc.  Use polyimide, preferably with silicone adhesive.  This may not be helpful if the board has internal planes, i.e. you need the heat applied to most of the board, not just a local area.


Ed: also, regarding reflow, it's not so much that it may be expensive, as it's just awkward.  To put it on the production reflow oven, it needs to be fixtured, and the fixture may need PCB rails that have been stripped off at this stage of assembly.  So you'd have to cook up a replacement first.  Reflow also doesn't help with removing the component, or cleaning the pads.  Hot air and soldering iron is the correct rework method here.

Tim

[wraper]

With all constrains listed, I would remove tiny nozzle from hot air and simply use it without any nozzle.

[ocset]

and I can give it a nudge with tweezers to ensure it's seated

Thanks, you  were kindly talking about QFNs here, but i assume (hope) its the same for DPAKs?...as you are right, i dont have much experience with hot air guns, and wouldnt know when ive applied enough heat to totally solder all the drain tab to all of the drain pad.

That kind of "wiggle" that a QFN  does after its  been nudged under hot air, is the only way i would  know a DPAK had been soldered properly.

With all constrains listed, I would remove tiny nozzle from hot air and simply use it without any nozzle.

Thanks, the thing is, i think this would need tape over the adjacent components (0805's) to stop them blowing away....and as T3sl4co1l said, the tape can wrench components of the board when they get hot.

[wraper]

With all constrains listed, I would remove tiny nozzle from hot air and simply use it without any nozzle.

Thanks, the thing is, i think this would need tape over the adjacent components (0805's) to stop them blowing away....and as T3sl4co1l said, the tape can wrench components of the board when they get hot.

It's really hard to blow away 0805 even if you try to do it intentionally.

[flolic]

8cm diameter PCB is actually quite small board, and no matter how much copper layers it has, it would really be peace of cake to rework it with any decent hot air.
I estimate it would take me less than 5 minutes to replace 5 DPAKs using hot air gun, with that tiny nozzle removed.

[SMTech]

Its an interesting challenge and personally I don't think you would normally attempt this kind of rework with the tools you describe if this was anything resembling a common requirement. A proper rework station would have pre-heat on the underside and then an overhead hot air nozzle or focused IR beam, the D-paks would come off in no time and you would have both hands free for operating tweezers, applying flux & cleaning up.

Of course you can try and emulate that with the tools you had, but its the kind of situation where you wish you had an extra arm or two, that leaves you with time and the hot air, which with the thermal vias and copper you  describe would be slow whatever the nozzle, maybe a touch of dual wielding .

However its quite possible that any clever tricks we might come up with on here would fail the interview because they don't meet some part of the IPC standard, so if it is something that interests you it could be worth doing some training.

[Siwastaja]

Preheat, preheat, preheat.

Any way of preheating is OK. Hotplate, if you have it. Hot air from a distance away.

In a pinch, I have taped another soldering iron on the edge of the table, moved the PCB over the table edge so that the spot that needs reworking is an inch above the hot iron.

You don't need high temperatures to preheat. Even if you just preheat the board to 100degC, this already more than halves the heat required to do the soldering.

Preheating gives you the control and margin you need. Say, you can heat the PCBs to about 130-140 degC almost indefinitely (well, at least hours) without doing any damage to the components. Now you just need to supply some additional 50-100 degC of temperature to make the solder flow (depending on leaded vs. lead-free), which is now easier with a wimpier iron, or a wimpier hot air gun, without needing to resort to extreme tip/air temperatures to compensate for poor heat transfer.

[flolic]

Thermal capacity of that little PCB is so small that you don't need separate preheater, just swirl around with the hot air gun for a 30 seconds or so and the board will heat up enough.

[ocset]

Thermal capacity of that little PCB is so small that you don't need separate preheater, just swirl around with the hot air gun for a 30 seconds or so and the board will heat up enough.

Thanks, the DPAKs were surrounded by 0805's so the heat gun couldnt have been swirled around the DPAK. It could have been aimed on the DPAK body though.
..........._______________''''''''''''''''''''''''''''''''''''''________________________________

To be honest, getting the FETs off wasn’t the main problem…the problem was getting them back on, in a way that you can be certain that the entire DPAK drain tab area  is soldered to the entirety of the drain pad area on the PCB.


For soldering the DPAK on to the PCB.....There does appear now to be the concensus that  a (preferably) hotplate beneath the FETs would be needed to pre-heat the PCB.

Then it appears that a fillet of solder   and some flux would need to be placed on the drain pad, then the drain tab of the DPAK would need to be tinned. Then the  gate and source legs would need soldering down. Then one needs to hold a hot air gun with wide nozzle over the DPAK body… simultaneously one also holds a 60w soldering iron over the drain tab (for extra heat), then one observes the edges of the DPAK tab….and when the solder kind of looks “wet” and “melty” at the edges of the DPAK tab, then that is the point that you can be certain that the entirety of the DPAK tab area has  soldered to the DPAK drain pad…..Would you agree with this?

…Or should one nudge the DPAK whilst  heating it to check for that “wiggle” that  shows that  the entire area of the DPAK tab has been properly soldered to the entire area of the drain pad on the PCB?
...woops, doh!...obviously i should have known it wouldnt "wiggle" since the  gate and source legs would be soldered and so would fix the drain in place.

[nali]

Considering this was a job interview, I'd probably tell them the task really needed a pre-heater but if you really want me to hack at it I'll give it a go.

I guess we'll never know but were they looking to see your approach to a problem, your skill in making do with what you've got, or just wanted to get a bit of rework done for free?

I presume this wasn't the 200k/yr job? 

[Siwastaja]

Thanks, the DPAKs were surrounded by 0805's so the heat gun couldnt have been swirled around the DPAK. It could have been aimed on the DPAK body though.

Swirl it far enough so that you heat the whole board (or, if it's a massive board, still a significant portion of it! Say at least 3-4 inches in every direction from the area-of-interest).

Adjust the hot gun temperature so that you won't damage the board or even melt the solder yet. If it's unadjustable and too hot, say a general-purpose hardware-shop 500degC gun, swirl it quickly, and from far away.

If you use must use an undersized hot air gun, with nozzle too small, and need to "scan" the board, use higher temperature - say 300 degC - to compensate, but keep swirling it all the time so that nothing gets too hot.

If you have a large hot air gun and a wide nozzle, you could set it to a decent preheating temperature - say just 150-200 degC - and keep warming up the whole area at once. At such low temperature setting, you can take your time, no problem if it takes 5 minutes.

Once you have heated the whole area for many minutes, so you are sure it's evenly warm everywhere, then do localized heating to melt the solder on the transistors. You can use an iron for this, or quickly ramp up the hot air gun temprature, possibly change to a smaller nozzle.

With the large area preheated, you don't need to apply excessive temperatures, or keep the high temperature for excessive time.

For example, without any large-area preheating, you might need to set the hot air gun to 400 degC and still blow it directly at the components for 120 seconds to finally melt the solder. This possibly damages the board or components; in any case, you feel like you lack control. With the board preheated to 150 degC, you may just need 350 degC for 30 seconds, a massive difference.

For the very same reason, reflow profiles (in production) use a slow ramp and a fairly long temperature equalization step at a lower-than-reflow temperature (although you hear the reason being "flux activation", it's not the only reason).

Limiting the thermal stress on ceramic chip capacitors is important, hence the careful ramp on preheating.

[ocset]

Thanks,
The company has the Tenma 21-10125 hot air gun...

This is farnell 2064551.
https://uk.farnell.com/tenma/21-10125-uk/hot-air-station-550w-220v-uk-plug/dp/2064551#

Do you think this is good enough for this job? (hot-airing the DPAKs back on the board)

Also, please can you recommend a head/nozzle for it that would be suitable (or a range of possibles?)

[Siwastaja]

It's adjustable, good enough for the job as long as it works.

For preheating / heating the back of the board, I do that without any nozzles at all. Maximize the preheated area. Keep the temperature under damaging threshold, though. Up to 150degC is usually fine with practically all components and PCB materials even if you work slowly.

[nali]

I had that model. Whereas it works - hot air is hot air is hot air - it's not so great as 1) the airflow is a bit er, industrial for want of a better word, and 2) the diaphragm pump makes for a pulsed airflow which tended to blow away smaller components although 0805s shouldn't be a problem.

Oh, and take the temperature readout with a pinch of salt...

[ocset]

Thanks
This video makes  soldering a DPAK on look so easy. They just  use a soldering iron alone…


However, at 00:26 they swab the DPAK with what looks like a cotton wool bud, which cant be good from an ESD viewpoint?

At 00:39 they use “AIM no-clan gel flux” on the pad, but no solder tinning of the pad whatsoever…(why no solder fillet placed on the drain pad?)

Can any "no clean gel flux" be used here?

At 01:30, they solder the DPAK drain from the end of the tab only, but the solder appears to wet all round the side of the DPAK.
Do you know what power of soldering iron woudl be preferred here?
It makes you wonder how  the solder managed to flow under the DPAK?

I wonder if pre-tinning the pads with solder paste would be best?..then the DPAK could be pressed into the paste?

[wraper]

This video makes  soldering a DPAK on look so easy. They just  use a soldering iron alone…

That PCB is not something you normally see. DPAK on the PCB has piss-poor thermal pad with no vias and non exiting heat dissipation capability. Try doing this on motherboard or graphics card and you will not be able to solder it properly this way, not to say removing old device from PCB with soldering iron with no preheating.

[wraper]

At 00:39 they use “AIM no-clan gel flux” on the pad, but no solder tinning of the pad whatsoever…(why no solder fillet placed on the drain pad?)

Do you mean pre-tin the pad? Why would you want to do this? Then you will need to remove solder from it before placing the device. Otherwise you won't be able to place it flat.

At 01:30, they solder the DPAK drain from the end of the tab only, but the solder appears to wet all round the side of the DPAK.

Solder flows underneath it due to capillary action.

[ocset]

That PCB is not something you normally see. DPAK on the PCB has piss-poor thermal pad with no vias and non exiting heat dissipation capability. Try doing this on motherboard or graphics card and you will not be able to solder it properly this way, not to say removing old device from PCB with soldering iron with no preheating.

..Thanks,  i think you are right..actually, i would do it like this, but have a hot air gun aimed at the fet whilst i did it like the video......then it would be OK?

[Siwastaja]

It's easy to heat up the FET with hot air, the trick is to also heat up the board. Which is easy with so called "minimum footprint" with no thermal vias, as shown in the video, but colossally difficult with an actual high-power design with a lot of layers, large copper pours and thermal vias. There you need to apply heat to the board over large area, preferably from below as well.

[T3sl4co1l]

I would guess (from the thumbnail, I haven't watched the video) that the board in the video has an internal plane, so might not be too unrepresentative.

I find even 2-layer boards, with stitched ground pour, need a fair amount of preheating, that you can't just blast hot air at it and get parts off in seconds.  At least not with conservative hot-air settings.  (Of course, the iron does a fine job in that regard, assuming you can get all joints melted on a part at once.)

Tim

[flolic]

This is my video from years ago.
Desoldering with soldering iron is possible, sometimes...