spring fit solderless iGBT? Skiip

[coppercone2]

So the IGBT in the machine I am servicing now appears to be a press fit assembly. Its held on through 2 screws into a nylon holder to press it into a PCB with pads that are covered in solder it looks like.

this IGBT blew up and looking at the mechanism its really weird.
https://www.semikron.com/innovation-technology/packaging-technology/press-fit-technology.html

Is this popular? is this good? Its filled with telephone company goo.

My IGBT is not the same as the article, it actually is a SMD device, it has pins much like the new multimeter 9V pins, that press into the exposed solder pad contacts (I think thats what they are). Yes, a big press fit SMD device, no plated through holes required.


 I noticed also, two of the many pads have arc damage, since I ordered a new IGBT, is there anything to watch out for in preparing this contact surface for a new IGBT module? I am kind of thinking arc damage at contacts is a bad sign for this technology...

I have this one specifically,
https://www.semikron.com/innovation-technology/packaging-technology/skiip-technology.html

Could this be the reason why the welder failed? Arcing contact maybe caused some problems.. I would say "well lets solder it on", but get this, its a 3 piece assembly,

so

the semiconductor is press fit into the spring, that press fits into a FR4 PCB board, and its all held together by two screws in the middle to make a sandwich. So if you even did solder it yourself some how, there is still ANOTHER mechanical connection inside the module, so the IGBT has TWO contacts per PIN. Its a 600V 6A unit.

My plan is to clean up the pads with alcohol and grease lightly and place in the new iGBT.

It happened to blow up right where the logo is, so I recommend putting two inscriptions on large power components, I had to put the pieces back together with tweezers like sherlock holmes to read the part # lol

[Berni]

Press fit can actually be more reliable than soldering.

Most connectors in automotive are press fit due to thermal/shock/vibration resistance. A lot of very big >50A PCB mount terminals/connectors also use press fit. You also get copper to copper contact making for a lower resistance connection.

I suppose they can be more susceptible to corrosion due to the thin amount of material involved, but that might be about the only down side.

[T3sl4co1l]

As with any other technology, it must be properly applied.  The first barrier to good press-fit THT connectors is getting precise, finished plated, hole I.D..  Usually the datasheet says something like +/- 0.001", whereas general fab is 3 to 5 thou.  Then they have to be assembled carefully, and probably the connector should be glued in so the pins aren't wiggling around under vibration.

They can be quite reliable, but they're also used at scale.  They're popular for mainframe backplanes in part because you don't have to try and solder anything into the 32+ layer, 1/4" thick board.  The leads don't even stick out the other side, there's nothing to solder to.  But these applications also use thousands of pins, and say they have a 99.9% success rate -- it might sound good individually, but in total, you have a very good chance that one or two pins haven't made it.  What then?

I'm not aware of any power devices with sprung surface contacts.  That sounds kinda dumb.

Tim

[coppercone2]

well I don't believe its better then solder because when I took it off one pad has 4 arc spots on it, its galled/melted from the 4 spring contacts that bring that 'foot' of the IGBT down to that pad. It must have developed a high resistance.

What is the specification for a SMD power spring fit pad? I do have to restore it. the whole thing of using the PCB directly as a contact seems dumb. the pads are also not gold plated, maybe thats a good thing for repairs, but they appear... tin plated? Is that even good? You would think they would sell a nice base plate that you solder to the PCB then insert the IGBT into that, or at least have solder in receptacles. Why does it go directly into the PCB? Thats so janky. The design in general does not feel 'low inductance' either. I guess the springs are flat and not helix which helps, but still.

I can re-tin it with an iron, and I can also clean it up good with a solder wick.. but is that surface good enough for the semikron spring contacts after my restoration? I can also try to take the pad down to bare copper with unionized abrasives (rubber abrasives), maybe, and then plate it myself, either with rub on silver or some kind of electroplating pen sponge thing. Do you think retinning, solder wick mop job is good enough for a contact? It looks too dull to be nickel on the pad.

This whole deal seems bootleg for a tig welder.  Maybe I am not getting something, but tin is just as soft as gold, and 'digging in' won't really hurt the igbt, the spring tips are rounded and free, if it shifts a few thousanths to 'pull them side ways' i don't think any sort of plating would get in the way and for some reason its difficult to imaging something bolted on with 2 1/4 inch bolts sliding around gracefully between the surfaces . I think if it was gold plated maybe this fault would be avoided, or if it had a real receptical thats not a PCB pad covered with tin.


most multimeters use gold plated contacts. I don't see why you can't use gold here. it seems like the same concept. I almost want to get a gold plating brush and plate the pads myself

[T3sl4co1l]

No, solder will not restore it.  It's junk, replace the assembly.

Tim

[coppercone2]

I am not replacing the backbone PCB board, its too expensive. each of these PCB is quoted at >2000$

there must be a way to get a bit of arc damage off the contact pads. It's not like conductivity is lost there, it just looks shitty in comparison to all the other ones. If I just cleaned it with alcohol and a cotton bud it would be kinda OK, but shitty compared to the other pads. I don't mean repairing the IGBT, I mean the PCB.

[coppercone2]

I ground down on it with rubber abrasive dremel tips till it was bear copper and looking clean of the defects, then I cleaned it up good and rubbed in some kool amp silver plating powder to plate the pad and swabbed it with deoxit. hopefully its ready to accept the new IGBT now

I also used the lightest mesh of abrasive rubber to clean up a few of the other pads, in this case in the form of a soft deformable cone shape, as to keep the cleaning cosmetic/electrical rather then highly abrasive. they are dental abrasives called 'greenie', just make sure you have the collet to fit the odd shank size.. can also put it in a proxxon mini mill to make a final soft 'polishing pass' on stuff to clean it up a little. the 'brownies' are alright for minor PCB charring and surface soot removal also.. but all very mild, the non dental rubberized abrasives are much tougher, so for non surface contact, use that if you are loooking at it, it can even debur stainless steel cuttings but it has a short tool life, however excellent for cleaning holes. I think I will use the greenie to clean up some screw terminals and crimp rings.

[coppercone2]

the square on the right is the restored one, they are all silver colored but the camera makes it copper for some reason. it had 4 weldaments that were polished off, only a very slight shadow of the outline remains, and its all silver plated now with kool-amp. 

with flash on

for a before picture, imagine a spot welded 2x tab ripped off a 18650 or other cylinderical cell

not the prettiest contacts ever but I think they will be OK now.

[T3sl4co1l]

From here looks more like it's just smut from the IGBT going explodey, but yeah.

Tim

[coppercone2]

those are both after cleaning, before cleaning you can feel the copper deformation on a fingernail on that pad, now its cleaned up.

[WattsThat]

The company I work for makes extensive use of press fit IGBT’s in variable frequency drives. They use them on units up to about 150 amps. The power boards are considered disposable since when the IGBT’s fail, the collateral damage is usually significant enough to scrap the board, especially so since they always blow the elephant snot everywhere when they let go.

When you build thousands and thousands of boards per month, the labor savings is significant. The reliability is better as well, especially with respect to shock and vibration.

[coppercone2]

damn I don't mind disposable boards for disposable prices but this price is heirloom

but most people don't want to pay for repair facilitation I guess, thats the usual


is there a reason gold contact surfaces are not used? Why is it left silver

[Berni]

Gold plated pads are mostly only used on high density boards that go into BGAs and similar. They need the precise pad flatness that it offers. Most simple low density boards just use HASL due to being cheaper. The gold won't really offer that much long term corosion protection because its a very thin layer that easily gets damaged. For proper protection you would need to do the so called "hard gold plating" that is used for making contact pads on things like card edge connectors, FPC cables etc.. But this is even more expensive of a process while also requiring all the plated parts to be electrically connected since its done using electroplating.

As for blown power electronics being treated as disposable is not all that ridiculous. Quite often large power devices blowing up send a violent pulse into the gate/base terminal, this then can cause significant damage to the drive circuitry. If things get even uglier with a arc flash this could also lick the board and blow stuff up, even the carbon deposits from something burning in the power device could cause an arc over at the driver board and blow it up. It could take quite a bit of troubleshooting and component replacement to get a blown driver board back to life, so at some point the repair technicians time ends up costing more than the BOM cost of the board. And even if you do fix it you end up with a visually unappealing charred board that might not be quite as reliable anymore.

The issue is that these more specialized products are sold with massive markups, so while they will sell you the board for 2 grand they likely put it together for 10% of that. They see the technicians time more valuable than the board, while you would gladly spend days troubleshooting it and replacing components if it means saving 2 grand.

[coppercone2]

yes, i replaced over 50 components for like $50

I found that pace resistance soldering tweezers can pull cut IC legs off a PCB after its cut out. super fast, then you just use the vac to clean the hole. Maybe a problem because of zapping parts but if you want to replace everything its real fast.

 I have yet to try the pace microchine accessory to fix badly damaged multilayer boards..

[T3sl4co1l]

Tin plate on tin plate is fine, most power connectors do that.  Gold is mainly worthwhile when wiping current is not available, i.e. for signals.

Mixing both is unwise, leading to fretting corrosion.  I'm guessing the device leads are tin plated.

Tim

[ajb]

well I don't believe its better then solder because when I took it off one pad has 4 arc spots on it, its galled/melted from the 4 spring contacts that bring that 'foot' of the IGBT down to that pad. It must have developed a high resistance.

Depending on the failure mode, it's not necessarily the case that the spring contacts would have failed before a soldered joint would have.  It's possible that the initial failure caused localized heating that reduced the spring force on the contact, resulting in the arc/heating damage you saw, and the catastrophic failure would have been a result of that electrical contact eventually going open circuit (or high enough resistance to cause similar effects).  A similar situation might have caused a soldered joint to just de-solder itself, or the presumably higher thermal conductivity of the solder joint (versus the spring contacts) might have transferred enough heat into the board to damage the board itself.  In either case, once the device starts getting that hot it's already doomed.