Need help powering up a car radio (ntg4 rer / MYGIG Uconnect 730N) on the bench

[SilverSolder]

If you roasted the IC at 430C you probably destroyed it. You really should not be trying to reflow BGAs unless you know what you're doing, and certainly you should practice on some scrap rather than on the device you're trying to repair.

Come on, @james_s....   Live a little! 

[james_s]

Come on, @james_s....   Live a little! 

If there's nothing to lose, then sure, go for it, but in this case:

- Pushing on the IC caused a change in behavior, very strongly suggesting that the problem (or at least a problem) was solder joints.

- There are people out there such as Rossman who are highly skilled at fixing problems of this nature and can do so with minimal risk of additional damage.

- This was a potentially expensive device that belongs to someone else, who entrusted the OP to work on it, probably hoping they could fix it, or at least know when to stop and not make it worse.


Chalk this up as a learning experience. We've all looked back at some point and wished we'd realized we were in over our head and known when to quit before making something worse, but it still sucks. Hopefully the owner of the radio has the attitude that it was already broken and had already decided that it was not worth paying an experienced technician to repair it so nothing to lose.

[dc101]

Well I found by pressing on the main FPGA chip in the unit that I was able to influence if the illumination came on or not.  So I tried to reflow it with one of those Chinese hot air stations but I think I may have done more damage then good as the current draw is now around 1.4A and no lights.  I was not able to get the chip hot enough to even move a little even at 430C and 6 air flow.

For heavy boards with large ground planes, lots of layers etc... you really should invest in a board pre-heater for hot-air rework.

It's hard to say what happened, the chip could be dead, some of the solder balls may have melted and bridged 

[poot36]

I have used 430C on Macbook motherboards before to remove and replace smaller BGA chips with out issue.  One Youtuber I watch uses 480C! and they do not have any issues.  even the small SMD resistors and caps did not move so I doubt that I got it hot enough to cause damage but I am not sure.  I do know that the flux I use can sometimes be partially conductive so I have now cleaned that up as best as possible and am letting it dry in the toaster oven at 60C.

[james_s]

Well the fact that people get away with it sometimes doesn't mean it's a good idea. As someone already mentioned, with something like this you'd typically need to preheat the board, usually somewhere around 100C and then once it's all good and toasty you use hot air to heat the part you're trying to rework enough to finish the job.

Unfortunately I suspect that you're going to have trouble finding a pro willing to touch it after making the problem worse with a repair attempt although you've got nothing to lose by trying. If it is indeed an FPGA then it should be possible to replace it, most FPGAs are programmed at power-up by an external configuration ROM so you can replace them with a blank part of the same type.

[poot36]

Well cleaning the flux and with ISO and drying it out did not help.  The chip is a Xilinx Spartan xa3s1500 so it looks like it does load its configuration from a external chip so that is good.  However it costs over $250 so I think that is off the table.  Oh well.  I also told the coworker the risks of trying to reflow the chip and they were ok with the potential of it getting worse.  It was very intermittent to start with and more often then not would not work at all.

[james_s]

Wow that's a really big FPGA, it's the top of the Spartan3 family IIRC.

$250 where? FPGA prices are really weird, if you go and buy one from Digikey or whatever you can pay a fortune, sometimes more than it costs to find an entire dev board from China. Given the reusable nature of FPGAs you might find some totally unrelated equipment that uses the same part, the Spartan3 is an older series, it was popular maybe 10-15 years ago so a device that has one might not cost a fortune, or you might find some NOS parts on ebay. Nobody is going to design new equipment around such an old part so demand will be limited to someone who just happens to need that specific chip to fix something.

[dc101]

I think the hard part will be trying to find something that uses the same Spartan 3 automotive series. There's still plenty of places trying to offload their ancient Spartan 3 starter kits, like Digilent, but again those are completely different versions of the chip.

[james_s]

Isn't the automotive range the same part, just certified over a wider temperature range? I'm guessing they just bin them according to demand. A non-automotive part won't be guaranteed over the whole temperature range but that doesn't mean it won't work. It's just a radio, not something critical like an ECU or airbag controller. I'm surprised there's an FPGA in there in the first place.

[dc101]

I honestly don't know what the differences are in the automotive series, but I would suspect it's more than temp binning. The industrial 3E is rated 100C which is the standard temp for the automotive series. I didn't read the entire datasheet for the automotive series, but at a quick glance it seems the main differences are the FPGA attributes, system gates, logic cells, user io etc... the XA3S1500 has 1.5M logic gates compared to the XC3S200E (Xilinx Spartan 3E Starter Kit) with 200K logic gates. To your point though, assuming you could find a non-automative series FPGA that had the appropriate attributes, I think you're correct in saying it should work.

[poot36]

Here are some inside shots of the unit, sorry about the low quality the lighting is not very good.

[poot36]

Overall shot.  This unit has GPS, Nav, a HDD, supports DVD playback and backup camera so it would need a powerful processor to do that.  It even has a DSP chip as well.  You can copy 2000 songs to the internal HDD from CD's as well.  It supports MP3, WMA, JPEG files and has Bluetooth for your cellphone with voice activated commands and Sirius radio.  There are 3 RF connectors on the back of the unit.  No anti-theft function though which I found a bit odd.  I am going to try and figure out how to rig up a board preheater and then use my heat gun at around the 350C to 400C range and see if I can get the FPGA to move slightly (I hope...).

[james_s]

Why do you want the FPGA to move? Positioning is absolutely critical for a fine pitch BGA part like that, you don't want it to move, really it needs to be removed completely, re-balled and soldered back in place. I would really suggest sending it out to someone who is experienced at doing this before you completely destroy the PCB. Your chance of success doing it yourself at this point is pretty close to zero.

[poot36]

@james_s I just want it to move a very slight amount and then snap back into place to verify that I have melted all the balls on the bottom.  I have seen this technique online before.  I may end up taking it off anyways as I may have bridged some solder balls already.  I am still learning that is for sure but if you do not try you will never know what you can do.  Even if I wanted to send it out for a re-ball I have no clue who or where I would send it as in my neck of the woods trying to find some place that repairs generic electronics is very hard to find.

[poot36]

Well I used an old electric frying pan as a preheater and got the board up to around 70C to 80C or so and used 400C and airflow 8 on my heat gun with no nozzle and I was able to get the chip to move like I wanted.  However now the current draw is even less (I did not have one of the RF boards connected so that could explain it) so I assume that I will have to take the chip off the board to inspect it.  I also found out that there is another large ST std5700hue (have not been able to find a datasheet for it) part on the under side of the PCB that is slightly overlapping with the FPGA on the top of the board so that chip is also suspect as well.  This is turning out to be an interesting repair attempt to say the least.

[poot36]

Just ordered BGA stencils and some 0.6mm BGA balls as well as some (probably fake) Kapton tape online.  Just have to wait for it to come in and then see what I can do.

[SilverSolder]

Beginning to sound like a bit of an adventure! 

Obviously only by trying it, will you be able to figure this out...

[poot36]

I just hope I ordered the correct size of solder balls as the data sheet for the chip was not very clear on the size.

[poot36]

The solder balls just came in today and man are they tiny, 0.6mm is very small.  Going to be fun placing them when the stencil comes in.  Good thing there are 25000 in the small bottle they came in.

[SilverSolder]

How do they actually get placed - do you just drizzle them over the stencil, basically?

[dc101]

Flux the chip, secure the stencil and then yes exactly, sprinkle the solder balls over the chip (usually over another container to catch the excess). smooth with a flat tool, check for missing solder balls and then use a little hot air to slightly melt the solder balls to secure them in place. remove stencil.

[poot36]

@dc101 I saw that technique on YouTube.  Will give it a try and see how it goes.  Also got in the stencils today so just waiting on the Kapton tape to arrive and this heat wave to go away and I will be able to give it a go.

[poot36]

So I finally got in all the parts to give the re-ball a try and it went went wrong in ways that I would have not imagined, the stencil got stuck to the chip and the only way to get it off was to heat up the solder again and pull it off!  This did make most of the balls that I had placed turn into pointy little peaks so I will have to do the re-ball all over again.  The stencil that I used had 0.6mm holes in in and did line up correctly with the pads on the chip.  I also used 0.6mm solder balls, is that the wrong thing to do?  Here are the steps I did to re-ball the chip.  Cleaned all the remaining solder balls of the removed chip using solder wick and flux and made it as flat as I could, added a thin layer of flux to the chip, placed stencil on chip and secured down with Krapton tape, poured solder balls onto the stencil and moved them around until all the holes were filled, re-flowed the solder balls with the hot air, attempted to take off the stencil only to find out it would not come off and that some of the solder appears to have stuck to the stencil.  Should I have taken off the stencil after I placed the solder balls and then re-flowed it with hot air at a low air flow setting or did I not get the chip hot enough with the hot air as when I did mange to take off the stencil with the hot air the ground and power pads were the ones that turned into peaks indicating not enough heat to flow it into a ball?  Do I need to preheat the chip before using the hot air on it to flow the solder balls?  This has turned into a interesting learning experience.

[poot36]

Well I think it may be screwed, I inspected the FPGA chip and found a missing pad that relates to a I/O pin on the chip.  I have also confirmed that it is connected to something else in the radio.  Not sure what it does but I wonder if it was the original cause of the radio not powering on all the time.

[poot36]

Well I cleaned the excess flux off the chip and found that the pad is still there so it was just the flux covering it that had turned brown with the heat making it look like I was missing a pad.  It is game on again.