I'm pretty sure this is a known problem. JJP support sent me this pic while troubleshooting a different problem. This pic is not my board, BTW.

Cap_on_Power_Supply.jpg

Just to be clear, the note about R91 is unrelated as far as I know. Frank sent me that pic to show me where to check for the piggyback cap while troubleshooting lamps. It just happened that the board in his pic had overheated.

I must admit, I'm a bit concerned about mine. I haven't looked yet, but I'm fully expecting to find the smaller inductor on it. Waiting for it to get hot enough that JJP will replace it is not the approach I want to take. I really would like if JJP would at least release a service bulletin.

Yours is larger than the ones that we have seen burned up, but it's also not the same as the one on the Rev2 replacement that danisme shows.

The sticker on yours is the part number for a commonly available inductor, so it does give me a lead for datasheets to look at. The value does not match the 14.7uH that is called out on the WOZ schematic.

http://www.digikey.com/product-detail/en/2301-V-RC/M8879-ND/775418?WT.srch=1

It's a Texas Instruments LM5117.

BTW, the RGB LEDs do not affect the 12V source. They have a separate switcher for 7.5V. It looks like the bulk of the 12V is the computer.

This section of the board IS the 12V switcher, but I'm saying that the RGB LEDs are not what is causing it to overheat. It does feed the single color lamps throughout the game (spotlights, coin slots, etc.), but those would be minimal compared to the current draw from the PC.

Yes, they call it the Unified Power Source. I'm not sure how long it was before they started replacing the ATX supply with them, though.

They are using the current source option, but my guess is that the circuit is NOT properly spec'd. I believe the smaller inductor cannot handle peak switching currents at the higher end of the supply's range, it starts warming up, then goes into thermal runaway. There is a 12A fuse on the output that apparently is not blowing, so the surface mount parts flowing off the board are acting as the fuse. If they didn't, I believe the results would be worse than we have seen.

I'm not leaving mine unattended until the issue is resolved. On the bright side, it is contained in a metal box.

I opened my game today, and sure enough it has the smaller inductor. At this point, it doesn't show signs of high heat.

20151109_173557.jpg

Here's another picture 20 minutes after turning the game on, still in attract.

20151109_175536.jpg

That's 127C! Since we don't know the part number of the installed inductor, I don't the what is rated for, but the bigger one I linked to earlier is rated for 105C max.

It would dissipate a bit more, but the bigger factor is that a larger coil will have a lower resistance.

There is not a plastic base on mine. It is just a torroid wrapped with wire.

What you see in the pic is the SSD, but there is a fan in the back left of the box. Having the lid off surely affects it a bit. Having the lid off and the playfield up would allow heat to escape somewhat, though. It's probably a wash.

I ran out of time, but I can give it a try later. I'll put the lid back on, play a game, then open it as quick as possible to take a new reading. If I had a thermocouple, it would be better to take measurements during. Alas, I don't.

Seriously, it's time to stop effing around with this one. This isn't light boards that might lock up.

I put my game back together, glass on. Played a game, failed to melt the witch. Pulled it back apart as quickly as I could, then got a reading of 114C. I'm sure it cooled some as I was pulling the glass and lifting the playfield.

I would guess that yours doesn't get as hot due to the larger inductor.

Nope, but Tom's inductor is different than the Rev2 one. The one on the Rev2 board has two parallel wires, more tightly wound.

That's one of the upgraded boards. Mine is much smaller.

According to the manual, they are using Alpha & Omega AON7244:
http://www.aosmd.com/pdfs/datasheet/AON7244.pdf

The TI LM5117 eval board uses NXP PSMN5R5-60YS, and they claim the eval board can deliver 9A. Here's the datasheet for the NXP FETs.
http://www.nxp.com/documents/data_sheet/PSMN5R5-60YS.pdf

Rds for the two transistors is nearly the same, so the JJP design should not generate much more heat than the suggested design. The AO FETs, however, are much smaller, like you say, and will not dissipate the heat as well. I would agree that it would have been better to go with something with a little more margin than the AO FETs. FWIW, they did not seem to be getting hot in my game, while the inductor was blazing.

It seems like they may have underestimated the amount of current that would be drawn from the 12V source, but then oversized the fuse and set the current sense on the LM5117 higher than the parts can really handle.

Any heatsink would probably help as long as you could transfer heat from the inductor in the sink. Because of the irregular surface, it would be difficult without something like a thermal transfer pad.

Edit: Reading again, it sounds like you actually meant the FETs. The same answer applies, though. It's going to be tough to get a sink to stay on them and transfer the heat from the FETs into the sink. Normally, you need to have big areas of copper on the PCB as Ed mentions.

OK, I had to make a good number of assumptions to narrow down to these, but here is a comparison of likely inductors that could be used. I used the 10uH criteria from the one known 2301-V-RC, approximate size, RoHS compliance, physical appearance, etc. to narrow to these parts. It should be of no surprise that they all ended up being Bourns inductors like the one known part, and they are all similar other than being different physical sizes and having different current capabilities.

I do not know if any of these are actually being used other than Tom's board that has the part number shown.

1-Fixed Inductors Inductors, Coils, Chokes DigiKey - Google Chrome 11102015 94609 AM.jpg

Looking at the pictures, I would estimate that the 0.860" diameter part (2101-V-RC) is likely similar to the smaller inductor. That would put it around 11A current rating, and just over half the capability of the larger inductor on Tom's board. The peak current in the inductor will be higher than the output current, but I haven't done the calculations.

Q20 and Q21 are the hi and lo side FETs, not two regulators. Judging by the heat damage on the board, my theory is that it is the inductor is generating the bulk of the heat, and it's getting hot enough to desolder the transistors. The transistors do not look to me like they are all that damaged, and the board doesn't look as bad on the side of the transistors that is opposite the inductor. I can try to get back in there and measure the transistor temps tonight. I'm also curious about how much current is typically being pulled from the 12V source.

NOTE: If you have any mods connected to the 12V source, they are making the situation worse.

If Q21 fails short, you basically end up with 24V on your 12V output. If it fails open, you have no 12V, like a fuse blowing.

If Q20 fails short, the current sense should kick and and use Q21 to shut things down. If it fails open, I think it would limp along in discontinuous mode, which would generate more heat due to inefficiency.

Is yours using both through holes in the PCB (dual wire)?

I couldn't find an off-the-shelf one wound like this either. The closest I came was a hobbyist website with instructions on how to wind your own. The fact that they had holes in the PCB says they had considered this type of inductor from the start, though. Maybe I'll ask our power supply guru at work.

I feel like I'm harping on the subject, but it has really struck a nerve with me. It's an issue that I feel is dangerous, JJP apparently knows about it, and they have decided to keep quiet about it. If the inductor is all that has changed, it's a $3 part and about 15 minute job to swap it.

Another set of measurements I took this evening:

Q20 - 73C
Q21 - 64C
L9 - 126C

I'm pretty sure all the new ones are fine.

...and you shouldn't feel that way. I haven't went back to read your comments, but I don't remember you ever saying that there was a design mistake or that it was a hazard. I take that responsibility on myself.

I've read LTG's response on the other forum, and I understand his position. He doesn't say whether it's also the position of JJP. I have the utmost respect for Lloyd's experience, but in this case, I think he's making a mistake.

Is it likely that it will cause a house fire? Probably not. Is it a big design blunder? In my opinion, yes. The circuit is obviously overtaxed, and the protections (fuse, current monitor) are inadequate. As an Engineer, it bothers me to see this sort of problem, especially with a known solution, go unresolved.

Is it "super rare"? In this thread, we have identified 5 cases of failure. Without knowing real numbers, I'm going to estimate 500 machines with the small inductor based on the number of machines built during the time frame after the ATX supply and before the problem was identified. That's 1%, and I'm sure there are more. Further, I believe ALL of them are being run out of spec, and it only gets worse the longer they are run. There will be more failures.

I had to go back and look at my posts to be sure, but I never said "fire" other than to say it probably would not catch fire. I do, however, turn off my soldering iron when I'm finished using it. I don't leave it plugged in, inside my game. Granted, your soldering iron is hotter than the melting point of solder (around 280C), but when these power supplies fail, that's how hot they're getting, at a minimum. There are parts flowing off the boards.

I get that you want to defend the game you love and the company that built it. I love the game, too. I want to keep mine and continue playing it. That doesn't mean I'm going to ignore this type of problem.

If I had this problem with one of my products, I would replace them proactively. I truly do think that as they get used more, the failure rate will start to increase. Running the inductors out of spec will cause them to degrade, leading to more heat. Mine is running at 20C over the maximum, and that is in a climate controlled basement.

I'd be happy to schedule an appointment to send mine back and to be out of commission for a couple weeks while they do it. I know this isn't an option for most, but I'd also be willing to change the part myself if I knew they wouldn't end up blaming me for any other problems that happen down the line.

That's a hard question to answer.

First, all we know is what we can see by comparing boards. The inductor is obvious, but other changes may not be. Without JJP telling us what to do, we're taking an educated guess.

Second, the inductor that looks the best (dual winding), I haven't been able to find anything off-the-shelf that even looks like it.

The Bourns 2301-V-RC has turned up a couple of times, and judging by the specs, it does look like a good choice. It could need other parts changed along with it to run optimally. It is available at Mouser:
http://www.mouser.com/Search/ProductDetail.aspx?R=2301-V-RC

Correct. Those are all through hole parts, and WOZ is surface mount.

Just when JJP is promising to do the right thing, one of their biggest defenders muddies the waters again. It's pretty obvious it was a joke, but still...

I guess that confirms it is only the inductor that changed. If I knew the 2301-V-RC is what they've settled on, I'd just add one to my next Mouser order and be done with it. Having them cross reference all those boards and try to track down all the owners sounds like a huge pain.

Mine is a Rev 3 as well, but I get your point. Rev 2 boards may need other changes.

They don't know the contact information of every serial number.

There's now a selection on the support page for the inductor issue. I submitted a ticket yesterday, and they shipped a replacement already.

Have you sent them that pic?

I wonder what this difference is. The physical size of that inductor limits its capability. I would assume the current draw from the 12V source would be the same. A different switching frequency can affect the switching current through the inductor, but likely not all that much.

Pay me to be a thorn in their side? I don't see that happening anytime soon. Really, I think getting ahead of this helps them in the long run, but I still doubt they appreciate the push.

Yeah, it's the bottom half of the UVLO divider.

That's exactly where mine topped out as well, and your thermometer is much nicer than my $15 cheapy. It's possible your inductor is one capable of higher temps, but it would still be heating everything around it.

CPUs and GPUs are much more complex than the inductor, and all parts have their own max temp rating. That said, these inductors, other than the high temp rated ones, have a 105C max.

This argument boggles my mind. SMT parts, in general, are not the problem. At my day job, we are currently designing a RADAR system that is designed to be used in space; shot into orbit on a rocket. We're using SMT parts for the majority.

Back to the JJP design, the part that's failing in this instance is through hole. With all that weight, it is appropriate that it is through hole.

The coil has been underspecified. Did you read Ed's comments?

The SMT transistors appear to be properly specified (Ed may disagree), but may not have enough copper around them on the PCB to properly dissipate the heat. Once I get the new board that is in route to me, I'll take some new temp measurements at the transistors and get a better idea of how hot they are getting without the inductor heating them.

The new board I received has the inductor with a double winding.

20151124_171625.jpg

Much cooler temps after swapping the boards.

20151124_173759.jpg

Inductor - 67C
Q21 - 60C
Q20 - 54C

I'd suggest submitting a support ticket. There is a specific selection for the inductor. I also sent an email first, with no response. The ticket, however, I got a very quick response.

Based on your pictures, yours looks fine to me. Since you have a thermometer, it's easy to test, so that may help you rest more at ease.

Pretty close to what I was seeing. My max was 127C (260F).

The melting point of solder is a bit higher; around 190C (370F).

Oh, and JJP is actively replacing these boards now. If they haven't contacted you, submit a trouble ticket.

Have you tried responding via the ticket? Ask them what it means, and at a minimum, it bumps it to their attention again.

The new boards do have updated specs. There's information in this thread, but basically it's a larger inductor that doesn't heat up as much.

It looks to me like it could be repaired, but it would be tough to know which parts in the area may have been damaged. It's also tough to install those FETs that have fallen off.