I got a batch 30 of IBM/Lenovo Thinkcenter boards in for repair from a great metropolitan newspaper.....Yes, that is true, I just won't reveal the city, as it'll give away the client (and it is a large well known city).
I've repaired a lot of boards with cooked MBZ and MCZ caps over the years (mostly overheated Dells), but this was the first I've encountered this issue. Their IT guy was pulling his hair out, he told me only a few of them had bloated caps, but all of them either wouldn't POST or wouldn't finish booting the OS before rebooting or BSoD'ing...even the ones with caps that 'appeared' perfect. He already replaced PSU's and eliminated that factor...so they show up here....
I first took one that actually had a couple bloaters on it, right next to a couple FETs in the middle of the board, and replaced only the bloaters, and then bench tested it. The board would POST, but would freeze under MEMTEST in about 2 minutes....so I tried the same thing with another one, same result. There was one last board left that had a couple bloaters, I replaced the bloated caps only, and that one wouldn't even POST. ...so now what...
I started pulling random caps of the same value of the bloaters and checking them. I was bewildered at the results!! These are so far out of spec, I thought my Sencore had gone screwy, so I checked it with a new cap just to make sure that wasn't the case, and it wasn't the tester that was out!!
All these caps datecodes range from 2006~2008. They are genuine, not fakes, and original to the boards. They have never been recapped.
They were all 1000uF 10v Rubycon MBZ Series:
You will note that the bung is flat, no leakage or damage:
Same with the vent, it's pristine:
Ok, here's where it gets interesting!! The value is SUPPOSED to be 1000uF! it's reading 2.78uF
LOL @ 27 OHMS ESR:
I pulled several hundred of these over the course of a couple days, this is just one cap that I profiled in this article. The rest were just as screwy. I found there was no pattern. Some were so far out, the LC102 wouldn't even get a reading, or show an open load. They would come from random/different sections of the board, clearly they didn't overheat, yet they were worse than a teapo as far as being out of tolerance but not leaking/bloated!!
The aftermath:
I recapped all the boards completely removing ALL their MBZ's and replacing them with Nichicon HM's of the same value. The boards now run flawlessly, and are back in operation. I did check the other caps on the board, there were some Rubycon YK 220uF 16v and a couple Rubycon 470uF YK's as well. Those were all well in spec, it was just the MBZ's. Scary trend for the recappers that don't have the equipment to accurately check a capacitor. Every failed Rubycon I've ever dealt with in the past had vented/bloated/leaked.....
I've repaired a lot of boards with cooked MBZ and MCZ caps over the years (mostly overheated Dells), but this was the first I've encountered this issue. Their IT guy was pulling his hair out, he told me only a few of them had bloated caps, but all of them either wouldn't POST or wouldn't finish booting the OS before rebooting or BSoD'ing...even the ones with caps that 'appeared' perfect. He already replaced PSU's and eliminated that factor...so they show up here....
I first took one that actually had a couple bloaters on it, right next to a couple FETs in the middle of the board, and replaced only the bloaters, and then bench tested it. The board would POST, but would freeze under MEMTEST in about 2 minutes....so I tried the same thing with another one, same result. There was one last board left that had a couple bloaters, I replaced the bloated caps only, and that one wouldn't even POST. ...so now what...
I started pulling random caps of the same value of the bloaters and checking them. I was bewildered at the results!! These are so far out of spec, I thought my Sencore had gone screwy, so I checked it with a new cap just to make sure that wasn't the case, and it wasn't the tester that was out!!
All these caps datecodes range from 2006~2008. They are genuine, not fakes, and original to the boards. They have never been recapped.
They were all 1000uF 10v Rubycon MBZ Series:
You will note that the bung is flat, no leakage or damage:
Same with the vent, it's pristine:
Ok, here's where it gets interesting!! The value is SUPPOSED to be 1000uF! it's reading 2.78uF
LOL @ 27 OHMS ESR:
I pulled several hundred of these over the course of a couple days, this is just one cap that I profiled in this article. The rest were just as screwy. I found there was no pattern. Some were so far out, the LC102 wouldn't even get a reading, or show an open load. They would come from random/different sections of the board, clearly they didn't overheat, yet they were worse than a teapo as far as being out of tolerance but not leaking/bloated!!
The aftermath:
I recapped all the boards completely removing ALL their MBZ's and replacing them with Nichicon HM's of the same value. The boards now run flawlessly, and are back in operation. I did check the other caps on the board, there were some Rubycon YK 220uF 16v and a couple Rubycon 470uF YK's as well. Those were all well in spec, it was just the MBZ's. Scary trend for the recappers that don't have the equipment to accurately check a capacitor. Every failed Rubycon I've ever dealt with in the past had vented/bloated/leaked.....
Comment