XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

[nick122]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

@momaka nice! i had over 20 rrod'ed 360s. Most of em had cold solder joints under the gpu but there were a few of them with blown and shorted kzjs

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Okay, I've been slacking a little, since I recapped and tested the second 8600 GT video card literally next day after I made my last post on 10/15. But I promise I will get them in soon (I'll try by the end of the week).

Ok, I'll start "winter session" in the next weekend for my hobby: to repair a nVidia 8500 video card from ZOTAC.
This topic inspired me, thanks!

Glad to hear that!
GeForce 8500 is also a decent video card for H.264 acceleration and YouTube. But for games, it is closer in performance to the 8400. At least it seemed that way to me from my brief tests this summer when I fixed a PC for my cousins.

@momaka nice! i had over 20 rrod'ed 360s. Most of em had cold solder joints under the gpu but there were a few of them with blown and shorted kzjs

Yes, the Xbox 360s have many faults, some of them not repairable. I used to work in a PC repair shop that did mostly Xbox 360s and PS3, so that's why I have so many of the Xbox 360 heatsinks.

[alindumitru46]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Ok @momaka, I found a XFX same like you (for repairing).

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Cool! If the price is right, get it and give it a try.

[alindumitru46]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Unfortunately, I understood from the seller that XFX 8600 card cannot boot into Windows, shows some artifacts then move quickly in black screen.
I noticed a slight discoloration on the back in the gpu zone.
Maybe it can be fixed? Be required a reballing/reflow? Of course, I can not do
a reballing .

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Looks like your 8600 GT above has been recapped already (though I still see one Sacon FZ at the corner).

If the seller says it has artifacts, I do NOT recommend you buy it, unless it's extremely cheap and the parts you get out of it are worth more to you. I say this, because once a GeForce 7/8 series card artifacts, it's usually done . There is a very small chance that a reflow may bring it back, but don't count on it. And reballing, in my opinion, is a no-no on these cards, particularly if you use a suction/vacuum pen to pull the GPU. What I have noticed with GF 7 and 8 cards is that if they are hot and you insert any kind of pressure on the GPU die, it will screw up the chip even more (I had a GeForce 7900 GS go completely short-circuit on all power rails).

So, I'm sorry to be the bearer of bad news, but I think that GeForce 8600 GT is destined for the bin. ... or parts pile.

Also, the discoloration on the back could mean that either the card has ran way too hot at some point or that someone has already tried reflowing it.

[alindumitru46]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Good points, and I was afraid of the same things you have said.
Thanks.

[Drack]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Nice mod momaka, i like the looks of that heatsink must say.

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Nice mod momaka, i like the looks of that heatsink must say.

Thanks. I've actually improved on the mod even further. Just needs a few final touches before I show pictures.

Also have to post results of that second 8600 GT I got.

And if anyone in the USA is interested in getting one of these cards, here's an auction for a 512 MB version of this card going on eBay right now:
http://www.ebay.com/itm/XFX-NVIDIA-G...3D222314166776
Two of the FZ caps haven't even popped yet, so there is a very good chance of reviving this video card with a recap.

[hikaruichijo]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Great, mod, looks very reliable and long lasting.
Those tingtening bolts look like they were instaled at the factory.

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Great, mod, looks very reliable and long lasting.
Those tingtening bolts look like they were instaled at the factory.

Thanks!
Well, I still think it needs more work, but it's too cold in the garage for any work at this time of the year, so who knows when I will get around to doing it.

[Drack]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

One of those projects that never end lol, I demand those pics haha (just kidding)

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Time for an update again.

Anyways, that is all for now again. I've been playing with the 50 mm fan from the original HS and have it installed on the modded heatsink above. Might make a fan shroud and see how that setup works. Running it "open" on the heatsink is okay for the low temperatures in my room now. But I think summer could be problematic, so this needs more testing.

So I did that (added a fan shroud, that is) and got some pretty decent temperatures again: about 56-57°C absolute max under load. And this was in a room temperature of 84-85°F / 29°C. Here are the pictures of the further-modded heatsink:



Also, a temperature curve with an X-axis showing about 13 minutes worth of temperature logging (I actually had the card running for much longer, though, so that it comes up to normal running ambient temperature before doing my testing):

Worth mentioning is that this is an old graph without the fan shroud. With it, I got pretty much the same temperature curve, but with the load temperature slightly lower - mostly sitting at 55°C and only occasionally peaking to 56-57°C. This is with the GPU completely maxxed out under load.

That said, I think the next heatsink revision will be to flip the GPU heatsink 90°C so that the airflow is across the video card instead of taking air from near the PCI/PCI-E slots and blowing it towards the side of the case. But most importantly, I think I need to change that 50 mm fan for something else. Sure it looks nice with that fan shroud on top of it, but it is way TOO LOUD. I also can't turn down its speed, because this video card offers no control of the fan... and even if it did, then I would likely have higher temperatures, which is what I don't want either.

I think the solution would be to find a 60-70 mm high-pressure fan and run it at a lower speed. Makes me wish I had saved some of the recycled iMacs at work - those have pretty decent high-pressure squirrel-cage fans in them. I guess I will play with it more again. Looks like Drack is right - this is going to be one of those projects that never seems to end.

I'm pretty sure the PCI-E 16x slot on my 939Dual-SATA2 is clean. I use that PC almost exclusively for testing video cards. Thus the PCI-E 16x slot sees a lot of "traffic".

Well... turns out IT WAS my PCI-E slot that likely caused the "dead video card" symptoms I mentioned on the first page. After testing various video cards for a year now, I've found that sometimes I have to take them out of the slot and re-seat them a few times before the motherboard recognizes them. So I guess it is no longer safe to say that the PCI-E slot on my 939Dual-SATA2 is fine. But most of the time, it still works, so I am not going to replace that motherboard just yet. And at least I know now, that when the video card appears dead, I re-seat it several times, and that usually fixes it. If not, I test with a known-good video card and then try again. If that doesn't work, I try the video card under test in a different computer (so that's why it's always a good idea to keep a few spare PCs laying about ).

Also, I purchased another XFX GeForce 8600 GT last week, again with bad Sacon FZ caps and all. Should be arriving any day now. So if you like seeing busted Sacon FZ, stay tuned for more.

And a thread for that is coming up in just a little bit.

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

More shenanigans with this video card.

Between last time I posted above and maybe up until January of this year, this 8600 GT video card has mostly been hung up on my wall doing nothing. I might have used it once or twice for an hour in that whole time period, but that's it. Then in January, I dug out my old gaming PC that I used while in university and decided to use that for a bit. I wanted to put a better video card inside it, but something that wasn't too power-hungry nor something that is too weak to play games. Thus, I settled on this 8600 GT as it fit the bill perfectly due to some mods in that PC.

So in mid March, after I had the card running in said PC above, I opened the PC to connect a second HDD and also take a picture of the setup. That's when I noticed this:
https://www.badcaps.net/forum/attach...1&d=1524023055

Yup, that's a bulged Nichicon HZ cap on the GPU core rail ^ there. Looks like my recap job didn't last too long after all.
But there's the reason why: it's not because this video card has something wrong with it, but rather because the failed cap above is a Nichicon HZ 6.3V / 2200 uF cap with a date code of H0532. Funny I noted this, because in February, I also noticed that I had a bunch of other Nichion HZ 6.3V / 2200 uF caps in my stock that had bulged by themselves in storage, and they ALL had H05xx date codes.

I guess that proves it for me: Nichicon HN and HZ caps with H05 date codes are not to be trusted - especially the 6.3V ones (with the 16V, 1500 uF ones from the same Xbox 360 consoles, I've never had one fail).

Now, given what I stated above, you'd probably think that as a rational human being, I would replace that failed Nichicon cap with something better... but you'd be wrong. Wanna know what I used to replace that cap? - Another identical Nichicon HZ. Date code on this "new" one is H0533, IIRC - i.e. only a week newer supposedly. But it read good ESR and acceptable capacitance on my ESR meter (still within 20% tolerance on the high-side of capacitance), so it hasn't gone too [electrically] leaky just yet. Makes good spark, too, when discharged, so I know it's not fooling my meter.

The reason behind this decision: I still have a few more of these "unreliable" HZs with 05 date codes that still read in spec. Thus, rather than throwing them away, I allocated them towards non-critical repairs of my own stuff and/or for testing purposes only.

With that said, let's see how long this one lasts. I imagine the one on the RAM could fail too, as it has the same date code as the failed cap above (they even came from the same Xbox 360 motherboard, based on the markings). But either way, that's not a problem for this video card, because both the GPU and RAM buck VRMs have "backup" caps in parallel with these HZs that can easily handle all the load by themselves even if the HZs fail. Also, it seems that XFX designs their buck VRMs really well, because even when the electrolytic caps go fully open, their cards still work, and often without crashing. Likewise, this XFX GeForce 8600 GT didn't crash either when that HZ cap failed. If anything, I probably did a few gaming sessions on it without knowing.

In conclusion, I think these XFX 8600 GT cards are pretty safe to repair if all they need is new caps. On that note, I couldn't hold myself and got (yet) another one of these cards for cheap. But I will probably post it in my other XFX 8600 GT thread, as it's a bit cleaner from stray discussions (i.e. me babbling ).

[Wester547]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

I told you using Nichicon HZ and HN with 2005 datecodes was risky, didn't I? Out of all your Xbox 360 stock caps, you should have gone with 2700uF 6.3V Rubycon MFZ instead (if you have any left).

I hate to be that “told you so”, “annoying as hell” kind of guy... but I just can't help it. I guess the 16V rated ones last longer because they have a thicker anodic dielectric, so they deteriorate at a slower rate (not dissimilar from the 6.3V vs. 16V rated KZG and KZJ).

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

I told you using Nichicon HZ and HN with 2005 datecodes was risky, didn’t I?

Oh yeah, I knew the risks well by now , especially after seeing so many of my H05xx stock HZs vent in storage (5 or 6 total now, if not more... out of maybe 15-20 caps total).

But considering I pulled many of these caps out around 2012-2013 and kept them in storage since (did only one voltage reform in 2015, IIRC), that's still not too terrible. There are many cheap Chinese and Taiwanese brands that fail faster than this, even when new!

Out of all your Xbox 360 stock caps, you should have gone with 2700uF 6.3V Rubycon MFZ instead (if you have any left).

I think I have only about 50 of those left.
I also have a good number of HZ caps with H06, H07, and H08 date codes - probably about 30-35 in total. Not running out any time soon, especially considering the fact that even a lot of old PC hardware started using more 8 mm caps in the mid to late 2000's.

I guess the 16V rated ones last longer because they have a thicker anodic dielectric, so they deteriorate at a slower rate (not dissimilar from the 6.3V vs. 16V rated KZG and KZJ).

Probably correct.
I guess this thread could double as a running experiment of how long these HN and HZ caps last. The 8600 GT in this thread has one 16V / 1500 uF HN with H05 date code, as does the latest 8600 GT I recapped. I will keep you updated.

[Drack]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Well it looks like I forgot to check this thread for almost 2 years... Hehe
That's a very good looking card, some young PC is going to be lucky to have it on it's pci-e port
Now seriously though, nice mod momaka! I think it even looks too nice to be called a "guetto mod".
Any updates?


Drack.

[Sparkey55]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Well... turns out IT WAS my PCI-E slot that likely caused the "dead video card" symptoms I mentioned on the first page. After testing various video cards for a year now, I've found that sometimes I have to take them out of the slot and re-seat them a few times before the motherboard recognizes them. So I guess it is no longer safe to say that the PCI-E slot on my 939Dual-SATA2 is fine. But most of the time, it still works, so I am not going to replace that motherboard just yet. And at least I know now, that when the video card appears dead, I re-seat it several times, and that usually fixes it. If not, I test with a known-good video card and then try again. If that doesn't work, I try the video card under test in a different computer (so that's why it's always a good idea to keep a few spare PCs laying about ).

I have gotten motherboards with wonky working VGA slots like that too. I first try cleaning them with alcohol and a toothbrush then a squirt of DeOxit and if that is not enough then I manually bend the gold contacts inward a little bit to give more pressure on the gold fingers.

[Drack]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

I think that is actually kinda common with asrock motherboards, i have 2 g41-vs3 rev2.0 that once in a while will do that (one with the pci slot and the other one with the ram and pci-e)

[momaka]

Re: XFX GeForce 8600 GT [PV-T84J-UDD3 V5.0] – recap information and shenanigans

Now seriously though, nice mod momaka! I think it even looks too nice to be called a "guetto mod".
Any updates?

Thanks.

No, I don't really have any updates on this card. I recapped it, used it for a while in the same PC I mentioned above in post #34. Then said PC above started crashing due to other issues - bumpgate problems with nVidia nForce 6150 chipset. Got mad at that PC and gave the chipset heatsink a few "light" punches. Somehow, that got the PC going again... though now the PC is a bit more picky: it will work fine with ATI cards but not nVidia. Any nVidia card will not even allow the mobo to POST. Go figure.

With that said, the XFX GeForce 8600 GT of this thread is back on my video card decoration wall again. It works fine - I tested it in another PC. Might do a volt-mod on it sometime in the future to lower the GPU voltage. I'm hoping to get slightly lower temps. (Not that the card is overheating or anything - I just want to see what the GPU core's limit is for low voltage on the current 620 MHz factory over-clocked clock.) But that will be who knows when. I got a ton more junk cards to fix / play with ATM.

I also posted the 3rd XFX GeForce 8600 GT in the other XFX GF 8600 GT thread, as I promised I would.

Recently, I saw a few more of these cards pop up on eBay again, after seeming to go almost "extinct" for some months. They sold for cheap, too (around $5 per card). Was tempted to get a 4th one, but didn't. Funny enough, the GTS versions actually sold cheaper. And even more funny, I saw 9600 GT and 8800/9800 GT sell for about the same price. So I suppose it's worth admitting that these XFX 8600 GT cards really aren't worthwhile at all to fix (unless you do it for fun, like me).

I have gotten motherboards with wonky working VGA slots like that too. I first try cleaning them with alcohol and a toothbrush then a squirt of DeOxit and if that is not enough then I manually bend the gold contacts inward a little bit to give more pressure on the gold fingers.

That'd be worthwhile doing on a good motherboard. My 939Dual-SATA2 mentioned above is probably beyond saving. It appears its Southbridge has issues, because first the SATA2 controller went out a long time ago. Then the SATA controller followed. Then the audio. Then the LAN, which still works but it can also make the mobo crash when it feels like it. Same goes for the PCI slots - they work and I can still use them for a crude "go/no-go" test on video cards. The AGP slot is okay most of the time now, but it wasn't before. So who knows how long it will work. Miraculously, though, the mobo is pretty stable when used with a PCI-E video card - that is, if it detects it. Seems more and more I have to play with the slot to get it to recognize cards. Then again, I've tested over 60 video card on it by now. Can't blame it for getting a little "loose".

Oh well, I've written that mobo off as worthwhile to fix. When it gets really bad, I'll probably re-purpose it for testing ATX PSUs.