Biostar N7NCG 400 v7.2

[BigTroll]

Hello all I was planning on recapping my N7NCG 400, the VRM section is what has me concerned. It has 4 3300UF 6.3 OST RLX capacitors and 3 2200UF 16V caps also RLX. I was wondering if I can go 10V caps for the 2200UF I wanted to do some older nichicon HM for this VRM. The last thing I want to mention is the 2200UF 16V caps look to be OST black RLS on other boards and I thought the RLS wasn't a ultra low ESR cap so I thought maybe I could get away with regular low ESR for the 2200 caps like a panasonic FM or something.

[BigTroll]

After doing research I decided to try the new Panasonic FS 2200uf 16v in place of the OST RLX. Will update when they get here and have pics of the before and after recaps. otherwise I got NOS Nichicon HM for the rest of the board.

[momaka]

May we see some pictures of the board? The images I'm getting through searches on the internet are being inconsistent. If that board has a 4-pin 12V CPU connector, then NO, you cannot use 10V caps to replace 16V caps, because it means those 16V caps are likely connected on the 12V rail.

And yes, Panasonic FS should work fine for the recap - both for the OST RLX and RLS. RLX is supposedly motherboard-grade ultra-low ESR, but their specs trail those of Nichicon HM and Rubycon MBZ by a fair bit. Meanwhile, series like Panasonic FM and FR are not that far behind the RLX specs. So in a pinch (especially now that mobo-grade ultra-low ESR caps like HM/HN/HZ and equivalent are no longer made), Panasonic FM, FR, and FS should work alright. I think the Rubycon-equivalent of those are the ZLS. For Nichicon, it should be HW. All are still active parts, as far as I know.

[Dan81]

Originally Posted by momaka

May we see some pictures of the board? The images I'm getting through searches on the internet are being inconsistent. If that board has a 4-pin 12V CPU connector, then NO, you cannot use 10V caps to replace 16V caps, because it means those 16V caps are likely connected on the 12V rail.

And yes, Panasonic FS should work fine for the recap - both for the OST RLX and RLS. RLX is supposedly motherboard-grade ultra-low ESR, but their specs trail those of Nichicon HM and Rubycon MBZ by a fair bit. Meanwhile, series like Panasonic FM and FR are not that far behind the RLX specs. So in a pinch (especially now that mobo-grade ultra-low ESR caps like HM/HN/HZ and equivalent are no longer made), Panasonic FM, FR, and FS should work alright. I think the Rubycon-equivalent of those are the ZLS. For Nichicon, it should be HW. All are still active parts, as far as I know.

I googled it and it doesn't have a 12v 4 pin CPU connector.

BigTroll: You can measure the voltage on the places where 16v OST caps were and check the voltage going through. Most likely it's an design similar to ASUS' A7N8X-E Deluxe where they got away with an boatload of 3300uF 6.3v caps and no 4pin 12v connector.

[BigTroll]

Hi guys so I did order the 2200uf 16v pannasonic caps that are 10mm. I finished recapping the board but have to wait on mouser to ship the panny caps and they are taking forever.

[BigTroll]

The recap seems like a success so far 2 passes in memtest where it kept erroring out before, will let it run all night and see how she does. Its a athlon XP-M overclocked to 2.4ghz with 2gb ram, 200gb WD IDE, DVD-RW, ATI X800 PRO.

One More thing I want to mention using a temperature gun, the OST caps were running 100-110F when taking the temps, the new nichicons are only 89-95F or so. I took the temps while the system was running memtest for 30 mins both times.

[ChaosLegionnaire]

the jap caps can hold and store energy significantly better than the osts so the regulators and mosfets need to work less harder to keep up with the demands of the switching circuit, hence the cooler temperatures.

[Wester547]

Or maybe the OST’s parameters (ESR, capacitance, leakage current, etc...) just degraded over time (which is quite possible given the previous memtest errors), if the specs of that batch weren’t exaggerated to begin with. I take it those older Nichicon HMs were made after 2004 or 2005. I think the low temperatures are probably because of the low TDP (albeit overclocked) Athlon XP-M.

[momaka]

Originally Posted by BigTroll

The recap seems like a success so far 2 passes in memtest where it kept erroring out before, will let it run all night and see how she does. Its a athlon XP-M overclocked to 2.4ghz with 2gb ram, 200gb WD IDE, DVD-RW, ATI X800 PRO.

Nice recap job and awesome classic rig!
I like how you have a red AMD motherboard, red ATI video card, and... red LED fan. Go Team Red!

BTW, that ATI X800 Pro looks like the old gen, before ATI started using a separate AGP bridge chip. Make sure to keep it cooled well (might want to invest into a bigger heatsink). Otherwise, it'll probably fail over time.

Originally Posted by BigTroll

One More thing I want to mention using a temperature gun, the OST caps were running 100-110F when taking the temps, the new nichicons are only 89-95F or so. I took the temps while the system was running memtest for 30 mins both times.

Not to say your measurements are not true... but all I will say is that I don't usually trust those laser temperature "guns". Depending on the color and surface texture of the material (i.e. dark vs. light and smooth/reflective vs. rough), the temperature can vary quite a bit. Granted cap sleeves are typically pretty smooth, so there likely won't be much error in those readings (at least in the temperature differences). Still, I rather use my type-K thermometer.

[BigTroll]

Originally Posted by momaka

Nice recap job and awesome classic rig!
I like how you have a red AMD motherboard, red ATI video card, and... red LED fan. Go Team Red!

BTW, that ATI X800 Pro looks like the old gen, before ATI started using a separate AGP bridge chip. Make sure to keep it cooled well (might want to invest into a bigger heatsink). Otherwise, it'll probably fail over time.


Not to say your measurements are not true... but all I will say is that I don't usually trust those laser temperature "guns". Depending on the color and surface texture of the material (i.e. dark vs. light and smooth/reflective vs. rough), the temperature can vary quite a bit. Granted cap sleeves are typically pretty smooth, so there likely won't be much error in those readings (at least in the temperature differences). Still, I rather use my type-K thermometer.

So ive had some issues, My computer would not get through 3d mark 03 it would lock up with the ATI card, it would make a few passes through 3d mark 01 but would eventually lock up after that. I figuered part of the problem could be the nforce2 AGP Gart driver not playing nice with ATI, so I put in my best nvidia card I had left a QUADRO FX 1000, it made it through 3D mark 03 no problem and looping 3D MARK 01 it would last about 2 hrs then crash with a message about not having enough vram but the computer was not hard locked.

Was hoping someone had some suggestions.

[Sparkey55]

What Operating System are you benching on?
Are you using 3DMark 2001 or 3DMark 2001SE?

[BigTroll]

Benching on a fresh install of XP Service Pack 3 with no updates done to it. and using 3DMark 2001SE. I will try and let it run again so I can get a picture of the error message.

[momaka]

Is AGP fast-write enabled in BIOS on your motherboard? Sometimes, I've had that cause issues with AGP video cards. I haven't noticed any loss of performance with it turned off for high-end cards either, so I usually keep it off now.

Also, what about "Cache Video Memory" (or similar option) in BIOS? Try disabling that too, if you have it.

If bad comes to worse... why not play some oldschool games on it for a few hours to see if it would crash? If it doesn't crash then, then I'd say the system is "stable enough".

Another idea: maybe try OCCT or some other similar CPU-only (or CPU + memory) stress test. If the system crashes with that, then most likely your GPUs weren't the issue.

[ChaosLegionnaire]

Originally Posted by BigTroll

I figuered part of the problem could be the nforce2 AGP Gart driver not playing nice with ATI, so I put in my best nvidia card I had left a QUADRO FX 1000.

when jumping ship with video cards, its also important to use a third party driver uninstaller like driver cleaner or DDU (display driver uninstaller. requires .net lamework 2.0) to wipe all traces of the previous video driver in the registry and filesystem. otherwise the drivers from the opposing video card maker wont work properly due to conflicts with each other. i think nvidia and ati/amd do this deliberately to stop people from jumping ship. sooner or later, someone is going to catch on to this shit that they are doing and file a class action lawsuit for anti-competitive behaviour/practices...

[momaka]

Originally Posted by ChaosLegionnaire

when jumping ship with video cards, its also important to use a third party driver uninstaller like driver cleaner or DDU (display driver uninstaller. requires .net lamework 2.0) to wipe all traces of the previous video driver in the registry and filesystem. otherwise the drivers from the opposing video card maker wont work properly due to conflicts with each other. i think nvidia and ati/amd do this deliberately to stop people from jumping ship.

Nah, I'm pretty sure there is no such a thing.

On my video card test PC (yes, I have a PC dedicated for testing video cards ), I keep both nVidia and ATI drivers running simultaneously. The nVidia drivers are actually pretty smart and don't start up if there is no nVidia video card. Meanwhile, the ATI ones are always running and often bitching if they encounter a non-ATI card (or one they don't recognize, because it's too old or too new)... but nevertheless, they don't stop the nVidia drivers from running if there is an nVidia card in the system. Likewise with nVidia drivers.

Thus, I rarely have to do any driver installations when testing various video cards. Most of the time, the already-running ATI and nVidia drivers pickup the video card after first boot, then install the drivers (after some 5 minutes of waiting due to slow HDD in that PC), after which I reboot the PC and everything is running properly with the right driver. What's more interesting is that neither the nVidia nor ATI drivers care if I am testing AGP or PCI-E 16x or PCI video card (and the motherboard in that test PC has all three.)

[BigTroll]

I have had to use driver cleaner pro or something to remove some bad nvidia drivers version 81.98 or something, they would crash screen savers in windows 98 and nvidia's installer would not removed them.I think it might be a good idea to try that since I used about 5 different versions of catalyst on the system.

So update, I got some used ram off ebay, nicer Kingston with heatspreaders. running 2GB with 2.5-3-3-6 timings and running memtest but they seem to be doing okay. I also bought a cheap FX5700 off ebay. I really would like a 6600GT or 6800 but anything high end AGP is outragous in price. So i'll try running 3d mark looping on it after i get it in.

I did play colin mcrae 2005 on it with the original ATI card for 30 mins or so with no problems, heck that X800PRO even pushed it at 1920x1080 on a newer monitor I was very impressed. no way a Geforce FX could do that...

Also wanted to mention I did turn off AGP fastwrite in bios and I don't think I am caching the video bios or anything, I set AGP aperture to 256 if the card was 256 or 128 if I was running the 128mb nvidia card however I could change it to 512. I am honestly not sure what would be the best aperture setting.

[momaka]

Originally Posted by BigTroll

I also bought a cheap FX5700 off ebay.

Oh, man, I hope it's not one of those Chinese variants. Not that there is anything wrong with them in terms of performance. But they always seem to come with Sacon FZ capacitors - some of them even ventless Sacon FZ!

Originally Posted by BigTroll

I really would like a 6600GT or 6800 but anything high end AGP is outragous in price.

6600 GT you can still find for around $15 shipped (in the USA) occasionally. Just have to be patient and spot an auction that no one else bids on. 6800 GT is a mixed bag - sometimes they go for a more "reasonable" price (under $30), and often times not (typically twice that).
Meanwhile, ATI seem to be cheaper. In fact, for the price of a "cheap" 6800 XT or GT, you can typically get a new HD2600 or HD3650 from China. Both the HD2600 and HD3650 will come close or even beat the 6800 in many games. Only problem with these new Chinese ATI cards... they too will come with Sacon FZ most likely.

Otherwise... yeah, prices on high-end AGP cards - especially nVidia - regularly go through the roof. Funny, though, nowadays the old stuff like GeForce 3 and 4 TI series sell for even higher prices than some other high-end AGP video cards (I've seen a few go for even more than the GeForce 6800 cards. )

Originally Posted by BigTroll

I did play colin mcrae 2005 on it with the original ATI card for 30 mins or so with no problems, heck that X800PRO even pushed it at 1920x1080 on a newer monitor I was very impressed.

And considering those old CMR series are better optimized for nVidia... that is indeed pretty impressive from the X800 card.

[Sparkey55]

If you want to bench on an PCI-E mobo get a Dell workstation card like the FX3500. It is equal to a 6800 and can be had real cheap.

[Sparkey55]

Originally Posted by BigTroll

So ive had some issues, My computer would not get through 3d mark 03 it would lock up with the ATI card, it would make a few passes through 3d mark 01 but would eventually lock up after that. I figuered part of the problem could be the nforce2 AGP Gart driver not playing nice with ATI, so I put in my best nvidia card I had left a QUADRO FX 1000, it made it through 3D mark 03 no problem and looping 3D MARK 01 it would last about 2 hrs then crash with a message about not having enough vram but the computer was not hard locked.

Was hoping someone had some suggestions.

What PSU are you using for graphics card benching? Check the caps.

[BigTroll]

Originally Posted by Sparkey55

What PSU are you using for graphics card benching? Check the caps.

I got a Enhance ATX-1136F its a 360 watt it was new old stock, I did open it and found jamicom and teapo it looked to be a 2006 build, I ran it on a PSU tester before i put it in the system. it seems to be doing fine but I will check the caps soon.

I also got the FX5700 its a ASUS V9570, lubed the fan it had small rust spots on the dvi connector and looked like it was in a bit of water at one point but It ran 3dmark01se 700 plus tests before it locked up but that pretty good, it lasted at least 6 hours in 3dmark before crashing. I will continue to work on this system. and I appreciate your feedback everyone.