DFI Lanparty B / Ultra Infinity (nF2)

[trodas]

I have two DFI boards and caps on them keep getting worser and worser, even the major ones are Chemi-con KZG ones, known as good brand - they end up badly. There are some speculation, if this is due to the mainboard design and "crazy level of noise" the mobo generate, but for whatever reason this happens, it does happen.
Take care and replace your caps with something worthy. Especially beware users of Taiwan made DFI mobos, as they clock better and so on, but use Teapo caps almost everywhere and Teapo is know as bad brand of caps!

And now finally a explanation why and how does the Taiwan and China boards differ.

First at all, let me say that all the time DFI fierce fully deny that there is any difference. Now I got my hand on Taiwan made mobo, and having two China made ones already, I spotted quite a few differences. First at all, the CPU mosfets is slightly different. On both mobos are there two series by 3 for each one of them. The first is 09N03LA and the second ones (closely to the CPU) are 06N03LA. But on China mobo are these 09N03LA - BC335 and 06N03LA - AN330, while on Taiwan mobo they are 09N03LA - CG339 and 06N03LA - AA330! And even by the very first look, they are different - the 09N03LA - BC335 on China mobo have a bit rounded edges, see there:
http://ax2.old-cans.com/s.php?p=badt...6&c=8&d=1&v=v2
...while on Taiwan mobo they both having sharp edges.
So, there is visual AND a little specification difference as well. Now let's move to the Vdd/Vdimm mosfets. As expected, the differences did not ended up with Vcore mosfets, but the very same differences are in these as well. China mobo use (looking at the image link bellow, from left (Vdd mosfet) to the right (3 Vdimm mosfets) 09N03LA - BC335 and 3x 20N03L - VF328 while the Taiwan one use probably - mosfet destroyed, anyone could help there - 09N03LA - BC335 (?) and 20N03L - XR320, 20N03L - VB320, 20N03L - XR320.
http://ax2.old-cans.com/s.php?p=badt...d=1&id=137&c=8

While some might say that these are quite small differences, it is indisputable that these are DIFFERENCES in fact, witch prove DFI lying about it. (no wonder at this point, the manufacture policy surely will deny any differences on principle, it's logical...)

And now for the main event. At the very first glance at the Taiwan made DFI board I noticed the (except the Made in Taiwan is not blanked with white anymore) caps are different. Yes, the 5 main ones (Chemi-con KZG ones, 3300uF 6.3V) and the there 1500uF 16V by same brand are the same, but the rest is different from what I have on my China ones. Upon a close inspection I found to my HORROR, that these caps are Teapo ones! Now for these who aren't familiar with Teapos - they are known as bad caps brand. Definitively bad. They die. Theirs good specs in short time get worse and worse and... and the mobo become unreliable. For anyone think that I was not right, then see there:
The Hall of Shame - Badcaps Photo Montage
Teapo - https://www.badcaps.net/forum/showthread.php?t=424
List of Bad Cap Manufacturers
https://www.badcaps.net/forum/showthread.php?t=388
Teapo - http://www.teapo.com.tw


God help the mobo...

Now conclusion - what all this means? Well, since Taiwan mobos are praised for their overclocking results, it obviously mean that the choice of the components for them was better - at least for the time being. ST caps specs can't stand against Teapos (well, for the time being it get reversed, but at first... amazing OC was possible!), witch explained the overclocker rumors about better Taiwan boards.

So, recommended action - replace immediately ALL the Teapos with some good brand caps (Samxon GA (when become available), Nichicon HZ, Rubycon MCZ, Chemi-con KZJ, Samxon GC, Nichicon HN, Panasonic FM) and then just enjoy your board back on rock-stable and reliable operation.
Even that I did not find anything better that the Panasonic FM ones (http://digi-key.com has them, still looking out for either Nichicon HZ or Rubycon MCZ for extreme OC), my mobo started to work reliable and as soon, as I add also the blocking caps, I try some OC and post results.

But definitively and finally, the mystery is solved. Taiwan mobos are at start notably better for OC that China ones. Later they suxx much more that China ones, since ST aren't bad brand of caps, though their specs suggest why the China ones clocked lower that the Taiwan ones did.

If anyone can provide me better caps, I still have 2 DFI mobos waiting for recapping, we can measure the differences. Even if I have to buy a oscilloscope

[RJARRRPCGP]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Over at ocforums.com, some people had bad luck with DFI motherboards, usually the Athlon 64 ones. One person says that his DFI motherboard keeps on killing PSUs!

http://www.ocforums.com/showthread.p...&highlight=PSU

[gonzo0815]

Re: DFI Lanparty B / Ultra Infinity (nF2)

This seems to me more that there are bad caps in the PSU, or the DFI Board s are lacking sufficient input filter caps so the PSu caps are exposed to high ripple current. As there are at the moment no fortron PSU affeckted i think that the problem is bad desing or bad caps /to few in the psu as usual. And if the DFI board drains so mutch current on the 5 and 3.3v if those Vcc jumper is set, then modern ATX 2.0 PSU´s are not build for that kind of current pattern. I think if, you want to do such extrem things the DFI board is build for, then you should realy have think twice what PSu can sustain that. It probably is more than go out an purchase an expensive 600w PSU. Personally i think that OCZ and others desinged even thos expensive units wit a low 5 and 3.3v line, to save money as they thought on ATX 2.0 mobos you never need that mutch current from those lines. Unfortunately the DFI board want`s both, mutch on 12v and a lot of 5 and 3.3v.
Fortron is IMHO different, but however newer FSP`S dont have that mutch amps on that low voltage lines too. But FSP are in generall underrated and probalaby can manage it to deliver the current pattern DFI needs. But on the long run, i think that those FSP units will develop bad caps to. So my suggestion would be to use high power units from FSP or Amacrox, think they will be the right units for those DFI boards.

[Chris1992]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Or a Zippy, I recall seeing one with 3.3@35, 5@50, and 12@32.

[CaysE]

Re: DFI Lanparty B / Ultra Infinity (nF2)

trodas, I hate to break up the party here, but these are probably revision differences, not manufacture facility differences. My Chinese board, rev AA0 uses Teapo caps as well (the mosfet models match your description, though). My Taiwanese AC0 used Teapo caps, but I can't confirm the mosfet models since it has been RMAed. I will be receiving an AD0 (unknown origin at the moment) on Tuesday directly from DFI, so I will let you know what I find.

You will also notice that later revisions of this board (AC0 and later, at least) have a capacitor soldered between two of the VDIMM mosfets. Please post what revisions your boards are, it is on the sticker on the LPT port.

Attached is my Chinese board, as you can see by the whiteout. The "Made in China" sticker is on the bottom of the board, as well. It is Revision AA0, and does not have the capacitor between the two VDIMM mosfets. Every cap from 100uf - 1000uf is Teapo. (Don't mind the broken IDE connector... I bought it like that. )

[trodas]

Re: DFI Lanparty B / Ultra Infinity (nF2)

RJARRRPCGP - and it is not all, since the DFI mobos has other issues as well. Todays I step upon this one:
http://forums.mushkin.com/phpbb2/viewtopic.php?t=2564

CH-UTT chips are having problems to start properly when they don't get enough voltage. DFI NF4 motherboards sets the voltage automatically down to ~2,53V when you're starting your system, which in this case are not sufficient for the Redlines to start operating correctly.

Uh, oh.
And the unpaid public beta bios testers for DFI get a present too - upon a cold start, the board push 3V (!) Vcore to the poor CPU. As you can probably quess, the CPU die:
http://www.xtremesystems.org/forums/...ad.php?t=79638
http://www.xtremesystems.org/forums/...ad.php?t=83032
And possible there? http://www.xtremesystems.org/forums/...ad.php?t=94412

The sounds from the DFI boards continue to happening:
http://www.xtremesystems.org/forums/...ad.php?t=93030
...witch indicate still bad design - too much ripple, lact of filtering and stuff...

And the DFI RMA hell is quite interesting too:
http://www.xtremesystems.org/forums/...ad.php?t=91837

However - enought bashing. At least I know what to not buy anymore.


gonzo0815 - I think my Antec NEO has pretty neat both 5 and 3.3V lines:
http://silentpcreview.com/article177-page1.html
...but I try get hold of oscilloscope and check what the voltages looking, after all


CaysE - you did not breaking anything. After all, I got plenty users complaining about the very same thing, so the caps is dying in this mobo. Probably bad design - can't be, that all have crappy PSUs, right?
I got two China rev. A and one Taiwan, Rev. A. All have the mentioned cap soldered between the Vdimm mosfets. The exact revisions are a bit problem to obtain now for me, because I remove every (and I mean EVERY) stickers from things upon arrival and what is worser - WHAT LPT slot?

[RJARRRPCGP]

Re: DFI Lanparty B / Ultra Infinity (nF2)

But I didn't get any wording about the socket A DFI motherboards having circuit problems.

The problems were only reported with the DFI Lanparty NF4 motherboards, AFAIK.

[CaysE]

Re: DFI Lanparty B / Ultra Infinity (nF2)

I didn't think anyone used CPU shims anymore. Any problems after removing the LPT and serial ports?

[tiresias]

Re: DFI Lanparty B / Ultra Infinity (nF2)

And the unpaid public beta bios testers for DFI get a present too - upon a cold start, the board push 3V (!) Vcore to the poor CPU. As you can probably quess, the CPU die:

3.00 Vcore... that's even higher than what I'm running this stock-cooler-cooled (and regularly overheating) socket A sempron at!

I've never used DFI boards myself - since I've never really believed in overclocking any computer used as a "serious" machine/workstation, let alone server. In such applications I have used (and would in the future use) one of the Intel SE/Supermicro/Tyan boards which don't even "know" how to overclock.

For the (my kids', mostly!) gaming machines though, that's another matter - since in that case, raw speed is more important than stability! Perhaps I should suggest they consider a DFI for their next upgrade - although from what you tell me, perhaps it would be better if they just stick to those decent, and affordable, Asus/MSI boards!

....

PS. Besides, Diamond Flower International does sound suspicious, eh?

[Rainbow]

Re: DFI Lanparty B / Ultra Infinity (nF2)

DFI were never quality boards - and using colorful PCB and slots apparently did not help

[trodas]

Re: DFI Lanparty B / Ultra Infinity (nF2)

RJARRRPCGP - well, neither do I. However it looks like that the old nF2 boards in time kill themselves. All users around the globe can't do the same wrong with their DFI boards, however with the exception of one fella using preproduction board (probably higher quality components and as observed - a little bit different design there and there...), having todays almost disfunctional DFI board.
A coincidence?
And it is also a coincidence, that these board generated "stupid amount of noise", both acustic and electric in the voltages? (have you seen the Vcore voltage terrible spikes I posted and SoddenFX made in my thread about recaping the mobo?)


CaysE - wrong. It come with the SI-97A heatsink and I love and use it. Preventing the damage well. Possibly also improving the cooling, but that is bellow 1°C for sure And hey, I'm a old-skool guy, so...
And no, none problems after removing the serial ports and the LPT ports and recently I ALSO removed the sound ports - picture soon, if there is interest All you have to do is to make sure no spit of tin goes anywhere where it did not belong. Think about them just about shorted the wires that still did not go anywhere, so - why keep them?
(just please did not shortcut any live of 5V wire to ground when desoldering them - and that's all)


tiresias - well, Vcore 3.00V is killing Vcore for such CPUs. Stop thinking about how old-good Barton managed 2.2V and some WR are made with 2.41Vcore. The modern suxxkas did not take this anymore. My friend killed AMD 64 X2 3800+ by using stock Vcore (1.45) and 3.7Vdimm. The difference between the voltages break the CPU, killing it. So, you see, of just 2.25 voltage difference is killing zone now. Dual cores are like to break at poor 1.7xVcore.
Not good news, eh? At least most Opteron 146 manage with stock Vcore 3Ghz, so... My future CPU.

As for overclocking, well, I consider this a game. After all, things will run well at stock, if I fail, so - why not? Stability is just a problem of testing and backing down for stable use, that it is. Combination of various testing methods give me enought stress-testing so I could be confident that things won't fail on me, because of overclocking.
Of course, for serious usage is impossible to recommend such games, hehe

And I did not share your view of crashing possibility on kids gaming machine. The flustration in crash in game is great, and then damage to the machine might happen Anyway, not even for this purposes I would recommend DFI, since they are, well, just overclockers board. Overclockers who can fiddle with them, but they are nowhere near for real usage. Even the nF2 ones was long too beta-like, but it get improved after long time. Now the nF4 ones are not beta, they are alfa and still being alfa, no matter what they say. This is what Scotty write me about his DFI board and my suggestion to test my new OCZ rams in it:

I'm afraid if I put the OCZ memory in my machine that I wouldn'r be able to get it out Seriously, the DFI NF4 board can be a pain when it comes to optimizing new memory. Some memory works well in one set of slots, some in the other slots, some BIOS versions work better than others depending on the brand of memory, etc.. If I got started on the OCZ it would take me a week to get it dialed in, and I think you want it sooner than that =)

Don't take this words lightly.

Another fine example could be my case with nF2 DFI board (supposedly fixed and stable, yenah right?) today. What was the problem? Oh, well. The board handle exchanging CPU's extremly well. This is pretty good, since there is so many settings, so the board itself reset just the FSB, Vcore and multiplicator when new CPU is detected. Worked like a charm on all the normal CPU's. And now come my new AXP-M 2400+. Replacing oldie Barton 2500+, running 1.900Vcore and 2200Mhz. I inserted the new CPU and the board refused to post. It did not hit me at first, I get depressed that the nicely looking CPU is dead, so I put the old Barton back. Still no post. And then I remember. Oh yea, I had this issue before. In fact, DFI send me a new bios for free, to make my mobo work - that well the old bios ficked up the - well - bios itself. So, there comes the super-special DFI 2 steps bios cleaning rutine, and after hour (if you try fake it and cut the time down to 15min, it simple won't work, I tried and it only extended the wait-time to 1,5h...) I got old Barton running. Then I loaded the optimized defaults. Cleaned the bios, while inserting new CPU - and all worked out like a charm.
So, you see - DFI is undubtely the fastest boards for nF2, however they still have some quirks like two-stage cleaning bios. Cut of the power completely. 30min cleaning just by jumper, another 30min w/o battery and then first jumper back to original position and then put the battery back. Failure to do any of this in proper time and order result in not cleared bios and not posting mobo. Amazing, is not it?
And I would strongly advice agains the MSI (terrible crap they do) and Asus (slightly better that MSI, however my recent experience with their AMD64 mobo was rather displeasing...). I will go with the Sapphire RX480 "Grouper", and I did. I will report how it work later, when it come and I finish playing with nF2 platform and leave it to the folding, hehe So far it looks like a stable platform with very good speed and none of the quirks DFI mobos has plenty of them. Knock knock knock on my wood desk.
Oh, well, there is two things. The rams has to be close to each one for double channel run, so, even if you insert just two sticks, you have to insert them into bank 1 and 2, not 1 and 3 or 4. Worser cooling, but I have a solution... And the second thing? Ability to kill any cpu, that even remotely look like AMD 64 you insert into the socket by setting in bios 4.4Vdimm voltage. Even if you crank the CPU all the way up to 2V (may God bless your cpu, it it can make it...), you still set the voltage between core and ram controler on the same die difference to 2,4V and remember - difference 2.25V killed the AMD64 X2 3800+ instantly.


Rainbow - well, I disagree. You know, I'm NOT any big DFI fanboy or something. I gived up and refuse to comment anymore on the nF4 DFI boards as well. However I can't turn a blind eye on the fact, that DFI win all the WR, while the next-to-it Abit NF7 boards kept dying and dying with attempts to reach the DFI speeds. masterchorch killed 3 or 4 Abit NF7 trying to get stable 3.4Ghz, ultimately killing the best AXP-M cpu ever - not even SAE can do 3.4Ghz, but still - with Abit he did not scored ANY WR, except for the max CPU-Z screenshot: http://www.xtremesystems.org/forums/...ad.php?t=77316
So, the ability to reach - at least bench-stable - highest clocks is proof of the best design for nF2 ever.
Period.
No if's, but's or anything.
Yes, there are a hell lost of quirks and problems, however with just a little modding, they do miracles. Most of the problems can be fixed and the board design is good, except the noise. But after adding the filtering caps - noise problem is fixed.
So, you can't get anything better for nF2. Period.
For nF4 I can't recommend them, tough... So, about what DFI mobos are you talking about? Even DFI P4 mobos hold most WR's, so... :

This is not a WR, but just a small try with SuperPI on my DFI nF2:

[Per Hansson]

Re: DFI Lanparty B / Ultra Infinity (nF2)

trodas, on one hand I agree with you on the exsessive noise on the vcore and vdimm on the DFI NF4 boards, I've got an Expert and had to hotglue a coil to make it stop making noise...

But on the other hand; the DFI mobos would not be able to OC the DIMM and CPU so extremley high if the vcore and vdimm voltages where not cleaner than on your averge system... So I think in the end the layout must be better anyway... Though I don't have a scope to compare the three A64 boards I have here right now, that would be a very very interesting to do though... (at stock and overclocked settings)

[trodas]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Per Hansson - well, you got point, tough But maybe the worse things is not the excess spikes, but voltage fluctuation under load, and this is almost nonexistant on DFI. So, this attribute to the great overclocking abilities, while the spikes aren't affecting OC at all, BUT affecting the noise the coils generate and also killing the life of the caps?
Could that sound as good explaination?

Your noisy coils = another reason why I drop DFI in the AMD64 board choice.

BTW, untill I recap my another DFI LP B mobo (this time taiwan one), it has excessively noise coils, especially when I ping using extremly precise hrPING (try for yourself and report back, pls!) several address. I used:
ping 192.168.0.1. -t (my router/fw)
ping 192.168.100.1 -t (my cablemodem)
ping 82.114.192.1 -t (my ISP gateway)
ping seznam.cz -t (main czech portal)
...and I can't believe the noise I got! And now - quiet mobo! So, perhaps your caps are also on theirs way to silicon heaven?
(wait - does caps have place in silicon heaven, after all? aren't they tossed down to hell by all the other silicon components they burned off because of them? )

BTW, the taiwan mobo run with the mobile 2400+ at 200x13 now and stable and fold for me

And I moved the better 2600+ mobile to my main machine in preparation for WR attempt in 3DMark 01 and see for yourself:
http://valid.x86-secret.com/show_oc?id=84643
Now is not that sweet?

[Per Hansson]

Re: DFI Lanparty B / Ultra Infinity (nF2)

trodas; well, my DFI Expert is currently not alive; it's awaiting a waterpump for it's cooling system...

Right now I'm using the Epox 9NDA3+ I bought from tiresias on this board...
https://www.badcaps.net/forum/showthread.php?p=13374

[trowakage]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Does anyone have experience recapping the NF2 Infinity?

I have a feeling trodas you wrote most of the recapping info on this thread here:

http://www.pclincs.co.uk/Forums/show...5&pagenumber=2

I've been trying to put all the information from that thread together so I can recap my infinity with Panasonic FM's, but some of the numbers don't match up between the picture illustrating the cap placements and the digikey shopping list.

Could anyone help? (I tried emailing you trodas but got no reply)

[tazwegion]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Use the capacitor values already installed on your motherboard as your 'shopping guide' when looking for replacements, use the lowest ESR value you can

Additionally pay special attention to the polarity indicated on the PCB silk screening, as some use the shaded symbol to represent negative whilst others use it to indicate positive orientation

[Jor3llBR]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Bump.

I want to do the same on my DFI LP nF2 UltraB Taiwan, but the information is scattered or missing.

Can we have some links here for more details? Can someone re-host the pictures?

And also instead of doing all Trodas cap replacements, can I just replace the TEAPOS for Pana's FM (if the others are in good shape of course) and if so can I use Trodas diagram or they only work if doing them all?

Thanks!

[JaNoZ]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Hi guys i am new to this forum, i found it using google.

I got an NF2 Lanparty B made in taiwan with teapo's from a friend
I got it faulty, but i would like to try to revive it for my use.
My intention was to have it for spare parts (flash chip ) for my other (taiwan) NF2 LP B mobo that i have.

I hotflashed the bios in my other board coz it got corrupted by the board failure and cause the first reason not to boot 4 leds burning, and now 1 out of 10 times it succeeds in posting after a few resets and so on.

I measured the mosfets for shorts.... none.

Now i got to this page, and you guys are telling me maybe that recapping that sh!tty teapo's and kzg's would give me more luck to save the board??

can someone help me out here, because i would never figure the caps would be the problem / you cant check the caps in any way ? other than with an oscilloscope right?

[gonzo0815]

Re: DFI Lanparty B / Ultra Infinity (nF2)

Are you familiar with the whole cmos and other quirks this board is famous for?
If not i would suggest to make sure it isn`t one of this issues.

You can check the caps with an ESR meter, but if you are not into electronics i think it would just be cheaper to try a capacitor swap.

First swap those Teapos and may be the mentioned bad KZG`s (not all only those which are indicated by trodas).

[gustie]

Re: DFI Lanparty B / Ultra Infinity (nF2)

HELLO NEWBIE ALERT.

My first post here...

Im another owner of a DFI LP NF2 Ultra B motherboard

i wasnt aware there was taiwan or china differences... is it as easy as looking for the country of origin on the motherboard? or do i have to look around at parts to ID this?

My system sometimes halts whilst playing music and will just continue to repeat the last noise in a loop... mouse and keys become useless and the reset button is the only way to stop this.


I was told it could be the capacitors on the DFI forum and pointed here...

I cant check the capacitors until i get home... but from what i read its best to replace ALL of them rather than one or two...

i think one had a dink in the side tho..

WHAT is the best capacitors to put on this board? i dont have the money to just upgrade the whole lot as its AGP,socket A etc... IDE blah blah and i tend to only use it for music and music production. so it fits its purpose (when it works properly!)


I have the bios to run through also as i used to have Hellfire REV3 Bios (which was very stable) but i was advised another bios would be better... which maybe could be my problem otherwise... who knows

i think i'll have to flash to a DFI bios and try that too... or go back to the hellfire...


i wouldnt mind a crack at soldering my own caps on! (could be a bad move lol) so what is the best type of iron to buy and what is the best flux or what ever you need to use also??

any help will be greatly appreciated

CHEERS