[Sammelthread] AMD K7 - Sockel A (462)

[Jone]

Wo wir gerade dabei sind:
Sind kaputte DFI eigentlich so schwer zu bekommen wie funktionierende? Also ich meine nicht für Reparaturversuche, sondern wirklich nur fürs Regal zum angucken...

Da die meisten defekten als "hat vorm Ausbau tadellos funktioniert" verkauft werden, auf jeden fall

 

[Tzk]

Ich meinte wirklich "defekt - gibt kein Bild". Also ein Board was auf keinen Fall mehr funktioniert Ich würde mir sowas tatsächlich an die Wand hängen oder in die Vitrine packen.

 

[Sparksnl]

I think I forgot to tell you guys I still have this one as well:



Sadly it is defective. NB contacts came loose... So it needs to be reballed. I just don’t know anyone who can do this.

Edit: now I look at it... It had FS caps, oops.That alone could cause issues. That needed to be FR.

 

[Tzk]

I'd upgrade straight to solidcaps and leave those FR alone

 

[Sparksnl]

Yes, I don’t even buy “normals” anymore.

 

[Hampti]

Richtig. Die KZG sind hier nur ein Teil der Seuche. Je nach Variation der Bestückung befindet sich ein 470er Teapo SC auf dem Board. Der muss, sofern vorhanden in jedem Fall runter und die 13 – 14 OST RLG/RLX an Chipsatz und RAM sollten eigentlich auch weichen, wenn man lange Freude daran haben will, aber das wird dann ein Thema für den Bastelthread @Hampti, falls du mehr Infos brauchst.

Danke für den Hint. Ja ich weiß da sehen einige nicht mehr gut aus. Bin aber nicht sicher ob ich mir das noch antun werde. Mal schauen. In den nächsten Monaten jedenfalls keine Motivation/Zeit dazu.

 

[maxpower1984]

Danke für den Hint. Ja ich weiß da sehen einige nicht mehr gut aus. Bin aber nicht sicher ob ich mir das noch antun werde. Mal schauen. In den nächsten Monaten jedenfalls keine Motivation/Zeit dazu.

Dahu gibt es immer nette Leute hier, die das erledigen können.

 

[Tzk]

Das war jetzt aber ein Wink mit dem Zaunpfahl

 

[WMDK]

That needed to be FR.

Even if that was the case and we'd see the board equipped with FR caps instead of FS, it wouldn't make any difference at all since all FR caps in 10mm diameter don't come with an ESR value equivalent to KZGs, HMs, MBZs and some boards tend to be very picky when it comes to ESR deviations. Of course you'll never know until you try but from my experience, on high-end boards, in maybe 7 or 8 out of 10 cases, Panasonic FR are simply not enough to replace those old Ultra Low ESR caps.

Considering the fact that we have a DFI board in front of us which usually only is a paraphrase for "I was born picky!" , I wouldn't take my chances with FRs or even FS to begin with One also has to keep in mind that on first sight, it doesn't seen to make a big difference if an ESR17 cap is used instead of an ESR12 cap but just do the maths and you'll realize that this seemingly small difference translates into an almost 50% increase of equivalent series resistance. An increase like that can easily cause a delicate circuit design to fail or flutter in stability while in operation.

Usually those types of caps are there for a reason and no sane manufacturer would have considered taking the more expensive approach if it wasn't required to do the job and rather take the cost savings instead. On some boards we even see UCC KZJ, Rubycon MCZ, Nichicon HZ and the likes which were even more expensive and more extreme when it comes to ESR and ripple specifications. So if I was you, I'd definitely swap those FS with a nice set of 2700µF/4V PLGs if you manage to find someone who can reball the northbridge and no other damage has been inflicted to that specific board.

 

[Sparksnl]

This was one of the first boards I (partially) recapped after I stopped soldering for at least 20 years. I didn’t look at low ESR back then. After recapping my Epox board I noticed it was not working anymore, so then I learned...
Since this board has 2 issues I never looked at it again and just put it in storage.
FR series are working on most boards in my experience, but these days I use solids mainly. I buy them with low ESR in mind.

edit: is 4v enough?

 

[WMDK]

edit: is 4v enough?

Yes. Those only see CPU vCore, so 4V is no problem.

 

[Sparksnl]

Ok, I never really measured it’s voltage on the caps. But now I think of it, it seems logical.
The thing is that I seem to be gifted with soldering. The electronics part is further away in my memory...

Edit: I ordered these last time:

16v 2200uF
6.3v 2200uF

Both for use near cpu. The lower ones when no space is available.

 

[WMDK]

Ok, I never really measured it’s voltage on the caps. But now I think of it, it seems logical.

There were some boards that came with both types of caps, depending on what they had on the shelf basically and you would see substitutions like 860µF/2.5V or 1200µF/4V instead of 3300µF/6,3V in the same spot with no other changes to the board than that. Thing is, even if they didn't need the voltage rating for that particular situation, 6.3V was the lowest rating available with standard caps.

Of course you should never differ from the original rating if you're not absolutely sure that it's a viable option so don't get me wrong here. Always stick to the original specs or even exceed them but never go below those specs if you're not 100% sure it can be done without side effects – be it short-term or long-term. In the other thread I explained how some decisions were made back then by the manufacturer on what cap to consider. For instance if you needed ultra low ESR and high ripple, there was no other cap available than the large capacity ones of the likes of 3300µF and so on even though they might not have needed the capacity but it was the only way to get the other specs right, capacity then was a byproduct really – at least in those amounts.

 

[Sparksnl]

Another recap for the NFII Ultra A did not help (code 0).

But DID I just fix my "defective" Lanparty B. I tried running it at 100MHz which did not help. I put it to normal mode again and tried running it, still nothing. I decided not to use on/off but RESET and this was key.
After resetting it, it ran again. I was actually filming it because I could post the error code. I thought: Hey... not the same code but normal boot.
So now I am happy with 2 working Lanparty B motherboards! This latest board I bought for €65 and the other for €85, both in 2020.
Both boards were bought as defective!

😁 😁 😁 😁 😁 😁 😁 😁

 

[The Sandman]

Congrats. Now check if it can do 260+

 

[Sparksnl]

Congrats. Now check if it can do 260+

I will do that soon.
My first board seems to be able, I already know that. I was just careful not to run it that much. But now I have 2...

 

[Tzk]

If you kill it, i'm interested in buying it afterwards...

 

[The Sandman]

The max 1.9V NB Voltage should be safe. Just make sure you have a fan blowing on the vrms. If you got really good BH-5 even 260-MHz should be possible 32M at 3.3Volts

 

[Sparksnl]

If I kill it I keep it for parts. 😁

I actually have good BH5, I have several that do 260-280 on NF4.

This is the movie where to my surprise it started working. Note the EE/EF code where it starts.

 

[Tzk]

I thought about buying that exact debug card for 7€ on amazon... Does it work good aka is it worth the buy?

Also, what about the Vdimm rework? I remember @The Sandman mentioning it a few times... Afaik there is some capacitor or resistor added, which helps keeping the Vdimm VRM alive.

 

[The Sandman]

FEE. The german word for Fairy 🧚‍♂️ . Mine only beeps or plays a weird melody if it fails. Or simply only the 4 red LEDS of Death when the bios is corrupted.

 

[Sparksnl]

Yes it is worth it, I use it for a few years now.mI can’t do without anymore. The nice thing is that it is.

This one kept on beeping even with a new bios chip...
Yes the capacitor is for more stability. He explained it some time ago in this thread to me.
But I probably will remove some parts there to be able to run above 3.3v...

 

[The Sandman]

DFI LanParty Ultra B (462) SMD's burnt - fixable?

Hey Guys, unfortunately my DFI died some days ago. I didn't realize (had to look at some pics I took a few days back) but some SMD components underneath the socket lever where already fried before the board ultimately refused to reboot from a warmstart. Any chance fixing it? The Board still fires...

community.hwbot.org

"It seems to be the same as C7G1 capacitor on the left. Based on the size it should be 0805"

the rework can only be found on the latest versions. It was soldered by hand. I think with that modification the 3.3V are safe to use. Infra had 3.7V on it and it didn't blow up.

 

[Tzk]

Well... 2x512 BH-5 and 3.6V killed my DFI Infinity. Even with heatsink on the tiny VRM chip and two 120mm fans blowing on it So you probably want to play it safe.

 

[The Sandman]

2x 512MB (double sided). are twice as power hungry as 2x 256MB. That could explain it. I wouldn't vmod mine. Going with TCCD. I'm getting maxpowers OCZ. Maybe I can run 270-MHz 32M with em or at least finish a 266-MHz run.

 

[Tzk]

Exactly. That's why i also only ran TCCDs on my 2nd Infinity which i sold a few years ago. I'd love to see more than 265Mhz 32M on any NF2 board

 

[bschicht86]

Die Debug-Karte - Gibts einen Link dazu?

 

[The Sandman]

Vielleicht bricht Sparks meinen Rekord oder fährt die 266-MHz noch vor mir. Material hat er ja gerade genug

 

[Sparksnl]

I will just remove the vdimm and use FireRAM. No worry's then.
edit: I bought that debug card on Aliexpress.

 

[The Sandman]

should be this one

Page Not Found - Aliexpress.com