Epox 9NDA3+ with exsessive coil noise

[markiemrboo]

Re: Epox 9NDA3+ with exsessive coil noise

Hi Markie,

We certainly believe you!

It just suprises me somewhat to see that - since I had, on the 2nd February 2005, shortly after buying a Rev1 9NDA, contacted the UK "representative" at epox.org (now dead, just a redirect in its place, it seems), and he had "assured me" (in salesman's terms, now I realise) that the only difference between Rev1 and 2 was a different CPU Voltage regulator chip.

lol

I guess the different regulator chip (AFAIK that is actually the only difference really) needed a slightly different layout to the old one. It is quite different around that area compared to the rev 1 boards, but I think that's about as far as the differences go

PS. All Sanyos it seems too - is that your own recap or was Epox having a good hair day?

Haven't touched this board matey. They're all stock caps

PPS. They even gave you bigger, chunkier, FETs! That's just not fair, is it?

Haha, maybe the regulator isn't the only difference then

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

the only difference between Rev1 and 2 was a different CPU Voltage regulator chip

Thanks for that info!

Now I know why the latest BIOS does not allow me to reduce my VCORE, only increase it...

BIOS v09/15/2005 changelog:
Added: Support F75121 and Atxp9 negative Vcore voltage adjust

Because my 9NDA3+ uses a Intersil ISL6559CB voltage controller, instead of the required Fintek F75121 voltage controller...

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

markiemrboo; So, can you describe the mod a bit more? What does it change, i.e. where is that resistor located in the circuit in relation to the F75121 voltage controller?

I've done a few voltmod articles in the past so I might be able to figure it out :P

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Just installed an Opteron 146, just for testing...

Idle: 1.447v
Load: 1.473v

Delta: 0.026v

Quite a difference from the ClawHammer... The coils are also about 90% more quiet now... I can still hear them but it's just a very very faint noise now...

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Overclocked the Opteron to 2.65Ghz, full load voltage is now 1,482v, so a 0,035v delta now... Quite interesting...

Can't get the Opteron to do any more either, I have not increased the voltage to reach 2.65Ghz but if I try 2.7 I get stuck on 1C during post... Does not matter if I change the voltages I have control over or anything...

I was previously running this CPU in a Asus A8N-SLI Deluxe, got the same OC result with it but blamed it on the mobo then, guess I'll have to rethink now :P

And I was hoping for 3Ghz on air with this CPU

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

Often a CPU will seem to have a hard "limit" as far as overclocking is concerned, above which large voltage increases are required for very little performance gain. Not coincidentally, this limit is often a little (but not a huge amount) above the highest speed that particular CPU core is being 'binned' at.

Still, a 1MB L2 K8 at 2.65GHz is pretty much as good as it gets as far as stable single-threaded performance goes. I'm glad to hear the singing-coil mystery has been, at least in part, solved!

How's the 'thermal-adhesive attached passive cooler' (doesn't that sound grand) on the southbridge holding up? I noticed it never really got particularly warm - in any case, it's probably performing better than the original, whiny-but-weak little fan that Epox originally taped on.

[trodas]

Re: Epox 9NDA3+ with exsessive coil noise

tiresias -

What is actually, electronically speaking, happening to cause the VRM's output voltage to increase with greater current draw?

The PWM stage is compensating for the more current the load and it does overcompensate it a little for the sake of stability. It is quite normal and common. Mine difer by 0.019V, witch is acceptable, IMHO.
The only one way to improve this voltage changes is to increase the switching frequency of the mosfets, so they more accurately and precisely control the voltage and if the overcompensating is still way high, then extremly slightly decease the value of the autosensing resistor that reporting back the Vcore voltage to the regulator chip...


Per Hansson - bad luck, mate. You need better stepping. Most Opterons 146 can do 3GHz w/o breaking a sweat or requiring more Vcore. Mine 148 does 3GHz easily on bad aircool on testing bench, so - just stepping - that's all.
Increasing Vcore on AMD64 CPU's did not seems to having the great effect like it used to has on the oldie AXP ones

Anyway, as for he coils - mine are very quiet now, hard to heard them sometimes and I have fanless (except one very quiet fan in PSU) machine, so... I think that when you exchange all your caps to good ones, up their capacity a little and use hi-quality ones (Rubycon MCZ, Nichicon HZ, Samxon GA - in bad case even the Panasonic FM ones do the job) AND add bridging 100nF caps on the bottom of each big one - you get rid like I'm of most of the noise Works for me, will work for ya.


japlytic -

Usually, a little bit of high-temperature silicone (non-aceatic so it does not attract ants) on the coils and board silences the 'fizzing' sound.

Well, been there, mate, no help. Only it increased unstability when O/C, thanks to the good thermal transfer of the silicone - it does heat-up the mobo PCB a lot and then... Go figure what mosfets like even more heat, hehe
So, quieting coil with any glues, silicones and such - that is the WORSEST THING anyone can do. Believe me, never ever ever recommend this to anyone. It not only not work, it make your machine potentialy unstable and it looks drop-dead-ughly
See there:
http://ax2.old-cans.com/s.php?p=wc&id=118&c=8&d=1&v=v2
http://ax2.old-cans.com/s.php?p=wc&id=122&c=8&d=1&v=v2
Ughly enought for ya?
And remember - this stuff is impossible to remove! I tried, but no way...

So, by experience - I strongly advise NEVER try quiet coild with silicone or other glues...!

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

tiresias; the chipset is running about 40°C at 265Mhz HTT, very nice...

trodas; hot glue does help on most coils, like on my DFI Expert and my 3com switch, there it totally killed the noise... Only a small dab was required, think I posted a pic of it here somewhere...

You might be right about that adding resistors under the bulkcaps helps, since the frequency is probably much higher than what the caps are able to filter anyway... But just adding something blindly, that is not something I want to be the first one to try...

Regarding the CPU and stepping; yes I know, fortunantley it's not my CPU but a friend of mine... He was hoping for 3ghz with it... The ClawHammer CPU was his, they both hit exactly the same frequency, quite ironic no? muahahahahaha, fucker

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

So, quieting coil with any glues, silicones and such - that is the WORSEST THING anyone can do. Believe me, never ever ever recommend this to anyone. It not only not work, it make your machine potentialy unstable and it looks drop-dead-ughly

lol! That sounds serious!

I think all japlytic was suggesting was to add a little hot glue or silicone somewhere on the coil to reduce vibration - a little like PSU manufacturers like to do all over the inside of their units.

[trodas]

Re: Epox 9NDA3+ with exsessive coil noise

Per Hansson - hmmm, dubt it helps, post a pic, pls.
And nope, not resistors, but small caps. This filter out the hi-frequency spikes/noise they aren't able to filter at all
And you can easily add/remove them, so... BTW, I would start with the there green (probably any ugly manufacturer) caps before the Vcore ones

He can still hit 3GHz, but with phasecooling only, hehe


tiresias - Im dead serious. Silicone won't help, the noise won't stop and since the noise is product of either failing caps or bad design, it won't make things better even if it temporary fix the problem, as the caps don't get better and the board design won't improve itself, so...
And please provide pic of what "inside PSU" glue stuff you meant. I was opening my new Enermax Liberty 620W ATX 2.2 PSU and there is no stuff like that in coils and there is inside surprising amount of toroid colis, after all

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

trodas; uhh, is I said the "ugle three green" caps are Sanyo WG caps, one of the best caps there is, and the other caps I put on are Rubycon MBZ, well enough for this board IMNSHO

And I wont bother to post a pic, I know it works, that is enough for me, and if it where dangerous I don't think you would find it everywhere in powersupplies and TV sets etc etc etc

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Per Hansson - hmmm, dubt it helps, post a pic, pls.
And nope, not resistors, but small caps. This filter out the hi-frequency spikes/noise they aren't able to filter at all
And you can easily add/remove them, so... BTW, I would start with the there green (probably any ugly manufacturer) caps before the Vcore ones

He can still hit 3GHz, but with phasecooling only, hehe


tiresias - Im dead serious. Silicone won't help, the noise won't stop and since the noise is product of either failing caps or bad design, it won't make things better even if it temporary fix the problem, as the caps don't get better and the board design won't improve itself, so...
And please provide pic of what "inside PSU" glue stuff you meant. I was opening my new Enermax Liberty 620W ATX 2.2 PSU and there is no stuff like that in coils and there is inside surprising amount of toroid colis, after all

Okay todas I was bored so here is your requested reply

His very funny Opteron now all off a sudden accepts 2.8ghz at 1.4v, before it did not even post at 2.7ghz 1.55v, go figure, burn in that did it probably...

So maybe it will do 3ghz in a few weeks ;-)

The silicon does help, it does remove the effect of the torroid core moving in the coil thus creating coilnoise, except on my very strange 9NDA3+, and coil noise is not something that happen with only bad capacitors, it just happens when there is ripple current and a small space for the torroid core to be able to move in phase with the ripple current of the circuit it is on...

Picture of a Nexus NX-3000 PSU with lotsa glue in it... Also a very dead PSU ;-)

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

Picture of a Nexus NX-3000 PSU with lotsa glue in it... Also a very dead PSU ;-)

Looks identical to the very cheap Fortron ATX300-GT's I have here, even down to the FSP logo on the U1 chip!

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Looks identical to the very cheap Fortron ATX300-GT's I have here, even down to the FSP logo on the U1 chip!

Yes I know, I found it in a review at x-bit labs too...

Though I think Nexus put their own fan logic circuit on it because it was very very silent, top rated at Silent PC Review, I really liked that PSU...

EDIT; nahh, even the fan logic circuit has the "F" logo ;-)

trodas: is this the type of caps you recommend using to filter the vcore, in parallell with the bulkcaps that are already there? Should I put them on both the 12v and vcore (1.5v) side or only one of them?

http://www.elfa.se/elfa-bin/setpage....htm?_65_752_29

[tiresias]

Re: Epox 9NDA3+ with exsessive coil noise

Though I think Nexus put their own fan logic circuit on it because it was very very silent, top rated at Silent PC Review, I really liked that PSU...

I'm always a little suspicious of those 'silent' PSUs... there's only a number of ways to make an air-cooled device 'noise-free', and not many of them are likely to increase a unit's 'MTBF'!

Edit: I did just notice a small difference that the standard, came-free-in-the-box Fortrons here don't have... yours has the red self-destruct button.

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Edit: I did just notice a small difference that the standard, came-free-in-the-box Fortrons here don't have... yours has the red self-destruct button.

Ahh, very good very good, always nice to have that self destruct button when the RIAA comes to visit

On another note I tried to use the thermal control PCB on it's own but it did not work, the fan ran at one speed only

The thermistor was not on the fancontrollers PCB, but on the main PCB in the PSU, so I moved it to the fancontrollers PCB (the cable from the fancontroller to the main PCB had one unused solder hole next to it so I thought it would work...)

[linuxguru]

Re: Epox 9NDA3+ with exsessive coil noise

That Nexus/Fortron is worth fixing - apart from the glue and the Fuhjyyus, it's well made. If you can get suitable 10 mm caps, and the controller IC is not dead, give it a shot and recap it.

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

linuxguru: Follow the link I posted, that PSU is dead alright, switching transistors, transformers and allot of components I can not identify because they are simply burned up

Picture of a Nexus NX-3000 PSU with lotsa glue in it... Also a very dead PSU ;-)

[Per Hansson]

Re: Epox 9NDA3+ with exsessive coil noise

Damn, it just struck me; why the Nexus NX3000 failed... It is because the voltage controller for the fan failed!!!

If you read the thread, (linked above) you can see me saying that the primary side heatsink was over 100°C hot, well, without proper cooling and a high load that is not exactly strange... So the reason it exploded was because the primary side switching transistors got so hot so they made the solder from the collector pin reflow and drip off, not the fact that it was a bad solder joint as I first throught!

So, in hindsight it is very nice to realize that there even is not an RPM sensor cable coming out of the PSU... Leasson learnt; buy a PSU with RPM sensor and hook it up to your mobo and tell it to emegency poweroff if it drops below a certain RPM!!! (Make new post in PSU forum and make that a sticky perhaps?)

[Tom41]

Re: Epox 9NDA3+ with exsessive coil noise

Slightly off-topic here, but right in the middle of that PSU pic adjacent to the cyan cap, there's a component I can't identify. On one dead PSU I saw, that component had elongated to the length of a finger!
Unfortunately, this was before I discovered badcaps.net and the cap problem,
otherwise I would have investigated and seen if any caps had vented. Maybe the component got so hot that it melted and dripped down (the PSU hangs upside down in most ATX cases). As it is, I just threw the PSU away without taking any closer look at it. Can anyone identify this component and what it's for?

Also, on that pic I can't see anything visibly wrong. There are no bulged/vented caps, nothing burnt - what can you see that I can't?