AN35N Ultra Vdd, Vcore, & Vdimm Solved

[Hell-Fire]

As title says...got Vdd and Vdimm worked out juse fine.

There is an overall lack of material on the net concerning mods for this board, but after receiving a request to mod one.....off I went.

There was very little info on the net, but I did bump into a thread or 2 where people said they performed the Vdimm mod and highest value possible was 2.9v. I got well past 3.2v without issue, so not sure what problems they were having with that mod.

I am working on the Vcore mod......but not sure the board actually needs it. It overvolts from what I can tell, and the bios allows upwards of 2v. That is not xtreme mind you, but the actual Vcore is closer to 2.2v I am speculating.

Regardless, I will be setting to work on it in the morning. It appears that several chips are tied together on this one, but I think I have an idea what needs to be done.

The RT9602 has pins Phase1 and Phase2 tied directly to Vcore.....as is Vsen of the RT9241. The circuit diagram shows pins #13 and #8 (Phase1 and Phase2) of the RT9602 forming one circuit, with pin #14 (Vsen) of RT9241 joining the same circuit.

A connectivity test indeed shows that the above pins are connected, and they are also connected to Vcc of the 2 GD75232 ICs as well.

Bare in mind that RT9241 also has a FB pin, which is the traditional mod location, but I am thinking that Vsen is the target pin for now.

Any ideas are obviously welcome.

[Hell-Fire]

Ok, some odd results.

Modding Vsen on the RT9241B Vreg had some weird action to it.

Turning the VR might yield a voltage increase, but then the next 1/4 turn would yield a decrease. The voltage has steadily increased, but it still bounces up and down as the VR is adjusted. It may be 1.795, then lowering the resistance more puts the voltage at 1.738v, then lowering again pushes it back to 1.79v...??

That is with the VR on pin #14 (Vsen) and GND. Maybe the VR should either be between Vsen and FB on the RT9241 Vreg...or maybe even between the the RT9241 and RT9602 since according to data sheets they both are connected to Vcore.

Any input?

[Hell-Fire]

**UPDATE**

Okey Dokey.

Got it all worked out it seems.

I was right to begin with. Modding Pin #14 (Vsen) on the RT9421B IC was the correct move. In my first attempt I was using a very small gauge wire and a 100k VR. I found that although the Vcore was all over the place, with the VR set to 20.22K ohms, the Vcore was +0.2v higher than the bios setting. That would yield a solid 2.2v when set to the bios max value of 2v.

I then started thinking about the wire I used and the 100K VR. So I decided to take another crack at it with a 50K VR set to 25K, and a thicker gauge wire. Worked like a charm and now the Vcore can be increased or decreased and is rock solid without flux.

I am tired, so I will update this thread with pics and the proper VRs to use later this week.

If anyone has this board and want to try out the mods, here are the VRs needed and the Vregs to modify:

Vdimm ---> LM431 (near dimms) Pin #8 to GND with 1K VR.
Vdd ---> RT9203 (between PCI and AGP) Pin #8 to GND with 100K VR.
Vcore ---> RT9241B (beside Fan #3 header) Pin #14 to GND with 50K VR.

[STEvil]

Good to see someone finally figured it out

[Hell-Fire]

Originally posted by STEvil
Good to see someone finally figured it out

Whats odd to me is that the only live pins on the LM431 are pins #1 and #8. All other pins are either tied to GND, or simply out of the circuit loop.

[celemine1Gig]

Well, THANK YOU VERY MUCH Hell-Fire!!! This Thread is exactly what I needed ATM. And respect that you figured out the VCore mod.
I'll try the mods as soon as possible.

BTW, did you also do the thermal diode mod on you board. That's what I'm planning to do soon.

Greetz

Ingmar

[Hell-Fire]

Originally posted by celemine1Gig
Well, THANK YOU VERY MUCH Hell-Fire!!! This Thread is exactly what I needed ATM. And respect that you figured out the VCore mod.
I'll try the mods as soon as possible.

BTW, did you also do the thermal diode mod on you board. That's what I'm planning to do soon.

Greetz

Ingmar

Nah, I didnt do the diode action as this isnt my board.

I did the mods on this motherboard and also mods to a 9800Pro for the owner.

I was planning on doing some MemTest86+ testing on it before giving it back to the owner, but my floppy drive is D-E-D dead. I did get boots up to 245fsb, but not sure if thats stable or not as I couldnt do testing.

The fact that it booted at 245 was nice if you ask me...if you think about the fact that this isnt considered a "performance" board.

The owner is running cascade and planning on a A64 setup soon....now thats what I call an upgrade....moving from this setup to A64.

[Hell-Fire]

**EDIT**

This is not my pic. Someone informed us that I had the wrong read point for Vdd. Here is an updated pic showing the correct read point.


Otay, here is a pic of the board after mods:

[STEvil]

bump

[STEvil]

update

Did the vdimm mod, I still max out at 2.9v with 3.3 raill @ 3.3.

With the 3.3 rail @ 3.7 I get 3.1v which is good for 232mhz 2.5-2-2-11 on my BH-6's

The board seems to be responding better to adjusting CPU mhz also with the addition of the vdimm mod...?

EDIT

VBT/VTT tracks, also.

[celemine1Gig]

Originally posted by STEvil
update

Did the vdimm mod, I still max out at 2.9v with 3.3 raill @ 3.3.

With the 3.3 rail @ 3.7 I get 3.1v which is good for 232mhz 2.5-2-2-11 on my BH-6's

The board seems to be responding better to adjusting CPU mhz also with the addition of the vdimm mod...?

EDIT

VBT/VTT tracks, also.

Hmm, that's quite interesting STEvil! I could adjust the Vdimm to a max of 3,05V while having my 3.3V rail @ ~3,4V (didn't go higher yet, without a major mod to the PSU).
And it's good news that VTT is tracking VDimm.
BUT, I bought a DFI Lanparty Nforce2 Ultra B anyway. So, we'll se, how the Shuttle, which is nearly one third of the price of the Lanparty, competes with it.

[Iggy]

I did the Vdimm and VDD mods the other day and I'm still trying to get the VDD to work. As soon as I connect pin 8 on RT9203 to any ground the 100k VR shows 0.00 ohms when it's actually set to about 30k.

[CaTalyst.X]

You cant check resistance on the VR once it is soldered in circuit bro.......If you set it at 30K before you solder it in, it will stay at 50K, the heat wont change the resistance or anything. BTW HF i am slowly starting to see the benefits of hot glue

-CaT

[STEvil]

My Infinity made it to 250 @ 3.45v 2-2-2-11 100% stable.

I think I need to do the VDD mod and lap the NB a bunch before i'm going to get 240 or maybe more..

I'm going to go and solder the +3.3 direct to the backside of the vdimm mosfet now to bypass the diode as the diode is in a location too tight for my soldering iron to get.. darn 325w iron

This board isnt half bad once you do some work to it

EDIT

I meant to post this a couple hours back actually.. just got side-tracked watching Aliens and stuff

Anyways, on a related note.. I cant get the solder to stick to the mosfet or the diode at all.. wtf?

[STEvil]

Ok, got the solder to stick (to a different location) so vdimm is effectively driven by the +3.3 rail now instead of only +3.0v

Attaching the +3.3 directly to the source feed of the mosfet allows me to go to 3.3v on the memory without having to crank up my +3.3v line past 3.55, but even turning it up to 3.8v didnt get me past 3.3v to mem. It also hasnt raised my max FSB, I cant run CAS2 with all the extra voltage, and setting vdimm to 2.7 (3.25 real) causes errors in memtest..

Time to lap the NB and do the vdd+vcore mods I guess

[celemine1Gig]

BTW, I wrote a little pictured guide, which will be online soon (site is not available for the public yet, but will be soon). Thanks to Hell-Fire and AA666 who figured everything out!

[STEvil]

Vcore goes to 2.2+ with EASE

Too bad its such a pain to solder to the mosfets... you gotta get the suckers cooking before the solder will stick


These things could have been nF7 killers.. maybe still could be with some bios tweaks?

[celemine1Gig]

Well, if there would be mod bioses like for the NF7 with CPC off and all the other fancy options, the shuttles really could've been killer boards. But ATM, they aren't.

[STEvil]

Pffft, dont need no stinking mod bios!!

Just need a FAST bios!

[celemine1Gig]

I finally got me a DFI Lanparty Ultra B. So I can now try to make my Shuttle run even faster, while still having a blazin fast system up and running on the DFI. Perhaps I'll even sell my Shuttle.

[STEvil]

Did you ever play with the C1 disconnect option in the bios? I havent touched it yet.. dunno what the heck its for

[Iggy]

You cant check resistance on the VR once it is soldered in circuit bro.......If you set it at 30K before you solder it in, it will stay at 50K, the heat wont change the resistance or anything. BTW HF i am slowly starting to see the benefits of hot glue

EDIT: but wait, why can I measure the resistance on my 1k which is controlling the Vdimm?

Thanks bud-I'm a vmod and electrical n00b, and I've still got alot to learn. I'll see if I can fit this on K6-2 hsf on the nb today (I have to clear the 9500's 2" tall ramsinks) and maybe finish the lap job on the nb. So far I've been able to boot at up to 230 with the VDD at stock and the Vdimm at 2.95 on my new Geil PC3200 "Value", which will do a 250 fsb, but I'll see how far I can really push this thing today. As always a modded bios would be a great help.

Too bad its such a pain to solder to the mosfets... you gotta get the suckers cooking before the solder will stick

No kidding-I finally had to switch to a larger tip on the Weller TC201 to get anything to stick to the Vdimm measurement point.

[STEvil]

I tightened my tip on.. it was kinda loose. Its working betting now

If you remove/bypass that diode you can get more vdimm.. extra .1 or .2v anyways.

[bh2k]

every little bit helps

[STEvil]

Oh the lucrid smell of burning silicone...

Did the VDD mod, but apparently using a 100k VR is very unsafe.

Went something like this:

*install mod*
wow I am the soldering god! (ego+2)
*put dmm in measure thingy, press power*
ooh, its working!! (ego+2)
*turn turn turn turn turn.............................................. .*
hhmm.. getting kinda low on t.. WTF 0.08v??
*shut down and turn the other direction a few turns*
ok, here we go again..
*power on*
good, still works! (ego +4)
*turn VERRRRY slowlSNAP!!!*
eep. (adrenaline rush)
*insert burned silicone smell here*
sheeeet (panic sets in)
*SHUT DOWN!!!!!*
omgomgomgomgomg
*turn the other way and power on again*
ok, vdd voltage looks right... but no monitor signal!!!
*numerous reboots/cmos resets/etc later...*
NOOOOOOOOooooooooo

Ego -eleventybillion.



Well, at least it will make for good pics and I can test for max vcore since it still turns on at least.


Conclusion: Using 100k VR is bad (??????????).


Crap... no system at all now.