max3 Vdimm stabalisation mod.

[Tony]

As ever mods void your warranty

[Charles Wirth]

I will give it a try and give feedback

Im also doing a cap mod on the 875NEO now that I got a good bios going.

God forbid I get stable voltages at 2+ volts, hehe hehehe hehe bwahahahahahaha

[Tony]

What you need is a HUGE PSU....Or build yourself a battery interface and run it off a huge car battery.

[PiLsY]

You after cpu or vdimm stabilisation m8?

[Charles Wirth]

both

I have a huge PSU

bad boy I picked up has 100A on the 12v line.

is that big enough?

[Poki]

Originally posted by FUGGER
both

I have a huge PSU

bad boy I picked up has 100A on the 12v line.

is that big enough?

BOAT ANCHOR
I had a 60amp for a while to drive a linear amp on a HAM rig. That thing was a pig, i can only imagine a 100amp

[PiLsY]

Fugger - regarding stabilisation of voltages.

Couple of things - adding caps and boosting coils will only get you so far. The real trick would be to replace the mosfets.

Secondly - a mosfet is a vreg. You can get adjustable ones. Depending on how much ampage you need there are various models available. If you can find the code off one of the power mosfets and look up the spec sheets then you can get an ampage rating. Find another mosfet with the same characteristics but higher ampage rating and adjustable. Dial in a few more volts. Its a very simple circuit to provide adjustment on a simple linear vreg. They have 3 pins for the most part - vIN, vOUT (also the back is vout, hence the cut off pin on the centre of the mosfets) and vADJ. R1 goes from vOUT to vADJ. R2 is from vADJ to ground.


Another method for adding some more regulation is to simply cut the mosfet legs and wire in an add in circuit with more voltage filtering. Find the vINP for every mosfet. De-solder the pin from the boarde and bend it upwards. Solder a wire from the board to a small piece of stripboard on. Solder the positive side of a 1000uF 16v cap to this wire. From the same point on the cap your wire is attached solder another wire going to the mosfet. On the negative side of the cap add a wire to ground. You can also experiment with adding simple voltage booster circuits. Use a zener diode to open up the circuit. For your 12v line use a zener with a trip point of say 12v and have a simple +0.5v input on the output side of the zener - bingo, 0.5v boost. Put thsi on the input side of the cap and itll smooth it right out - you should end up with a much more steady voltage.

Just throwing ideas around - contact me via pm or ICQ if you want to discuss further m8 .

PiLsY.

[Charles Wirth]

Thanks Pilsy, willing to do what it takes to reduce fluxuations.

I was checking out caps at Frys and noticed most of them had higher voltage rating on the side, 1000uf 60v and higher can I use these?

[bito]

BT.....is this the mod. that Abit are doing to stabalise the VDDR and VTT readings from the IC7 MAX3 ???
May give it a try if that is all they will do if i rma.

Thanks

[nutcase]

first, A Mosfet Is Not a Voltage Regulator.. It is a Transistor( Field effect Transistor).. There are 3 leads on a Mosfet: Gate, Drain And Source..

In the Regulator circuit on Mobo's, It does nothing But switch On and off a determined rate by the signal at the gate Pin..

The actual regulation and control are done by the PWM (Pulse Width Modulator) which on the MAX3 is the Winbond chip in Bigtoe's Pic..

The Winbond Chip does the Voltage Monitoring and Control of the Mosfet's. The Winbond Chip will Monitor the Voltage and produce a Signal to the Mosfets to Turn on and off at a specific Rate to keep the voltage constant. The winbond chip will adjust the signal rate to keep the voltage constant during fluctuations.

adding larger Mosfets will increase the current capability of the rebulator, But will require Mods to the PWM to get the voltage right.

The Cap That Bigtoe is showing appears to be a timing Cap to the Winbond Chip... I might have to try this..

The Caps from Fry's are No good for Most Regulators. The Caps need to be what is Called Low ESR for best performance and fry's Only sells Standard Type Caps.. Because of the High Frequencies switching Regulators use, High Quality Caps become important..

[drunkenmaster]

fugger, just outta interest, whats your thoughts on the msi 875. I had taht first, wen to max 3(was gonna sell, forgot), so have both and tried a 865asus right now jsut to play about. Anyway, i seem to recall having a lotta trouble getting ram to run anything but spd timings, and some other oddities with the board. I don't have a lotta spare time, and wanna sell one and run the other in a vapo rig for now. ASus i dislike, no pat above stock, modded bios only 1:1, etc, etc. Anwyay, is the msi as good as the max 3 clock for clock on newer bios's? you found mem adjusting fine? Any help really.

[PiLsY]

Hag6br - I was trying to keep things simple . As I said search by code off the relevant mosfet and replace with the next model up.

Mosfets regulate voltage by switching on and off a set frequency - ergo in the most simplistic terms they are vregs. As you said they switch on and off to keep the current constant . I suppose modulate voltage would be a more exact term, but hey if I give everything away I wont be needed anymore . To pretty much everyone here it doesnt matter if a mosfet is a resistor or a kitten - as long as they know its function and how that function can be improved thats all that matters.

As to alterations at the PWM, I was under the impression that as long as you replaced a mosfet with merely a higher ampage capable model that no additional work would be required on timing. Im self taught as far as electronics are concerned, so there are some things I dont know, and some things I think I know that arent strictly speaking correct . We're all here to learn afterall.

As for the caps at frys - theyre no good at all. You need to be shopping at RS or some other pure electronic wholesaler to get the right ones. I tend to use Audio Grade caps where possible, but failing that most electrlytic aluminium low ESR caps will do. You dont want 60v caps though. You want to pick the cap as close to the voltage being supplied as possible, but higher than it. For example, a 3.3v supply would have 4v caps, 5v supply 6v caps and 12v supply 16v caps. The voltage rating is how much they can deliver if needed and the capacitance is how much they can store. Higher capacitance means they can deliver current for longer, but conversely they also take longer to charge. Its a fine art balancing capacitance until you receive the optimum smoothing effect. Multiple low capacitance models in parallel would likely be better. I wouldnt recommend going much above 2200uF, and never above 3600uF. The extra charge time will mean a large voltage dip and spike on power up as the caps charge.

Hag6br - you seem like a knowledgable chap - have you any idea where the vref is supplied from on the IC7? If you check with a mm youll find that vref does some very weird things as voltage is applied. It should be strictly 1/2 of vdd(q) +-0.005v, but its out by a long long way.

Cheers,

PiLsY.

[EDIT] Drunkenmaster - pm me if you want to know how to get PAT enabled above 200fsb on that asus 865 board .

[nutcase]

I have not found anything on the Vref Voltage yet on the IC7. I am still trying to find a data sheet on the Winbond chip used on the ic7 but have had no luck.

The only thing I have found on Vref is in Intel's Design Guides for the 875P chipset. By Them they reccomend using A resister divider network at the NB and also at the Dimms. But, I cannot find these dividers on the IC7 Max3, so they did it differently.

Here is a link to the Intel Design Guide:

http://www.intel.com/design/chipsets...nex/252527.htm

I would like to get the Vdimm Stabilized on the IC7 Max 3 as I can tell that this Board is capable of some Great overclocking except for this problem.

As for Mosfets, Higher Power Mosfets require A slightly Higher voltage to switch at the Gate of the mosfet and sometimes require some mods to work correctly. But most new PWM Modulators now directly connect to the Mosfet, so it is not as critical as it used to be.

The PWM Timing Caps are probably the most critical item. If these change Value, All types of Wierd things happen (learned this from working on TV's) such as Way High Voltages, fluctuating Voltages, Etc..

[CATCH22ATPLAY]

I believe VTT is on pin #7 of that winbond chip. Someone correct me if i'm wrong. The left side of MC50 reads 1/2 of VDD also. Vref tracks exactly 1/2 of VDD and uses a resistor divider network at bottom left of dimm 3. Right above the big cap. But that's ONLY Vref on pin 1 of ddr. If only VTT and CK0 and it's sister and CK2 and it's sister and various other ddr pins which we'll call VTT (although it really isn't) tracked to 1/2 of VDD we'd be alright. Since the Japaneese or Chineese (i forget who) started theese IC7 mods we've come along way. They have a mod for VTT and it's been around a long time. I don't really feel like cutting a trace on my 4th IC7 (2nd MAX3). I can get 247 cas2-2-2-5 out of my Corsair pc3500. Strange but my Mushy 3200 used to be my best ddr. This MAX3 i currently use just doesn't seem to like it. May be that VTT crap F knows. I guess if i get bored i'll probably end up modding VTT.

Working on an alternative mod for VTT atm actually. It's not propper, but may help a bit and requires no cutting or soldering. I'm not hopeful that will work. Hmm now that i think about it i got another alternative i may try this weekend if this one tonight fails. Does require soldering. But at least you don't have to cut traces. Hmm since i got 2 dead IC7's and they are the same as a MAX3 i'll use em for trials. 1 of em actually still runs. But only at 1/2 cpu speed and you can't adjust anything. Condensation fried a capacitor on the northy substrate. I wish i could solder good enough to replace that cap. That was my best board. ARG i'll see wtf happens

[Tony]

I firmly believe we can sort this boards problems out.I have a feeling VDD and Vref will need to be sorted properly and we will then submit our findings to Abit...the board will then become a god!!

lets keep the ball rolling on this one, the iron is hot and ready , i just need instructions on what to do

[CCC]

i read in the abit forums that there is a new IC7-MAX3 v1.1 out that fixes the fluctuation isssues...

I finally got my r404 cooler from herefishy and hooked it up to my ic7max3/3.0c and when ever i start upping the volts i get blue screens everytime.. Its been frustrating the heck out of me... Then i noticed the ddr voltage, set at 2.9v, was fluctuating between 2.72v and 2.98v

It seems that everything on the board is fixed now, including the voltage flucuations, VTT and ECN fix. Thread started here:

HERE is the link....


Anyone know where i can pick up a v1.1????? Tried newegg and they wont guarantee it, zipzoomfly is supposed to call me back and let me know if they have any...

[PiLsY]

Catch22 - if you measure Vref from within pin1 on a dimm slot youll find a different story altogether . As vdimm is raised via bios vref goes horribly wrong. At 2.9v it resets to 1.25v and starts tracking over again. 2.5 to 2.8v is perfect, 2.9v to 3.2v vref starts again as if from 2.5v.

You have lots of voltages with ddr - vref (1:2) (internal io voltage), vtt (1:2) (internal io voltage termination), vdd (1:1) (external io voltage), vddq (1:1) (external referance io voltage), vdds (1:1) (no idea!) and then vTT for the external voltage too (1:1). All these voltages need to track perfectly at the appropriate setting, which they dont by a long shot.

Forgive me if the explanation of what does what is not totally accurate - its what ive gleaned from a samsung pdf .

PiLsY.

[nutcase]

Does anybody know where I can get the Data sheet for that Blasted WinBond Chip on the IC7 Max3?

Kind of hard to figure out a Fix without any Data...

[CATCH22ATPLAY]

Originally posted by PiLsY
Catch22 - if you measure Vref from within pin1 on a dimm slot youll find a different story altogether . As vdimm is raised via bios vref goes horribly wrong. At 2.9v it resets to 1.25v and starts tracking over again. 2.5 to 2.8v is perfect, 2.9v to 3.2v vref starts again as if from 2.5v.

You have lots of voltages with ddr - vref (1:2) (internal io voltage), vtt (1:2) (internal io voltage termination), vdd (1:1) (external io voltage), vddq (1:1) (external referance io voltage), vdds (1:1) (no idea!) and then vTT for the external voltage too (1:1). All these voltages need to track perfectly at the appropriate setting, which they dont by a long shot.

Forgive me if the explanation of what does what is not totally accurate - its what ive gleaned from a samsung pdf .

PiLsY.

OK here's the deal on my mobo and it may not be the same on yours. I use a multimeter and not hardware monitor. If i set vdd to 3.2 i check vdd at ddr pins and i get say 3.26 Then i check Vref at pin 1 of ddr and i get exactly 1/2 of that like 1.63 Vref is not being sent from the winbond chip and it's not being sent from the 2 upper right mosfets. A resistor divider network located between dimm 2 and 3 bottom right corner check the vdd after vdd mosfets but before it goes to ddr and from there the signal is divided in half with 2 resisters. example to get exactly 1/2 would be a 500 ohm and a 1k ohm resistor. Although i think they might be a 100 ohm and 50 ohm resistor. Now if your vdd fluctuates like mad then Vref will fluctuate like mad as well.

All the other 20+ pins that we'll cal VTT use the upper 2 mosfets and are being controlled by the winbond chip. Problem is the winbond chip only regulates VDD up to 2.80 max So max you can ever get VTT is 1/2 of that 1.4 volts. On the max 3 they are basically doing the same mod we did to increase vdd. They are tricking the winbond chip. If i set vdd to 3.2 volts in the bios and i check R50 on the upper side i get 3.26 but on the bottom side i get 2.80. Winbond chip thinks VDD is at 2.8 so regulates VTT to 1/2 of 2.80 which is 1.40 volts. Only fix i can find so far is to cut a trace and preferably NOT between MC50 and pin 7 of the winbond chip. But cut a trace somewhere else. Then solder a wire from left side of MC50 to one of the mosfets located in the upper corner. But also using a resistor divider network to properly control the exact voltage to 1/2 of actuall vdd. Cutting traces as we speak looking for a good place.

[CATCH22ATPLAY]

OK i found the correct place to cut the trace. It's on the backside of the mobo. It differs from the Japaneese way of doing it. In that my way will show voltage correctly even in the bios. All i gotta do now is try it and see what value resistor i need to use. It will go from the left side of MC50 to upper right mosfet of mobo.

[axiom]

good job Catch22, now all you need to do after finishing this is send your resume to aBit so they can hire you and fire all the other guys!

[CATCH22ATPLAY]

Originally posted by axiom
good job Catch22, now all you need to do after finishing this is send your resume to aBit so they can hire you and fire all the other guys!

HAHAHAHA Hell i better make sure it works first. DOH!!!!

[scanido]

great work CATCH22ATPLAY!

Good luck...

[axiom]

Catch22, any word on your mod yet? Is it working, did you finish it?

[CATCH22ATPLAY]

Finished? er yes and no. Yes in that it works fine. No in that i'm gonna change it a bit later on. But here it is. The first 4 pics basically show you where to cut the trace. Look at my arrow and you will see a trace by itself that goes up and to the right. Cut that one. Then ohm out from bottom leg of upper right mosfet on mobo to left side of MC50. It was 1 ohm and should be reading say 5 mega ohms now. I then solder a 500 ohm vario set at 320 ohms to left side of MC50 and bottom leg of mosfet located in upper right corner. Then i used a second vario connected to my first. Connect it to the first Varios side that connected to MC50. Then connect other side of second vario to ground. I used a 2k ohm vario and had it set to 2k ohms for a 0.20 increase in VTT. I had my bios set at 2.80 I increased VDD to 3.2 with my VDD mod. The VTT mod came to 1.60 with the 500 and 2k ohm resistors. Sorry about the webcam. I hope this helps. Winbond will read volts as 1.40 btw. So use a multimeter on upper mosfets to check voltage.
Volt modded the CATCH22 way














I removed MC50 completely. Also if you remove MC64 it really helps VDD mod a lot. I was able to increase VDD and VTT to 3.4 and 1.7 with theese mods on my non max3. It had no problems booting at all. Luck dudes and i'm ready for questions. Also did the mod on my max3.