Printable View
It seems that no one has posted any Vmods of this motherboard. I am trying to vmod mine, and I would appreciate if the experts here can tell me if I am on the right track. (I probably won't mind killig the motherboard, but I certanly don't want to kill my last pair of Corsair BH-5s and the FX-53!)
Anyways, AV8 uses the Winbond W83303D for VDIMM, and soldering a 20ohm connecting the emplty soldering point to ground gives me 0.3V more, and a 10ohm 0.5V more.
Vcore is regulated by a 6559, so a VR to the sixth leg should do the trick.
VDIMM ca be measured from the right leg of the center MOSFET near the DIMM socket and Vcore can be measured from one of the coils.
Please let me know if you have any thoughts about the VDIMM mod. Thanks.
For the Vcore, on the ISL6559CB the target pin(s) would be either #10 (FB) or #13 (Vsen). Do a connectivity test between the coil measuring Vcore and those 2 pins to see which one (if not both) is connected to the Vcore read point. A vast majority of the time the FB pin is the target on most mods, however sometimes (Vcore) the Vsen is the actual mod target as FB is tied to otehr controllers and not directly tied to Vcore.
On the Vdimm, I am at school and dont have that data sheet with me, but I am pretty certain its pin #19 or #21 that controls Vdimm for the W83303D IC. You have already found where you can add resistance to boost the Vdimm, so I would think that a 200-1000 ohm VR would work nicely where you have the fixed resistor in your pic.
Keep in mind that since you are adding a resistance where none exists (as far as you are not "directly" soldering your VR to another resistor), that you will increase the resistance on the VR to increase the Vdimm. Whereas you would typically decrease the resistance of a VR to increase the voltage. So, be sure to preset the VR to 0 ohms before soldering it into place.
GL with it and let us know what happens.
However, be sure you first check which pin is needed for the Vcore mod and then you will need to do a resistance check between that pin and ground to see what value VR is needed for this mod.
Hi Fire, thanks for your help. I've been testing the VDIMM mod for a couple of hours (with a pair of cheap DDR400s, of course)and it seems to work fine. I am getting 3.25V (2.8V in BIOS)with the 20 ohm resistor, as shown in my photo. Nothing seems to be melting yet so I guess its probably OK.
As for the Vore it looks like its pin 10. I'll probably do this in the next couple of days when I have more time.
Cheers!
At 3.25v Vdimm, you will have zero problems with that BH5.
Let us know how the Vcore works out.
HKY, About the Vmem-mod.
Have you soldered the resistor to left or right empty soldering point? It's hard to say from picture :-)
Ok, I did the Vmem-mod but for some reason it's not working very well.
I can get 2,86V without mods and the board can run around 218 MHz 1:1 with BH-5 memory.
When I raise Vmem to 2,88V I get BSOD very easily and the higher voltages are even worse.
:confused:
Pics of my mods:
http://www.io-tech.fi/pic/modit/av8_03_1.jpg
http://www.io-tech.fi/pic/modit/av8_04_1.jpg
http://www.io-tech.fi/pic/modit/av8_05_1.jpg
http://www.io-tech.fi/pic/modit/av8_06_1.jpg
does the Vtt follows when u raise the Vdimm?
No I don't think so.Quote:
Originally Posted by enduracell
I believe if the winbond controller is the same as on the other abit board then that chip also controls the Vtt. Whichever pin it is you should try and find it via the datasheets because Vtt will play a good role in your ram's overclockability.
it doesn't matter what u think, it matters what it is. KV8 pro had Vdimm on 3.2 and reduced it to 2.8 on 1.5 BIOS because VTT didn't follow over 2.8 and caused major instability. So look in the Bios monitor screen for VTT sense or measure it on a fet around the neighbourhood.Quote:
Originally Posted by Sampsa
heh heh, VTT doesn't follow.. with 2,8V from BIOS I measured 2,86V Vmem and 1,43V VTT from motherboard.
When I add the 20ohm resistor the VTT remains same.
Any ideas for VTT-mod?
try this one, if this R is VTTsense... :rolleyes:
@Pc Ice ovp on AV8 is @1.95 and the mobo gives 1.85 out of the box...
I just received datasheet for W83303B from Winbond:
http://www.muropaketti.com/temp/w83303d.pdf
Thanks guys.
Sampsa is right. VTT does not follow and the increase in VDIMM is not very useful. I only got about 5mhz more from 2.8V to 3.15V.
Enduracell, I'll try that later to see how it works.
Endu, that doesn't seem to work. Nothing happened even when I turned the VR down to 5 ohms.Quote:
Originally Posted by enduracell
That mod wouldnt work.
You need to remove the resistor, its a 0 ohm resistor just bridging the gap, and replace it with a 200-1000 ohm VR soldered to both solder pads in order to increase the Vtt. The VR has to be set to 0 ohms before soldering it in place, not max.
And just in case it still acts odd, you may have to solder the other vr leg to ground. Just make certain that the vr legs that are soldered to the the sodler pads are the legs reading 0 ohms before soldering the VR into place.
HKY, check out the datasheet. The pin number 26. is the 1.2V_SEN and I think it needs some modification.
I've no idea what enduracell tried to tell with his instructions.
Ok my mistake I haven't done it, just speculating. Do what Hellfire says knows far better than me those thinks :toast:
I think is BT_SEN (pin 7), correct me if I m wrong. Also check this out.Quote:
Originally Posted by Sampsa
Guys, remember that there are 2 instances of Vtt with A64 boards. One is driving the cpu and one drives the memory termination bus. If you try to boost the voltage by modding to pin #26, say good bye to your cpu.
It is the Vbt pin (#7) on the W83303D IC guys. Trust me.
Ok, thanks. I'll try it later this week and let you know what happened, if HKY doesn't do it first :)
Yea I'm going to try out Fire's mod in the next few days.
Sampsa I have no idea which is the correct pin just by reading the data sheet but I am betting Fire is right. (I found the VDIMM mod by trial and error on the board only....)
Ok here is what I did:
http://www.io-tech.fi/pic/modit/av8_09_1.jpg
I removed the 0 ohm resistor which was attached to pin 7.
http://www.io-tech.fi/pic/modit/av8_10_1.jpg
http://www.io-tech.fi/pic/modit/av8_11_1.jpg
I replaced it with 1 kilo ohm VR and set it to 0 ohm.
Unfortunately Vtt didn't change when I raised the VR to 1000 ohms.
What next? Solder the other leg of VR to ground?
I did the same thing with the same result, and 5 minutes ago I connected the other leg of the VR to ground and now the system won't boot up any more, even after I removed the ground. It stopped after "C1", (which means checking DRAM), and then return to "02" and then "C1" again and again. It can't even be turned off by the power on switch. The RAMs are OK because I checked them in another system.
I am afraid that either the motherboard (maybe the uGuru chip) or the memory controller of the CPU is dead.... I'll have to find another 939 board to check it out. Hopefully its the motherboard!
Doh :(Quote:
Originally Posted by HKY
So why would it help to solder the remmaining leg to ground?Quote:
Originally Posted by Sampsa
Its alright, Sampsa. The second I lay my soldering iron on the board I'm prepared to damage it. Just hope that the CPU is still OK.Quote:
Originally Posted by Sampsa
Anyway check this out. I think this might work. (Keep trying man!)
http://xtremesystems.org/forums/showthread.php?t=38946
HKY, Hmm you have a VR with 2 or 3 legs? Just 2 legs and and you soldered the other to ground?Quote:
Originally Posted by HKY
If I undestood correctly from the KV8-Pro Vtt thread, I should solder the remaining leg of my 3 leg VR to ground?
:)
The VTT mod works fine the way I did it:
http://www.vr-zone.com.sg/Shamino/av8/av8vtt.JPG
I never bother connecting to the other solder pad veside the pad cionnected to pin 7 but direct to the mosfet leg. 0 ohms and work up wards :)
I used a 3-legged 1K trimmer, similar to the one you used. I soldered the remaning leg to groud.
I've confirmed that the FX-93 is dead. Maybe I've damaged the memory controller. So it might bot be related to the vmod. I think you should do the mod without the CPU. Find the measure point on one of the legs of the MOSFETs around the DIMM and measured from there. That's what I did when I found the VDIMM mod. This time I was lazy because I want to read Vtt from BIOS and its a lot of work to remove and install the Mach2 - big mistake! (LA I wish you were here earlier!)
Ok Vtt-mod is working. I used the same method what Goldlocke used with KV8-Pro.
1. I removed 0 ohm resistor next to Winbond chip pin 7
2. I replaced it with 10 ohm resistor
3. I soldered 1 kilo ohm VR between 10 ohm resistor (the pad which is nearer Winbond chip) and GND
VR should be set to MAX and lowering the value will raise Vtt
I'm now installing Windows and I'll post the results later. Currently running Vmem 3,02 V and Vtt 1,51 V.
Thanks for your help everyone!
:toast:
Quote:
Originally Posted by Sampsa
:toast: Good to hear that! I'll do the same after I get another CPU! (Please show a photo for the completed work for easy reference for other users!)
Here is the picture of working Vtt-mod for Abit AV8:
http://www.io-tech.fi/pic/modit/av8_kaavio.jpg
http://www.io-tech.fi/pic/modit/av8_12_1.jpg
NOTICE! Remember to set VR to MAX (500 - 1000 ohms). Vtt will raise when you lower the VR value.
:toast:
:toast:
Unfortunately Vmem and Vtt mods didn't help much and 220 MHz FSB with tight memory settings is the best I can do. 225 MHz isn't stable even with 3,25 V to memory. I tried with Kingston KHX3500, Corsair 3200XL and Geil Ultra Platinum memories (all of them are BH5).
That's strange! I can run at 250+ HTT with a paired of borrowed Geil Ultra-X at 2.5-3-3-10 before vmod so the motherboard should not be a problem. Have you tried increasing the chipset voltage?Quote:
Originally Posted by Sampsa
Which 2 slots are you using? Also I think there are a couple of MOSFETs supplying VDIMM. Please measure the voltages of them to see if there is a difference.
Good job on getting this mod worked out guys. :up:
Well 2.5-3-3-10 memorysettings aren't actually what I'm looking for.Quote:
Originally Posted by HKY
I'm trying to run at 2-2-2-5-7-12 :)
I know we are all aiming at 2-2-2-5 with out ol'BH-5s. That's why we are trying to Vmod it. I am just saying that the motherboard should have no problem running at 250+ HTT, so I found it strange that the VMOD didn't help your RAMs to run t least in the 240s. Too bad I don't have a 939 now so I can' test it. I'll probably wait until November to buy a new one, as AMD should reduce the prices then.Quote:
Originally Posted by Sampsa
Sampsa VTT way works perfect for me, but Vdimm mod mention on this thread cause me major drops under loads and reduced constantly. I did it this way and I m running with it right now. Far more accurate and is rock stable under load (memtest#5) w/ one dimm@3.3 and has a little drop (compared to previews way) w/ DC setup (2x512) 3.26 under load when idle 3.3. :)
(this is actually Hipro5 way on IC7 :toast: )
http://www.xtremesystems.org/forums/...chmentid=18491
note: I also removed R 87A from first pad (it was around 760 ohm and it was grounding, I don't now if it's necessary to do so, but I just mention it)
And cut only the trace! is a small one sorting the 2 points the R that I placed, u shall see it w/ naked eye (or a magnifier glass)
could anybody write a collection of all those mods (with pictures and detailled descriptions of course...)??????????????????
VagpSen is Pin34 track the trace down to find out how it can be done.
Maybe I search for it but not this week.
But what do u want it for above 1.65?
The bigger problem on this mobo is that oc the ΒΗ-5/6 like crap no matter the volts u feed it :(
http://www.xtremesystems.org/forums/...chmentid=19922
I have a following problem. Contact platforms of the resistor r84 have come off. I need find out where there are contacts of this resistor to return it back.