i removed the 100k resistor.. 10k @ vcore: 1.27vcore still no os boot
1.26vcore os boot
1.27vcore no os boot
something else in the way..
it sure looks ovp pin goes to the back of the card to that resistor.. i could be wrong
i removed the 100k resistor.. 10k @ vcore: 1.27vcore still no os boot
1.26vcore os boot
1.27vcore no os boot
something else in the way..
it sure looks ovp pin goes to the back of the card to that resistor.. i could be wrong
You'll have to use DMM to find continuity between pin 44 and a nearby component.
EDIT - perhaps altering the VID resistor arrangement is a way around this issue? See the table in the data sheet.
Last edited by CryptiK; 02-01-2009 at 12:41 AM.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
^ indeed.. vid resistor arrangement for higher vcore should work
ive read more @ datasheet.. looks like ocp occurs @ 2v+
If the voltage is higher than
2V, ISL6327 will go into the OCP mode
do you think vrrdy (pin 45) has also smtg to do with this?
VR_RDY - VR_RDY indicates that the soft-start is completed
and the output voltage is within the regulated range around
VID setting. It is an open-drain logic output. When OCP or
OVP occurs, VR_RDY will be pulled to low. It will also be
pulled low if the output voltage is below the undervoltage
threshold.
Last edited by NapalmV5; 02-01-2009 at 01:40 AM.
check this out:
Overvoltage Protection
Regardless of the VR being enabled or not, the ISL6327
overvoltage protection (OVP) circuit will be active after its
POR. The OVP thresholds are different under different
operation conditions. When VR is not enabled and before
the 2nd soft-start, the OVP threshold is 1.275V. Once the
controller detects a valid VID input, the OVP trip point will be
changed to the VID voltage plus 175mV.
Last edited by NapalmV5; 02-01-2009 at 01:50 AM.
VR_RDY appears to trigger the phase shutdown after OVP or OCP is triggered, so although it does have something to do with it, it's not the cause of the problem. The main issue here is the data sheet doesn't show how OVP is configured.
As you saw, before a valid VID is read its 1.275v and after its VID + 175mv. Getting around this is not straightforward. At this stage all I can think of is VID alteration unless I can figure out how to get around the OVP sensing the actual voltage.
However it seems that this VID + 175mv is set internally, as for example with the OCP point they explain which resistors control this, but there is no mention of how OVP is controlled with resistors or an R-C circuit.
Last edited by CryptiK; 02-01-2009 at 02:36 AM.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
vid resistor arrangement ?
ssc has a resistor at the spot where the amp does not
my ssc: 1.23v
my amp: 1.13v
Last edited by NapalmV5; 02-09-2009 at 03:35 AM.
looks like they have to be as there are none on the front side i can see corresponding to the VID pins.
However all those spaces are filled and I cant see any VID's that have all 1's. Unless the resistors differ in value so for ex a 30K is read as open and a 50K is read as closed.
What do those resistors measure? Are they all the same? Just wondering where the 30-50K on your pic comes from. They may be part of a circuit that adds to the resistance so the actual resistor may be less than that.
Also I have a non ssc model coming in a couple of days and i'd like to add a resistor in the space to raise VID to SSC level if i can confirm they are the VID resistors and confirm the value of the resistor to add there.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
dont mind the 30-50k
thats what i used to increase vcore and vmem on the amp.. posted that @ news forum thread
i did measure but i forgot.. let me measure again
be right back
So what exact resistance did you use in that space on the zotac to raise vgpu & vmem? (I will do this on mine when it arrives).
I'm just thinking if we can see a high/low resistance pattern and match it to the open/closed pattern in the VID table we could figure out which resistor is which in the VID table.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
22k produced 1.26vcore/2.17vmem
so there are 16 spaces.. on the ssc 1/2/3/11 are occupied where 1/3/11 are occupied on the amp the rest are vacant
on my not so great dmm all resistors are of the same value: 1ohm
gotta get a better dmm for such finesse measuring
Last edited by NapalmV5; 02-01-2009 at 03:31 AM.
I can't see an obvious pattern there. Nvidia have been known to put VID resistors in strange arrangements, like 1,4,2,3 left to right for example.
Are the VID pins on the ISL6327 chip large enough to get the probe of a DMM on? If you can find which VID pins have continuity with specific resistors in that row, that should allow resistor identification.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
after more fiddling around i doubt any kind of vid resistor arrangement allows past the 1.27/1.3 limit.. they built it such.. no way around that if theres limit elsewhere
the 100k may be the vr_rdy why it allows 1.32vcore @ 100k added
all 3 right here are 10k which possibly have to do with the 1.3ovp
which one.. ill look into later on tonight
tried 10k resistor at all 3.. still no os boot
as for finding continuity my dmm is not up to the task.. will get one more appropriately
only so much info to go by from the datasheet.. im gonna let the experts have at it
once vcore freely gets more voltage.. clocks will take off
in the meantime.. cryptik/if you guys have anymore ideas.. let me know
after reading some more @ page 18-19
ill remove all mods and play around @ vid pinsISL6327 begins the
soft-start and ramps the output voltage to 1.1V first. After
remaining at 1.1V for some time, ISL6327 reads the VID
code at VID input pins. If the VID code is valid, ISL6327 will
regulate the output to the final VID setting
Last edited by NapalmV5; 02-01-2009 at 11:24 PM.
well looks like vid pins are more effective than resistor @ vcore
after several combos.. 1.00-1.25vcore most im getting
no resistor required just by connecting the points
so what is the combo to 1.3 1.4 1.5vcore ?
anyone ?
Last edited by NapalmV5; 02-02-2009 at 01:34 AM.
The OVP and VR_RDY pins are only logic level outputs to indicate the status of the voltage regulator, and shouldn't have any effect on the OVP threshold. The OVP threshold is derived by comparing VDIFF (effectively the output voltage of the regulator) and the reference VID value from a DAC + 175mV.
You might be able to tweek the reference voltage from the DAC by pulling the DAC pin up to a higher voltage (depending on the output impedance of the DAC), although doing this might damage the voltage regulator. If that doesn't work, the only other way to overcome OVP is to change the VID values.
I thought about that option but deemed it too risky considering the expense of these cards.
That's why I thought the only safe option is to attempt to modify the VID value through addition/removal or resistors/bridges. However its the identification of the appropriate resistors/spaces that is the tricky part without being able to test continuity between the VID pins themselves and the resistors.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
I did the VID mod to get 1.22 vgpu, 2.13/2.14 mem. I had 1.15/1.16 vgpu and 2.11v mem stock.
Did you figure out if bridging any of the other VID contacts raises the vgpu any more than just having 1/2/3/11 bridged?
Last edited by CryptiK; 02-04-2009 at 06:44 AM.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
any idea to 1.3 1.4 1.5 heck even 1.6 ?
bridge 4 and you may get 1.24-1.25v
after numerous combos on the amp i got to remove all 1/3/11 bridges.. i got as low as 1v and as high as 1.25v
havent had the time since then to try some more
its possible that the vid code validity goes hand in hand with the vid pins + vid bios.. im not sure since i didnt get to try
looks like nibitor maker finally released 4.8 but not the 4.9 going around that no one is sharing
dont know if 4.8 supports 285
still doesnt support 285
anyone out there willing to share 4.9 ?
Last edited by NapalmV5; 02-04-2009 at 12:48 PM.
just bridge the points by solder/wire
if you use 22k resistor @ one of the points or by bridging the points with a 22k resistor then you get 1.26vcore/2.17vmem
As Napalm said just bridge it with solder, the 'resistors' they use are 0 - 0.1 ohms anyway (basically a dead short between the 2 contacts). Max stable clocks did improve after the mod.
On my card (base model EVGA):
1/3/11 = 1.16 vgpu / 2.11 vmem
1/2/3/11 = 1.22 vgpu / 2.13-2.14 vmem
EDIT - bridged the 4th contacts (so now have: 1/2/3/4/11 bridged), now have 1.23 vgpu / 2.15 vmem under load, basically at SSC voltages now.
Raising voltage hasnt had a large effect on load temps, before mods on auto fan speed it was ~80*C in really heavy scenes of crysis, at 1.22v is was around 83*C, and now at 1.23v it's around 85*C (but today is warmer weather too). Fan doesn't go over 72%.
Last edited by CryptiK; 02-04-2009 at 11:29 PM.
Ci7 990X::Rampage III Extreme::12GB Corsair Dominator 1866C7GT::2 x EVGA SC Titans in SLI::Corsair AX1200::TJ07::Watercooled
Ci7 920 3849B018::Rampage II Extreme::6GB GSKILL Trident 2000C9 BBSE::EVGA GTX580::Antec Signature SG850::TJ09::Aircooled w/TRUE 120X
Great.....I have some silver conductive paint here, from the old Athlon days which i used for the same thing
Can be removed as well if need be. Thank you for sharing your efforts.
Thank you CryptiK and NapalmV5.
I bridged 2 and 4 as well today, for a nice little performance boost on my vanilla zotac. http://service.futuremark.com/compare?3dm06=9927577
Last edited by dengyong; 02-07-2009 at 03:55 PM.
nice ^
your welcome
@NapalmV5, is there any chance for pencil vcore mod for GTX 285, as softmod is impossible for GTX 285 and GTX 260 55nm?
Thank's!
|RIIIE||Xeon W3520#3845A935|EVGA GTX480||PC3 16000 Kingston HyperX 6GB||2x150GB WD VRaptor||WD6401AALS||HAF 932//CM 1250W||LenovoL220x||X-Fi Xtreme Music||Lycosa+Lachesis+Destructor+G27+P880|
WC-Loop1: CPU |EK-Supreme HF|Magicool Xtreme III||MCP355||MCRES| Loop2: VGA/NB| EK-FC480 GTX-Acetal / MCW30|MCP655|MCR220QP|MCRES|
Bookmarks