-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
9600 np volt mod
- Thread starter bkraptor
- Start date
- Thread Starter
- #23
Ok.....
Let me get a deep breath and calm down
I did both the VDD and VDDQ mods. Working on increasing the VDDQ (after I had previously increased the VDD to 3.6 from 3.35) I got weird artifacting while I was in the POST screen paused - junk text, weird lines, striked out letters, etc. At that point my multimeter read ~3V on the VDDQ which was 0.6V under VDD. I shut down the PC as fast as I could (forced power off by turning off the PSU). After that I turned off the VDDQ mod and the artifacts were still there. I then lowered the VDD to default voltage, but the artifacts wouldn't go away! I panicked and turned off the VDD mod aswell. Still the artifacts remained. I tried taking out all the current remained in the capacitators by pressing the power-on button while the PSU was turned off. That didn't work either. After that I took out the video card and inspected it. Nothing smelled funny and I thought it was ok to put it back in. And here it is working perfectly again. It did scare the **** out of me
Can you L337 M33P explain what actually happened?
Let me get a deep breath and calm down
I did both the VDD and VDDQ mods. Working on increasing the VDDQ (after I had previously increased the VDD to 3.6 from 3.35) I got weird artifacting while I was in the POST screen paused - junk text, weird lines, striked out letters, etc. At that point my multimeter read ~3V on the VDDQ which was 0.6V under VDD. I shut down the PC as fast as I could (forced power off by turning off the PSU). After that I turned off the VDDQ mod and the artifacts were still there. I then lowered the VDD to default voltage, but the artifacts wouldn't go away! I panicked and turned off the VDD mod aswell. Still the artifacts remained. I tried taking out all the current remained in the capacitators by pressing the power-on button while the PSU was turned off. That didn't work either. After that I took out the video card and inspected it. Nothing smelled funny and I thought it was ok to put it back in. And here it is working perfectly again. It did scare the **** out of me
Can you L337 M33P explain what actually happened?
- Joined
- Jun 5, 2003
- Location
- TEH INTERNETS
Increasing VDDQ to 3V meant that the input and output buffers/drivers were operating at higher voltage, so presumably the memory/GPU bus was operating at higher than normal voltages. I take it your GPU does not like higher bus signal levels.
I haven't heard of anything that would cause non-permanent damage to a chip, but if that had happened I would have also powered off, left it for a while and removed the mods.
Count yourself lucky
- Thread Starter
- #25
Bad news!
I thought I knew what caused the artifacts and wanted to replicate them to be sure. Now they won't go away!
Any idea what to do next?
PS: windows has close to no artifacts. the POST screen has little artifacts. The windows XP loading screen has tons of artifacts (thick lines).
I thought I knew what caused the artifacts and wanted to replicate them to be sure. Now they won't go away!
Any idea what to do next?
PS: windows has close to no artifacts. the POST screen has little artifacts. The windows XP loading screen has tons of artifacts (thick lines).
- Joined
- Jun 5, 2003
- Location
- TEH INTERNETS
Can you run any 3d-based apps such as 3Dmark?
I think the artifacts may be due to memory corruption...
I think the artifacts may be due to memory corruption...
- Thread Starter
- #28
Right now the card gives an image so corrupted that it is un-usable. Even the windows desktop and every window gets corrupted.
Is there any way I can find out what part exactly is deffective on it?
The only way I can describe the image output of the card is normal image with various coloured lines overlayed (garbage mostly).
Is there any way I can find out what part exactly is deffective on it?
The only way I can describe the image output of the card is normal image with various coloured lines overlayed (garbage mostly).
- Joined
- Jun 5, 2003
- Location
- TEH INTERNETS
The memory is most likely screwed. I get random lines when I OC the memory too high with ATi Tool. I could be one chip, it could be all the chips. A solution would be to replace the RAM but it is highly unlikely you will a) find the right RAM chips and b) successfully solder em in.
Just for kicks, what are the codes on the RAM? I can't see it in the pictures.
Just for kicks, what are the codes on the RAM? I can't see it in the pictures.
Last edited:
- Thread Starter
- #30
They are Hynix HY5DV281622DT-4.
I've taken off all mods and a weird thing happens. If I boot up with this card and I measure the VDD, it decreases from 3.3V down to ~1.8V where it starts to decrease slower, causing heavy artifacts on screen.
Is there any way a malfunctioning capacitator could cause that? (which I could replace, of course)
I've taken off all mods and a weird thing happens. If I boot up with this card and I measure the VDD, it decreases from 3.3V down to ~1.8V where it starts to decrease slower, causing heavy artifacts on screen.
Is there any way a malfunctioning capacitator could cause that? (which I could replace, of course)
- Joined
- Jun 5, 2003
- Location
- TEH INTERNETS
Ah now we are getting somewhere - good call
Lowering VDD would cause errors as the RAM is rated to run at that speed and at that voltage. If it is run at a lower voltage then the effect is to reduce the maximum usable frequency.
The gradual decline suggests to me that something is overheating - the little voltage regulation chip maybe? The capacitors are usually rated for 2x the line voltage (6.3V usually), and the MOSFETs have ratings that far exceed the voltages used. Anyway, I think that the artifacts of death are not due to the memory itself after reading this.
Have you tried underclocking the RAM to ~100MHz or so?
Lowering VDD would cause errors as the RAM is rated to run at that speed and at that voltage. If it is run at a lower voltage then the effect is to reduce the maximum usable frequency.
The gradual decline suggests to me that something is overheating - the little voltage regulation chip maybe? The capacitors are usually rated for 2x the line voltage (6.3V usually), and the MOSFETs have ratings that far exceed the voltages used. Anyway, I think that the artifacts of death are not due to the memory itself after reading this.
Have you tried underclocking the RAM to ~100MHz or so?
- Thread Starter
- #32
I can't underclock the RAM since the image is all garbage and I can't make out anything...
The regulator is not overheating, I touched it and it's not even warm. The capacitator I use when reading the VDD is rated for 10V and 470uF. The MOSFET has no ratings marked on them so I don't know, but they are marked APM3055L GF277.
The decreasing voltage happens almost at once, takes 2-3 seconds from 3.36V to 1.80V, where the decrease slows down to ~0.01V per second.
Is there anything I can tell you more so you can help me revive it?
Should I start looking for a replacement for the voltage regulator? (which I could also solder by myself)
[Later edit]
What I noticed is that the voltage doesn't actually decrease gradually, but it takes a big plunge from 3.36V down to around 1.80V just when the PC posts and the speaker beeps, after that it slowly decreases to around 1.73V where it seems to be stable. Hope it helps to determine what actually happened.
The regulator is not overheating, I touched it and it's not even warm. The capacitator I use when reading the VDD is rated for 10V and 470uF. The MOSFET has no ratings marked on them so I don't know, but they are marked APM3055L GF277.
The decreasing voltage happens almost at once, takes 2-3 seconds from 3.36V to 1.80V, where the decrease slows down to ~0.01V per second.
Is there anything I can tell you more so you can help me revive it?
Should I start looking for a replacement for the voltage regulator? (which I could also solder by myself)
[Later edit]
What I noticed is that the voltage doesn't actually decrease gradually, but it takes a big plunge from 3.36V down to around 1.80V just when the PC posts and the speaker beeps, after that it slowly decreases to around 1.73V where it seems to be stable. Hope it helps to determine what actually happened.
Last edited:
- Joined
- Jun 5, 2003
- Location
- TEH INTERNETS
YGPM
- Joined
- Dec 23, 2003
- Location
- Memphis, Tennessee
i had a similiar issue when i volt modded my 9600xt using this guide: www.unoid.net/9600pro/voltage.htm my xt had the same controller chips and the mod worked to increase the highest atainable core clock, but the memory would have weird issues. Even if i just did the core volt mod, the memory wouldn't clock much higher than stock. The higher i pushed the core voltage the slower the memory would have to run to be artifact free. Infact anything above 1.42vcore would cause the memory to have to be UNDERclocker. so i took the mod off.
- Joined
- Dec 23, 2003
- Location
- Memphis, Tennessee
www.unoid.net/9600pro/voltage.html that should work
- Joined
- Dec 23, 2003
- Location
- Memphis, Tennessee
btw i tested the memory voltage at the capacitor and it would vary slightly, i checked the voltage with my simpson 260 and it would read some serious variation going on with the digital multimeter i could tell that it was doing it aswell. about 3.3 stock it would drop to the high 2v range but that would cause the memory to have to run slower. Apparently the core was robbing the memory of power.
I run into the same issue when I raise the vgpu to above 1.42v.
as it is now ,I can run 630/745 on my Sapphire 9600xt. That's with vmem (2.95v) and vgpu (1.42v) mods.
Is there a way to solder the 3.3v line from the PSU to maybe a mosfet on the card somewhere to provide the extra juice that the increase in vgpu takes away from the mem?
however, I was reading in Xtreme forums that there is also a vddq mod as well that could bring even more life out of my card. however, I haven't gotten a response as to what chip that would be from.
L337 M33P, would you be able to tell from this pic? (hope this pic helps some other people also if they're looking for locations)
as it is now ,I can run 630/745 on my Sapphire 9600xt. That's with vmem (2.95v) and vgpu (1.42v) mods.
Is there a way to solder the 3.3v line from the PSU to maybe a mosfet on the card somewhere to provide the extra juice that the increase in vgpu takes away from the mem?
however, I was reading in Xtreme forums that there is also a vddq mod as well that could bring even more life out of my card. however, I haven't gotten a response as to what chip that would be from.
L337 M33P, would you be able to tell from this pic? (hope this pic helps some other people also if they're looking for locations)
- Thread Starter
- #39
I took the card back from the "repair shop". They told me they had no clue what was wrong with the card and claimed to have replaced all memory chips (even though they're exactly the same now, thermal paste on them included ) w/o getting the card to work properly. They claimed to have tested ("measured" was their way of saying it) the APL431 chips and that they all worked perfectly. What disturbed me was that the card failed to even give out an image. Being curious out of nature I took the HSF off and noticed that the chipset was chipped on one corner (and I had thoroughly verified every aspect of my card before handing it to them, including the chipset). Pretty unprofessional of them, I must say
Well... there's at least one positive aspect: they didn't ask for any money in return.
I'll now have to wait a bit before I have enough money to buy a new card. I might wait for the new generation of cards to be out so I will be able to buy a cheap 9800XT or a FX5950.
) w/o getting the card to work properly. They claimed to have tested ("measured" was their way of saying it) the APL431 chips and that they all worked perfectly. What disturbed me was that the card failed to even give out an image. Being curious out of nature I took the HSF off and noticed that the chipset was chipped on one corner (and I had thoroughly verified every aspect of my card before handing it to them, including the chipset). Pretty unprofessional of them, I must say Well... there's at least one positive aspect: they didn't ask for any money in return.
I'll now have to wait a bit before I have enough money to buy a new card. I might wait for the new generation of cards to be out so I will be able to buy a cheap 9800XT or a FX5950.
- Joined
- Jun 5, 2003
- Location
- TEH INTERNETS
Time for a triple whammy of replies:
Man that sucks - I bet it would have worked if you'd replaced the controller chip, but as they saved you the trouble and killed it with the chipped core, you won't ever know
NST6563: I don't know what voltage lines are used to derive the VGPU and VDDR for your card. My design uses 3.3V for GPU, 5V for DDR and 12V for DDQ, but yours is different and by the sound of it you have one line powering both voltages. There should be some line capacitors to smooth out any disturbance the regulator generates. You could try measuring the voltages on some of the surface mount caps on the other side of the board.
Man that sucks - I bet it would have worked if you'd replaced the controller chip, but as they saved you the trouble and killed it with the chipped core, you won't ever know
NST6563: I don't know what voltage lines are used to derive the VGPU and VDDR for your card. My design uses 3.3V for GPU, 5V for DDR and 12V for DDQ, but yours is different and by the sound of it you have one line powering both voltages. There should be some line capacitors to smooth out any disturbance the regulator generates. You could try measuring the voltages on some of the surface mount caps on the other side of the board.
Similar threads
