迅维网

标题: Palit RTX3080 PCI-e not detected [打印本页]

作者: vusalcrm    时间: 2022-2-26 04:14
标题: Palit RTX3080 PCI-e not detected
hello to all friends.
When I looked at the motherboard visually, one integrated was broken.

1.jpg
登录/注册后看高清大图


After 3-4 days of research, I found what it was.
2 input and gate logic. NL17SZ08. marking code L2

this circuit is related to pcie rst signal. I could not find the same logic. i replaced this logic instead - MC74VHC
MC74VHC- 2 input logic and gate

2.jpg
登录/注册后看高清大图



all voltages are active on the video card.
gpu is normal hot. Ohms are all normal. no short circuit. only the motherboard is not detecting the video card.
I'm only getting images from the vga port on the motherboard

When I examine the eeprom with the oscilloscope, there is no reading.

作者: 中博科技    时间: 2022-2-26 08:30
帮他翻译了一下

各位朋友好  当我目不转睛地看着主板时,

一个集成电路板坏了  经过3-4天的研究,我发现了它是什么。 2输入和门逻辑。NL17SZ08。标记代码L2 该电路与pcie rst信号有关。我找不到同样的逻辑。我替换了这个逻辑——MC74VHC

视频卡上的所有电压都处于激活状态。 gpu通常是热的。欧姆都正常。没有短路。只有主板没有检测到显卡。 我只从主板上的vga端口获取图像 当我用示波器检查eeprom时,没有读数

作者: Poko    时间: 2022-2-26 09:05
Hello. If the GPU is not getting PEX_RST_BUF* signal, the EEPROM will not be read by the GPU so it's expeceted that you won't see anything on it.

If you are getting the RST signal from your motherboard at pin 1&2 at the AND gate, then you should be getting the buffered version of the RST signal at pin4, which goes directly to the GPU via a resistor ( looks like the one sitting diagonally to the right of the AND gate in your photo). You can take off the resistor and check the pad connecting to the GPU to see if well soldered on the GPU side.

If it looks like the GPU is getting power normally, and it's getting clock from the crystal on board, and the buffered RST is going to GPU, assuming some peripherals (power control I2C, temp, etc.) are good, a good GPU die shall try to read from BIOS EEPROM at this point.

Hope this helps.
作者: wang245582    时间: 2022-2-26 10:39
厉害了厉害了看不懂

                               
登录/注册后看高清大图

作者: 619881623    时间: 2022-2-26 11:18
拔掉内存电流不动
作者: vusalcrm    时间: 2022-2-26 23:39
Poko 发表于 2022-2-26 09:05
Hello. If the GPU is not getting PEX_RST_BUF* signal, the EEPROM will not be read by the GPU so it's ...

i check resistor 100k- it is OK
pull down res.  this is resistor 1 pin connect GND.
作者: aideheqq    时间: 2022-2-27 09:05
等高手来判断
作者: youyafei    时间: 2022-2-27 09:18
ROM芯片测不到波形,ROM对地电阻可正常,再测试gpu工作条件,还有27M晶体有没有波形。再不行50包邮我回收了
作者: 小莹虫    时间: 2022-2-27 09:24
你翻译给他说 鬼佬来迅维学校系统学习吧。别再瞎搞胡搞了!
作者: fumin668    时间: 2022-2-27 09:26
看不懂,谁懂?给我们说下
作者: Poko    时间: 2022-2-27 09:43
vusalcrm 发表于 2022-2-26 23:39
i check resistor 100k- it is OK
pull down res.  this is resistor 1 pin connect GND.

You are right, that is a pull down resistor. But did you verify the signal (non GND terminal) is going into the GPU correctly? You can do that be measuring the reverse bias diode resistance of it to GND using a DMM's diode function. If connection is good, it shall read 400 Ohms ~ish if I'm not mistaken.

Beyond that, did you verify this AND gate's pin 1 and 2 are getting the reset signal (PEX_RST_AND*) correctly? Not sure about your particular board but on my reference it is coming from another AND gate on the other side of the PCB (U54). It is an AND of the reset signal from PCIE connector (PEX_RST*) and 3V3_SEQ. You can bypass this AND gate by populating a 0 ohm resistor to R711.
作者: vusalcrm    时间: 2022-2-28 04:54
youyafei 发表于 2022-2-27 09:18
ROM芯片测不到波形,ROM对地电阻可正常,再测试gpu工作条件,还有27M晶体有没有波形。再不行50包邮我回收了

27mhz signal ok
作者: testpoint    时间: 2022-2-28 06:49
check the No.4 pin of NL17SZ08, if that is high level (3.2V), that is OK, if not, check the resistance by your DMM with no power. as usual, GPU pin broken (short) will cause that.
作者: vusalcrm    时间: 2022-2-28 16:38
testpoint 发表于 2022-2-28 06:49
check the No.4 pin of NL17SZ08, if that is high level (3.2V), that is OK, if not, check the resistan ...

and gate logic check
1-2 pin > 3.3v OK
4 out - 1.8Vv OK
5 VCC - 1.8v OK
作者: vusalcrm    时间: 2022-2-28 16:41
This model does not have a boardview file. but compatible with msi model
rtx3080

MSI GeForce RTX3080 MS-V389 Rev5.0 Boardview(CAD).cad

8.56 MB, 下载次数: 9, 下载积分: 下载分 -2 分, 下载 1 次


作者: vusalcrm    时间: 2022-2-28 16:46
Poko 发表于 2022-2-27 09:43
You are right, that is a pull down resistor. But did you verify the signal (non GND terminal) is g ...

This model does not have a boardview file. but compatible with msi model
logic 1-2pin 3.3v ok
u54 without  - R711 with - 3.3v send
作者: vusalcrm    时间: 2022-2-28 17:12
FBVDDQ - 63.8ohm - diode - 0.094
NVVDD - 0.3ohm - diode - 0.002
MSVDD- 0.5 ohm - diode - 0.002
PEX_DVDD - 8.1 ohm - diode -0.015
1V8 - 313 ohm - diode -0.341
5V - 319 ohm - diode - 0.526
作者: Poko    时间: 2022-2-28 18:19
vusalcrm 发表于 2022-02-28 16:46
This model does not have a boardview file. but compatible with msi model
logic 1-2pin 3.3v ok
u54 without  - R711 with - 3.3v send

so you can see the buffered reset signal going into GPU toggling when power on using an oscope, and bios Flash is not getting anything? No CS no MOSI no CLK?

Then some peripheral fault condition is stopping the GPU from initialization...
作者: qinhecat    时间: 2022-2-28 19:35
1V8 - 313 ohm - diode -0.341
1V8 power's res is  abnormal,normally it will be 10~30k ohm or above.The GPU may be damaged。
作者: vusalcrm    时间: 2022-2-28 23:15
Poko 发表于 2022-2-28 18:19
so you can see the buffered reset signal going into GPU toggling when power on using an oscope, an ...

CS no MOSI no CLK - I checked the start time with an oscilloscope.
There is no data reading other than voltage. there is only constant voltage


作者: vusalcrm    时间: 2022-2-28 23:21
qinhecat 发表于 2022-2-28 19:35
1V8 - 313 ohm - diode -0.341
1V8 power's res is  abnormal,normally it will be 10~30k ohm or above. ...

I will give low voltage from outside and try.
作者: vusalcrm    时间: 2022-3-2 03:33
I removed one leg of the 1.8v coil
The value on the pwm side is high ohm (9-10kohm)
output was low. 340 ohms. I supplied voltage from outside. It would draw 0.02 current. but i didn't see anything warming up
作者: testpoint    时间: 2022-3-2 05:41
if rst signal is OK, please be are the diode voltage to ground, (red test lead to groud about 0.3, black to ground should be infinity).
作者: vusalcrm    时间: 2022-3-2 15:54
testpoint 发表于 2022-3-2 05:41
if rst signal is OK, please be are the diode voltage to ground, (red test lead to groud about 0.3, b ...

If I'm not wrong, are you asking to measure the PEX_RST_BUF signal in diode mode?
showed no value.
作者: Poko    时间: 2022-3-2 19:18
vusalcrm 发表于 2022-3-2 03:33
I removed one leg of the 1.8v coil
The value on the pwm side is high ohm (9-10kohm)
output was low ...

I measured my 3080's 1.8v rail and it's 400 ohm in resistance and 341 ohm in diode...

Maybe you don't have a problem here.
作者: 俗俗    时间: 2022-3-2 21:32
我英语不好,爱莫能助
作者: vusalcrm    时间: 2022-3-2 22:57
Poko 发表于 2022-3-2 19:18
I measured my 3080's 1.8v rail and it's 400 ohm in resistance and 341 ohm in diode...

Maybe you ...

I watched a video on youtube. If the person measuring there had an ohm value of 300, it sounded like there was a short circuit. But I couldn't be completely sure.

then that line is normal.

PEX_RST_BUF
When I measure the reset signal in diode mode, it shows no value.
作者: testpoint    时间: 2022-3-3 03:17
vusalcrm 发表于 2022-3-2 15:54
If I'm not wrong, are you asking to measure the PEX_RST_BUF signal in diode mode?
showed no value ...

OK, GPU down, good luck
作者: Poko    时间: 2022-3-3 10:08
vusalcrm 发表于 2022-03-02 22:57
I watched a video on youtube. If the person measuring there had an ohm value of 300, it sounded like there was a short circuit. But I couldn't be completely sure.

then that line is normal.

PEX_RST_BUF
When I measure the reset signal in diode mode, it shows no value.

I also checked my 3090. 1.8 rail measures 230 ohm in reverse diode. I also checked PEX_RST_BUF*, it reads 293ohm in reverse diode.

It sounds like your RST signal got desoldered on the GPU side. I would reball the GPU if I were you. Good luck!
作者: testpoint    时间: 2022-3-3 12:16
Poko 发表于 2022-3-3 10:08
I also checked my 3090. 1.8 rail measures 230 ohm in reverse diode. I also checked PEX_RST_BUF*, i ...

if you dismount the GPU chip, just use the DMM to test the pin to determine the problem.
作者: vusalcrm    时间: 2022-3-3 15:31
I talk to the customer. According to what he said, welding works were being done in the neighborhood. then at this time i connected the video card to the computer and suddenly the video card stopped
作者: vusalcrm    时间: 2022-3-3 15:38
Poko 发表于 2022-3-3 10:08
I also checked my 3090. 1.8 rail measures 230 ohm in reverse diode. I also checked PEX_RST_BUF*, i ...

then the values for the 1.8 line are normal there is no problem here
I remeasured
this time i connected the red probe to the GND
I connected the black probe PEX_RST_BUF
0.657

later
I connected the red probe PEX_RST_BUF
black probe GND
10 kohm value showed
作者: vusalcrm    时间: 2022-3-12 19:13
Poko 发表于 2022-3-3 10:08
I also checked my 3090. 1.8 rail measures 230 ohm in reverse diode. I also checked PEX_RST_BUF*, i ...

293ohm in reverse diode. - I did not fully understand this measure. Can you explain a little bit?
maybe it doesn't belong to this topic, but i can't solve the logic of using diode mode with dmm. Is the value 0.400mv a voltage value applied to the line there?
作者: 37592179    时间: 2022-3-12 19:57
小莹虫 发表于 2022-2-27 09:24
你翻译给他说 鬼佬来迅维学校系统学习吧。别再瞎搞胡搞了!

看他样子不像瞎胡搞,技术应该有相当,他可能没有图纸,要来这边问人而已,在油管很多维修大神主播的,特别是俄国人的都是高手
作者: Poko    时间: 2022-3-13 03:14
vusalcrm 发表于 2022-03-12 19:13
293ohm in reverse diode. - I did not fully understand this measure. Can you explain a little bit?
maybe it doesn't belong to this topic, but i can't solve the logic of using diode mode with dmm. Is the value 0.400mv a voltage value applied to the line there?

This is what I meant.

                               
登录/注册后看高清大图

作者: vusalcrm    时间: 2022-3-17 15:36
thank you for your comments
Most likely the gpu is bad. I returned the graphics card to the customer.
作者: vusalcrm    时间: 2022-3-17 15:37
Poko 发表于 2022-3-13 03:14
This is what I meant.

https://www.chinafix.com/thread-1306241-1-1.html
I would be very happy if you let me know your opinion.
作者: a6464863    时间: 2022-3-21 13:25
I have the same issue in my galax rtx 3070 lhr。all voltage is normal,All resistance values are normal, and the crystal oscillator is 27m.but mainboard can't detect the gpu device。rstbuf have 1.8 voltage。maybe is the 30 series new problem。
作者: vusalcrm    时间: 2022-3-22 03:26
a6464863 发表于 2022-3-21 13:25
I have the same issue in my galax rtx 3070 lhr。all voltage is normal,All resistance values are nor ...

the problem of that integrated is chronic. because a few friends also faced the same problem




欢迎光临 迅维网 (https://www.chinafix.com/) Powered by Discuz! X3.4