Seller Note “Gtx 770 2gb graphics card powers on but faulty.”
Summary
- Resistances
- VCore – 15.2Ω
- VMem – 166.2Ω
- PEX – 948Ω?
- 5V – 506.6Ω
- 3.3V – 852Ω
- 12V PCI – 4KΩ
- 12V PCI-E 2.7KΩ
- Bought as a sister card to MSI GTX 770 2GB (2048 MB) Twin Frozr (A). Nice looking card, also has artefacting issues like the sister card, but not quite as bad.
- Vram: Elpida W2032BBBG-7A-F
- Initial reflow of A1 lasted over 7 months but eventually failed (from my early days when I thought reflows were ‘a fix’, they often don’t last)
- Replaced VRAM chip A1, seems to work now.
Next Steps
With no obvious power issues/shorts, a Mats memory test could be good to see if the artefacting is memory related, although it may be the GPU.
SUBPART RANK0 RD ERR RANK0 WR ERR UNKNOWN ERR
------------- ------------- ------------- ------------
FBIOA[ 31: 0] 0 0 0
FBIOA[ 63: 32] 0 514933 0
FBIOB[ 31: 0] 0 0 0
FBIOB[ 63: 32] 0 0 0
FBIOC[ 31: 0] 0 0 0
FBIOC[ 63: 32] 0 0 0
FBIOD[ 31: 0] 0 0 0
FBIOD[ 63: 32] 0 0 0
Rank 0 Failing bits:
A040 A041 A042 A043 A044 A045 A046 A047
Read Error Count: 0
Write Error Count: 514933
Unknown Error Count: 0This is very similar to the MATS failure on the other MSI GTX 770 2GB (2048 MB) Twin Frozr (A), part of the first memory chip.
Update 12/06/2022 – The memory chip later failed after 7 months+ of intense gaming!
Reflows don’t always last! This one lasted probably 7 months of sustained use. What is a little interesting this time is that I assumed the chip A1 was failing because I could see the same artefacts:
However, I think there could be some variability with heat and card flex, as it took several goes to get MATS to detect any memory issue. I re-ran with 10Mb, 50Mb and only 100Mb showed a slight error:
mats version 367.38. Testing GK104 with 100 MB of memory starting with 0 MB.
Errors found. Use -matsinfo for details.
This message will only appear once.
SUBPART RANK0 RD ERR RANK0 WR ERR UNKNOWN ERR
------------- ------------- ------------- ------------
FBIOA[ 31: 0] 0 0 0
FBIOA[ 63: 32] 0 1663 0
FBIOB[ 31: 0] 0 0 0
FBIOB[ 63: 32] 0 0 0
FBIOC[ 31: 0] 0 0 0
FBIOC[ 63: 32] 0 0 0
FBIOD[ 31: 0] 0 0 0
FBIOD[ 63: 32] 0 0 0
Rank 0 Failing bits:
A047
Read Error Count: 0
Write Error Count: 1663
Unknown Error Count: 0
BIT RANK0 WRITE RANK0 READ UNKNOWN
--- ----------- ---------- -------
A047 9 0 0
A047 5 0 0
A047 10 0 0
A047 9 0 0
A047 210 0 0
A047 279 0 0
A047 604 0 0
A047 537 0 0
ADDRESS EXPECTED ACTUAL REREAD1 REREAD2 FAILBITS TPSEIB ROW COL
---------- -------- -------- -------- -------- -------- ------ ---- ---
0001109d7c 55555555 5555d555 5555d555 5555d555 00008000 WA1027 0088 01f
0001109d78 55555555 5555d555 5555d555 5555d555 00008000 WA1026 0088 01e
0001109d74 55555555 5555d555 5555d555 5555d555 00008000 WA1025 0088 01d
0001109d70 55555555 5555d555 5555d555 5555d555 00008000 WA1024 0088 01c
0001109d68 55555555 5555d555 5555d555 5555d555 00008000 WA1022 0088 01a
0001109d60 55555555 5555d555 5555d555 5555d555 00008000 WA1020 0088 018
.......
000447d91c 55555555 5555d555 5555d555 5555d555 00008000 WA1007 0223 127
000447d918 55555555 5555d555 5555d555 5555d555 00008000 WA1006 0223 126
000447d914 55555555 5555d555 5555d555 5555d555 00008000 WA1005 0223 125
000447d910 55555555 5555d555 5555d555 5555d555 00008000 WA1004 0223 124
000447693c 55555555 5555d555 5555d555 5555d555 00008000 WA10c7 0223 1ef
if you are getting failure for first MBof FB then try option -no_scan_out
Error Code = 00000001
####### #### ######## ###
####### ###### ######## ###
## ## ## ## ###
## ## ## ## ###
####### ######## ## ###
####### ######## ## ###
## ## ## ## ###
## ## ## ######## ########
## ## ## ######## ######## Also, the card would sometimes load drivers, but the card crashes under load e.g. Kombuster. Hopefully, the card hasn’t developed a GPU-related issue and can be fixed by replacing VRAM chip A1 with a new chip.
After quite a lot of heat, 400+ degrees C of hot air to both remove and replace A1, the card now passes initial testing (Kombuster HD, MATS on 100Mb). I need to put the card back together and run more tests, but hopefully, it now has a more lasting fix.
Update 03/09/2022 Sigh.. failed again
The same chip, perhaps there’s something I am missing here? Maybe some of the nearby capacitors are faulty? The MATS report shows different bits from last time. The PC case where this has been running is somewhat unsuitable for a card that runs this hot (small, cramp, insufficient cooling).
mats version 367.38. Testing GK104 with 10 MB of memory starting with 0 MB.
Errors found. Use -matsinfo for details.
This message will only appear once.
SUBPART RANK0 RD ERR RANK0 WR ERR UNKNOWN ERR
------------- ------------- ------------- ------------
FBIOA[ 31: 0] 0 0 0
FBIOA[ 63: 32] 0 514906 0
FBIOB[ 31: 0] 0 0 0
FBIOB[ 63: 32] 0 0 0
FBIOC[ 31: 0] 0 0 0
FBIOC[ 63: 32] 0 0 0
FBIOD[ 31: 0] 0 0 0
FBIOD[ 63: 32] 0 0 0
Rank 0 Failing bits:
A040 A041 A042 A043 A044 A045 A046 A047
Read Error Count: 0
Write Error Count: 514906
Unknown Error Count: 0
BIT RANK0 WRITE RANK0 READ UNKNOWN
--- ----------- ---------- -------
A040 320 0 0
A040 94 0 0
A040 365 0 0
A040 58 0 0
A040 43101 0 0
A040 43734 0 0
A040 43815 0 0
A040 43835 0 0
A041 320 0 0
A041 94 0 0
A041 365 0 0
A041 58 0 0
A041 83557 0 0
A041 84190 0 0
A041 84271 0 0
A041 84291 0 0
A042 320 0 0
A042 94 0 0
A042 365 0 0
A042 58 0 0
A042 43101 0 0
A042 43734 0 0
A042 43815 0 0
A042 43835 0 0
A043 112 0 0
A043 94 0 0
A043 132 0 0
A043 58 0 0
A043 83557 0 0
A043 84190 0 0
A043 84270 0 0
A043 84291 0 0
A044 112 0 0
A044 94 0 0
A044 132 0 0
A044 58 0 0
A044 43101 0 0
A044 43734 0 0
A044 43814 0 0
A044 43835 0 0
A045 112 0 0
A045 94 0 0
A045 132 0 0
A045 58 0 0
A045 83557 0 0
A045 84190 0 0
A045 84270 0 0
A045 84291 0 0
A046 112 0 0
A046 94 0 0
A046 132 0 0
A046 58 0 0
A046 43101 0 0
A046 43734 0 0
A046 43814 0 0
A046 43835 0 0
A047 42765 0 0
A047 42542 0 0
A047 42808 0 0
A047 42506 0 0
A047 83557 0 0
A047 84190 0 0
A047 84271 0 0
A047 84291 0 0
ADDRESS EXPECTED ACTUAL REREAD1 REREAD2 FAILBITS TPSEIB ROW COL
---------- -------- -------- -------- -------- -------- ------ ---- ---
00009aae7c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1087 004d 0bf
00009aae78 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1086 004d 0be
00009aae74 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1085 004d 0bd
00009aae70 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1084 004d 0bc
00009aae5c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1087 004d 0b7
00009aae58 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1086 004d 0b6
00009aae54 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1085 004d 0b5
00009aae50 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1084 004d 0b4
00009a1e7c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e7 004d 07f
00009a1e78 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e6 004d 07e
00009a1e74 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e5 004d 07d
00009a1e70 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e4 004d 07c
00009a1e5c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e7 004d 077
00009a1e58 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e6 004d 076
00009a1e54 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e5 004d 075
00009a1e50 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e4 004d 074
00000acb3c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1047 0005 00f
00000acb38 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1046 0005 00e
00000acb34 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1045 0005 00d
00000acb30 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1044 0005 00c
00000acb1c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1047 0005 007
00000acb18 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1046 0005 006
00000acb14 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1045 0005 005
00000acb10 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1044 0005 004
00000acb08 00000000 00008700 00008700 00008700 00008700 WA1042 0005 002
00000acb00 00000000 00008700 00008700 00008700 00008700 WA1040 0005 000
00000a7b3c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10a7 0005 0cf
00000a7b38 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10a6 0005 0ce
00000a7b34 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10a5 0005 0cd
00000a7b30 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10a4 0005 0cc
.....
000049539c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10c7 0024 187
0000495398 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10c6 0024 186
0000495394 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10c5 0024 185
0000495390 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10c4 0024 184
0000495388 00000000 00008700 00008700 00008700 00008700 WA10c2 0024 182
0000495380 00000000 00008700 00008700 00008700 00008700 WA10c0 0024 180
00004bc93c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1067 0025 10f
00004bc938 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1066 0025 10e
00004bc934 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1065 0025 10d
00004bc930 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1064 0025 10c
00004bc928 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1062 0025 10a
00004bc920 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1060 0025 108
00004bc91c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1067 0025 107
00004bc918 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1066 0025 106
00004bc914 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1065 0025 105
00004bc910 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1064 0025 104
00004bc908 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1062 0025 102
00004bc904 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1061 0025 101
00004bc900 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1060 0025 100
00004b793c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1007 0025 1cf
00004b7938 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1006 0025 1ce
00004b7934 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1005 0025 1cd
00004b7930 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1004 0025 1cc
00004b791c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1007 0025 1c7
00004b7918 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1006 0025 1c6
00004b7914 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1005 0025 1c5
00004b7910 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1004 0025 1c4
000030937c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1027 0018 03f
0000309378 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1026 0018 03e
0000309374 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1025 0018 03d
0000309370 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1024 0018 03c
0000309368 00000000 00008700 00008700 00008700 00008700 WA1022 0018 03a
000030935c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1027 0018 037
0000309358 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1026 0018 036
0000309354 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1025 0018 035
0000309350 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA1024 0018 034
0000309348 00000000 00008700 00008700 00008700 00008700 WA1022 0018 032
0000309340 00000000 00008700 00008700 00008700 00008700 WA1020 0018 030
000030237c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e7 0018 0df
0000302378 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e6 0018 0de
0000302374 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e5 0018 0dd
0000302370 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e4 0018 0dc
000030235c 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e7 0018 0d7
0000302358 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e6 0018 0d6
0000302354 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e5 0018 0d5
0000302350 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10e4 0018 0d4
000062f2fc 00000000 0000ff00 0000ff00 0000ff00 0000ff00 WA10a7 0031 03f
if you are getting failure for first MBof FB then try option -no_scan_out
Error Code = 00000001
####### #### ######## ###
####### ###### ######## ###
## ## ## ## ###
## ## ## ## ###
####### ######## ## ###
####### ######## ## ###
## ## ## ## ###
## ## ## ######## ########
## ## ## ######## ######## I haven’t any more spare chips and feel like I will be seeing this card again, so I might try reflowing it. Here is the A1 chip with cleaned pads:
Actually, I will reball this one instead, hopefully this will lead to a better fix. Here is the reballed chip:
Great! Following replacement with the reballed chip, we now have a clear picture (previously there were artefacts even on the BIOS screen).
Heat has likely been a factor in the repeated failures of chip A1 on this card. For good measure, I’ll also change the thermal pads and repaste the core:
MATs tests are now passing (10Mb, 25Mb, 50Mb). Heaven benchmark is running nicely. I am very happy with this latest fix, but I do expect to be seeing this card again at some point. The little PC case and all the fluff I found due to it being used on a carpet, all add to the possible heat problems.
Update 15/12/2022 – A1 failing again in exactly the same way…
Well, here we are again! I would really like to discover the underlying issue here, the same chip just keeps failing, and this time the MATs 367 report seems identical to the last time.
- E1 – error count 515127
- Failing bits A040 A041 A042 A043 A044 A045 A046 A047
- Fail bits 0000ff00 and sometimes 00008700
- All the failing bits seem to be located in the top right corner of the A1 chip, on the core they don’t seem particularly near any corner or edge
Update 24/12/2022 – Heating the core to leaded solder melt point
A helpful guy on the Learn Electronics Repair Discord made an interesting suggestion that the issue could in fact be to do with the ‘solder bumps’ between the core die and PCB substrate. Heating the core to the leaded solder melt point is a way to explore this, as the solder bumps can apparently melt at a lower melt point (~180°C) than the core BGA solder balls (~230°C)
I did as follows:
- Covered the VRAM chip that reported an error in MATs (to avoid it getting reflowed)
- Preheated to 100°C+
- Used a wide hot air nozzle and heated the core to about 180-190°C for perhaps 20 seconds
The card now displays a clear picture, passes MATs and can take full load. So I guess this card may have a core ‘solder bump’ issue, because that shouldn’t be hot enough to reflow it. There is still the possibility that the heat somehow temporarily fixed the VRAM or core BGA.
Whilst the card is working, I fully expect it to fail and not before too long. As a secondary finding, I noticed when stressing the card that it it takes a lot of power (~250 Watts) and thermal limits even on an open test bench at 80°C. The VDDC voltage is also sporadic in it’s readings, so possible the VRM efficiency is impacted by some issue, which generates excess heat that in turn perhaps impacts on the VRAM / Core issues seen.