Sapphire RX-480 Nitro+ OC Repair Help

[Shonk]

Short story
Sapphire RX 480 Nitro+ 4 GB Died
2 x Blown 10a 8 pin PCIE Fuses
1 x Blown MLCC with a Short to ground C4132

I want to replace the MLCC C4132 and need a schematic or someone who knows what i should order
can anyone help?

atm its fixed and running with C4132 removed but want to replace it
as its now missing some filtering



Long Story
My Son has a 4GB Sappire RX-480 Nitro+ OC that i got for him in 2016 from the rainforest company
a few months ago (maybe 4) i took the back plate off so i could remove the front shroud to sort a fan vibration issue, the heatsink wasnt removed as i didnt need to and also didnt want to mess with the factory paste, on removing the back plate the pcb is a matte finish with no laquer top coat to protect it
like older graphics cards used to have, there was a black thick heat transfer pad to transfer heat to the back plate (green in image)

which had clearly badly degraded and was a gooey mess and it had leached an oil off of it onto the pcb the whole area in red had an oily shine, with it being worse the closer it was to the heat transfer pad i cleaned it all as good as i could with neat isopropanol and about 30 cotton buds but couldnt get the pcb to go matte looking again

Fast forward to two days ago whilst he was playing hearthstone (which doesnt tax the gpu in the slightest)
he lost graphics output ingame and with the pc still powered on it was beeping no graphics card

I diagnosed it as a dead graphics card and put a spare MSI R9 280 in there for him which worked fine

The RX-480 Nitro+ is still an ok graphics card and wasnt happy with just calling it a loss

I took the card apart to just the pcb
there are 3 x 10a fuses to protect the card from shorts
1 x 10a for power from the 16x pcie port
2 x 10a for the 8pin

The 2 x 10a for the 8pin where blown
The 1 x 10a for the 16x pcie port was not

I ordered 10 x Littelfuse Fast Acting SMD Fuse 1808 10A from ebay
they look identical to the fuses in there already

Next port of call was to clean the pcb and get all of that oily shine off the back of the pcb

I decided the best way to do that was to use neat fairy washing up liquid and a toothbrush gently on the whole card as its a good degreasant
I then rinsed it under the hot tap
I then cleaned the back with isopropanol and a toothbrush
I then cleaned it again with fairy liquid and a toothbrush
I then thoroughly rinsed it all for a good 10mins under the hot tap ensuring there was no traces of fairy under bga's etc, shook off as much water as i could and left the card gently placed on a hot radiator for 14hours to fully dry ensuring no cap/resistors where damaged (rotating it every few hours)

I probed the mosfet's vin with a multimeter in resistance mode with ground connected to the io shield
all mosfets would go up from 0 to around 2000 then show 1 within around a second indicating no connection apart from the bottom one which would
go up from 0 to around 1500 and stay there

On looking at the rear of the pcb one of the mlcc's has clearly blown and is shorting to ground (C4132)
not knowing its spec the best thing for me to do is remove it and check if there is still a short
which i did

once it was removed I probed the mosfet's vin with a multimeter in resistance mode with ground connected to the io shield, all would go up from 0 to around 2000 then show 1 within around a second indicating no connection

so all is looking good

Since i am waiting on 10 x Littelfuse Fast Acting SMD Fuse 1808 10A to come from china
my choice was to either wait a month not knowing if it was fixed
or bridge the fuses i decided to solder 13a fuse wire onto the blown fuses as thats all i have
and a 13a fuse is better than a direct bridge and atleast offers a little protection

I put the card back together
and tested it
its all working fine

but would like to replace the removed MLCC when i go in to replace the fuses in a month or so

does anyone know where i can get a pcb schematic so i can order a replacement

 

[justwondering]

Wow, amazing post, I can offer no help, but oddly enough I bought the same card from ebay a couple of weeks ago, except it is the 8gb version, it powers up / lights up, but the fans do not spin and no display.

I tried the bios switch etc.

I guess I should check the fuses as detailed in your very detailed post.

 

[Shonk]

Wow, amazing post, I can offer no help, but oddly enough I bought the same card from ebay a couple of weeks ago, except it is the 8gb version, it powers up / lights up, but the fans do not spin and no display.

I tried the bios switch etc.

I guess I should check the fuses as detailed in your very detailed post.

The fans were not spinning on mine cant remember if the led's were on

Checking the 3 fuses will help to narrow down the problem
i spent around 10 hours watching translated russian & german vid's on youtube
seeing what commonly goes and hoped for the most common problem
which i was lucky and it was

this guy said (in russian) that its common due to the oil from the bad heat transfer pad (paraphrase)

https://www.youtube.com/watch?v=9iiEbmLcNzg&t

thats why i put a short story not a lot of people will read the whole thing

 

[justwondering]

The fans were not spinning on mine cant remember if the led's were on

Checking the 3 fuses will help to narrow down the problem
i spent around 10 hours watching translated russian & german vid's on youtube
seeing what commonly goes and hoped for the most common problem
which i was lucky and it was

this guy said (in russian) that its common due to the oil from the bad heat transfer pad (paraphrase)

https://www.youtube.com/watch?v=9iiEbmLcNzg&t

thats why i put a short story not a lot of people will read the whole thing

yes very informative, hopefully just a fuse, the ebay seller has refunded me and told me to keep the GPU.. So I shall investigate further.

Just watched this one https://youtu.be/TTwPHoGIYJY

 

[timmy222]

Well done Sir, I thought it was a very good post.

 

[Shonk]

yes very informative, hopefully just a fuse, the ebay seller has refunded me and told me to keep the GPU.. So I shall investigate further.

Just watched this one https://youtu.be/TTwPHoGIYJY

yeah the single pcie 10a fuse went on that also watched that yesterday

its all down to the bad pad leaching oil onto the pcb
see the oily shine (green area)
mine was much much worse than that and covered the whole red area
his was in 2017 so 1year in
mine is 4 years in..

 

[justwondering]

yeah the single pcie 10a fuse went on that also watched that yesterday

its all down to the bad pad leaching oil onto the pcb
see the oily shine (green area)
mine was much much worse than that and covered the whole red area
his was in 2017 so 1year in
mine is 4 years in..

I shall check the one I have, when I am back in the office, hopefully it is fixable, happy to spend a couple of hours trying to fix it, as it was free, I also have a non working gtx 1080, which I keep meaning to attempt to fix too.

 

[Kotu]

Hi Shonk,

I went through the same thing with a Sapphire 390x tri-x, it's a shame to see Sapphire was still cutting the same corners on the next generation of the card.

My card got RMA'ed 4 times for the same issue before the warranty expired. The RMA got refused when it happened a 5th time (cost me £160 in postage having to keep sending the card back to Hungary).

I managed to repair the card myself a few times by replacing the IR3553 chips each time one died, but some of the replacements also died. The 8 pin fuses would die each time this happened too.

Was something else at fault on the card? I'm not sure, it seemed to die when under heavy gaming load whilst also running hardware accelerated videos at the same time.

If you look at the IR3553 chips you will see they only occupy about 66% of the space for the chip (the pads underneath the chip protrude below the chip) and the square around the chip is larger.

The IR3553 chips on the card only have 40A*6 capacity (240A), but you can actually replace them all with the 60A versions (IR3550) which has the effect of both increasing the current capacity to 360A and also makes the chips 33% larger (going from 4mm x 6mm to 6mm x 6mm) meaning they have a larger surface area to keep the VRMs cool.

I bought the IR3550/IR3553 chips from aliexpress.

My 390x has been stable without anymore random deaths since doing the IR3550 upgrade, again it's shameful that Sapphire chose to save about ~£1 per card (considering the circuitboard has space for the 60A chips) leading to these kind of issues.

 

[Shonk]

Hi Shonk,

I went through the same thing with a Sapphire 390x tri-x, it's a shame to see Sapphire was still cutting the same corners on the next generation of the card.

My card got RMA'ed 4 times for the same issue before the warranty expired. The RMA got refused when it happened a 5th time (cost me £160 in postage having to keep sending the card back to Hungary).

I managed to repair the card myself a few times by replacing the IR3553 chips each time one died, but some of the replacements also died. The 8 pin fuses would die each time this happened too.

Was something else at fault on the card? I'm not sure, it seemed to die when under heavy gaming load whilst also running hardware accelerated videos at the same time.

If you look at the IR3553 chips you will see they only occupy about 66% of the space for the chip (the pads underneath the chip protrude below the chip) and the square around the chip is larger.

The IR3553 chips on the card only have 40A*6 capacity (240A), but you can actually replace them all with the 60A versions (IR3550) which has the effect of both increasing the current capacity to 360A and also makes the chips 33% larger (going from 4mm x 6mm to 6mm x 6mm) meaning they have a larger surface area to keep the VRMs cool.

I bought the IR3550/IR3553 chips from aliexpress.

My 390x has been stable without anymore random deaths since doing the IR3550 upgrade, again it's shameful that Sapphire chose to save about ~£1 per card (considering the circuitboard has space for the 60A chips) leading to these kind of issues.

They must have changed something from 290 to 390 then
I have a Sapphire R9 290 Trix OC that i picked up from a miner in 2014 for £150 and ran it from day one bios modded to 1100core 5600ram with the power limit upped to 238w
it was my primary card until late 2018 when i got a Sapphire Vega64 Nitro+

Its still in my bedroom pc running like a champ but doesnt game much anymore

 

[Kotu]

@Shonk
I don't think checking the resistance from VIN to ground will give useful results as a lot of the VRM are connected in parallel through capacitors.

I suspect the issue is identical to the one I had, can you check continuity between VIN and SW on the IR3553s?

If they are shorted then one of the IR3553s has blown. They are all connected in parallel so it might look like they are all shorted, but it will likely be the one connected to the blown capacitor.

This would also mean the reason the cap was damaged is because the 12V input got shorted to the vcore for a short period before the fuses blew (unfortunately this could have also damaged the gpu core, but mine survived this happening a few times).

@justwondering

The fuses rarely blow on their own, something else breaks which causes the fuse to blow. If you just change the fuse and retry the card again then whatever caused the fuse to go in the first place might cause even further damage.

 

[Shonk]

@Shonk
I don't think checking the resistance from VIN to ground will give useful results as a lot of the VRM are connected in parallel through capacitors.

I suspect the issue is identical to the one I had, can you check continuity between VIN and SW on the IR3553s?

If they are shorted then one of the IR3553s has blown. They are all connected in parallel so it might look like they are all shorted, but it will likely be the one connected to the blown capacitor.

This would also mean the reason the cap was damaged is because the 12V input got shorted to the vcore for a short period before the fuses blew (unfortunately this could have also damaged the gpu core, but mine survived this happening a few times).

@justwondering

The fuses rarely blow on their own, something else breaks which causes the fuse to blow. If you just change the fuse and retry the card again then whatever caused the fuse to go in the first place might cause even further damage.

Its working fine once i removed the blown mlcc and sorted the fuses
aparently the reason these mlcc's blow on the back is down to the oil seepage from the pad

 

[Shonk]

Here's screenshots of the MLCC being tested
I measured them and they are around 2mm x 1mm x 1mm

So
SMD 0805 size. L 2mm x W 1.25mm x H 1.25mm

So have ordered 10 x 10uF 16V SMD Ceramic multilayer 0805 MLCC X7R

There are two types that i can see
A Z5U capacitor will operate from +10 °C to +85 °C with a capacitance change of at most +22% to −56%.
A X7R capacitor will operate from −55 °C to +125 °C with a capacitance change of at most ±15%.

with the X7R being the better one

 

[Shonk]

The MLCC's came man they are hard to solder they are so tiny and hard to keep inplace
ended up fluxing the area and using 2 soldering irons at the same time

still waiting on the fuses
card works fine with the MLCC replaced

also modded the bios whilst i was there so it will run cooler

Changed its id to RX 580 as the drivers dont check the bios signature on 580's but do on 480's
weird that a 480 and 580 have a different driver path

Changed Core Clock from 1306 to 1342 to match the 8GB card
Changed board Power from 140w to 150w to match the 8GB card

Changed the fan target from 75°C to 63°C
that thing used to be sitting at 70°+ with the fans at like 800rpm
it goes to around 2000 RPM now ingame

Updated the UEFI Gop