Are my VRM's ok?

[Alexsander]

Good morning y'all. I recently build a "old school/vintage" hardware (Managed to buy I brand new on the box) 770iCafe motherboard.

The thing is that, I don't know if everything is ok or if I managed to do some damage to it, I just wanna now if it still Ok, if I just reduced it's life span, I don't see anything stuffed or burned.. I've been stress testing this thing for almost 2 weeks with the following specs:

MB: AsRock 770iCafe
CPU: FX-8350
RAM: Kingston ValueRAM 16GB (4x4 1600mhz)
PSU: Mancer Thunder 500w 80plus
SSD: Kingston 240GB

Stress testing it with: Aida64, OCCT, CineBench and Prime95 (I ran every single one for at least 3 to 4 hours with no BSOD)

I didn't realize how much the VRM's were burning until I got my first BSOD playing League of legends. It started to happen pretty frequently after that (always while playing something)

I decrease both the voltage and the CPU clock to the following values and everything is running ok (I stress tested it with OCCT for 30 minutes while touching the VRM'S to see if they were too hot)

CL - 3.5ghz (1650V)
NB - 2200 (1500V)
HT - 2000

Everything OK.

I now know that the VRM's in that Mobo needs a cooling system specifically to the VRM'S, so I bought two micro fans (40x40x10 - 12v) and I'm gonna adapt them on the VRM'S. I just want a rest of mind from you all saying that it's not damaged (Don't wanna spend hours testing it after the adaptation just to discover that I managed to burn a VRM).

Thanks c:

(Please don't tell me that I should change the MB, I know that a 970 and further would be better, I build this config just because I like old Hardware and because that's the only option I have RN, I'll jump to Ryzen in the future)

 

[Flamethrower1972]

You could buy these.

Are you a human?

www.newegg.com

 

[Alexsander]

You could buy these.

Are you a human?

www.newegg.com

Yeah, that would be great as well, but I think the tiny fans will cool it better. They will arrive today and I'm going to adapt them there. Just wanna know if I damaged something or nah (The signs of a damaged VRM and so on). Thanks for the reply, I'll probably add one of these as well later.

 

[Flamethrower1972]

Yeah, that would be great as well, but I think the tiny fans will cool it better. They will arrive today and I'm going to adapt them there. Just wanna know if I damaged something or nah (The signs of a damaged VRM and so on). Thanks for the reply, I'll probably add one of these as well later.

No problem I have used them before on older PC'S and they worked well.

 

[EarthDog]

I wouldnt overclock or stress that system out... scary psu and not great vrms. Cooling is one thing... if they can't handle it... they can't.

Those tiny *** copper heatsinks won't help much over what you have. Active cooling or bigger than thise little things.

 

[Alexsander]

I wouldnt overclock or stress that system out... scary psu and not great vrms. Cooling is one thing... if they can't handle it... they can't.

Those tiny *** copper heatsinks won't help much over what you have. Active cooling or bigger than thise little things.

THE PROBLEMS IS FIXED GUYS C:

Please rate this thing 1 to 10 (I used a screw to fix them on the heatsink). I'm gonna use a 3D printed model in te future just to make it fancy.

But that's the resolution, no more heat. Matter fact, is it ok to use these screws on the heatsink? I think there's no conductivity between them since the heatsink is made of aluminum so... It's gonna be ok right?

 

[Flamethrower1972]

THE PROBLEMS IS FIXED GUYS C:

Please rate this thing 1 to 10 (I used a screw to fix them on the heatsink). I'm gonna use a 3D printed model in te future just to make it fancy.

But that's the resolution, no more heat. Matter fact, is it ok to use these screws on the heatsink? I think there's no conductivity between them since the heatsink is made of aluminum so... It's gonna be ok right?

I bet that is going to work great. A 10.

 

[Alexsander]

I bet that is going to work great. A 10.

Thank you my dude (I don't know if u are being ironic ; -; I'm not American) gonna be using this 3D printed model for these tiny fans in the Future. There's a dude here in BR that makes them. Thank y'all for the help, I'm gonna see if this fix the problem no more BSOD's hopefully.

 

[Flamethrower1972]

Thank you my dude (I don't know if u are being ironic ; -; I'm not American) gonna be using this 3D printed model for these tiny fans in the Future. There's a dude here in BR that makes them. Thank y'all for the help, I'm gonna see if this fix the problem no more BSOD's hopefully.

No bro you did a great job.

 

[Alexsander]

No bro you did a great job.

Thank you man have a great weekend y'all

 

[EarthDog]

Thank you my dude (I don't know if u are being ironic ; -; I'm not American) gonna be using this 3D printed model for these tiny fans in the Future. There's a dude here in BR that makes them. Thank y'all for the help, I'm gonna see if this fix the problem no more BSOD's hopefully.

I hope it works out for you...keep us posted.

 

[Flamethrower1972]

Thank you man have a great weekend y'all

You too bro.

 

[Alexsander]

I hope it works out for you...keep us posted.

I can see the difference in the temperatures but something is off... I was experiencing this problem and now it's more active.

I can stress test for hours, not a single problem. I did OCCT, AIDA64 for two hours each, not a single problem. But every time I run "sfc /scannow" or every time I do a big extraction such as the .zip archive from CineBench, the pc freezes at 95%. I cannot use the mouse or the keyboard, both of them are on (with their lights on) but irresponsible.

I already tested each stick of RAM and a full slot on Memtest86 2x pass. I did the Windows Memory Diagnostic tool and one thing to note is that the "standard" test, I can pass, but the extended one always freezes at 3%.

I don't think it's a problem on the VRM's because the CPU can be stress tested fine (although it could actually be, because when doing normal stuff the cpu isn't using that much energy)



But now I don't know if I Have a damaged CPU, a damaged VRM's circuit OR a damaged SSD (just to take note, but the speed of an extraction is usually really slow that could be either the CPU or the SSD itself, crystal disk says it's good.

Any guess?

EDIT:

This MB have 4 memory modules maximum of 16GB. If I use 8GB (4x2) on A1 and A2 It can extract the file without freezing. If I use it in B1 and B2 it can extract the file without freezing, but if I use all of them it can't, it freezes °\{• >•/° What's the explanation for that? Memory trimmings are set correctly for this module (Kingston KVR16N11/4). They are all the same. Just one thing is different, 2 RAMS the chips are Elpida made and the other 2's are Kingston made (Both of them are from the same Manufacturer at the end, just the company that provides some chips to Kingston)

EDIT2

Yeah, apparently the problem is Always happening when all 4 sticks of RAM are in use. Just to take notes, I already raised the CPU Vcore to fix it and nothing. Set the memory voltage to manual (1.5v) and nothing.

Memory trimmings are in Auto but one thing concerns me... The CL thinking is at 9 (it works fine with 2x4). The manufacturer says it works in a range of 9-11, could it possibly be that this CL time doesn't work for 4x RAM? I'm gonna reset the CMOS and put all of them back to see if something changes

EDIT3

dude, how long should I wait after removing the CMOS battery? I waited for 5 minutes and it didn't reset (PSU wasn't connected and I pressed power button several times). I changed the times manually, I'm doing the extraction now, gonna see if it works. If it don't I'll reset the CMOS and wait at least 30 minutes.

EDIT4

I'm perplex, it worked °\{• >•/° I just changed the CL timing that was at 9, to 11 and now it's extracting. I'm gonna reset the CMOS just to make sure it was the timing and run it again to confirm.

 

[Alexsander]

I can see the difference in the temperatures but something is off... I was experiencing this problem and now it's more active.

I can stress test for hours, not a single problem. I did OCCT, AIDA64 for two hours each, not a single problem. But every time I run "sfc /scannow" or every time I do a big extraction such as the .zip archive from CineBench, the pc freezes at 95%. I cannot use the mouse or the keyboard, both of them are on (with their lights on) but irresponsible.

I already tested each stick of RAM and a full slot on Memtest86 2x pass. I did the Windows Memory Diagnostic tool and one thing to note is that the "standard" test, I can pass, but the extended one always freezes at 3%.

I don't think it's a problem on the VRM's because the CPU can be stress tested fine (although it could actually be, because when doing normal stuff the cpu isn't using that much energy)



But now I don't know if I Have a damaged CPU, a damaged VRM's circuit OR a damaged SSD (just to take note, but the speed of an extraction is usually really slow that could be either the CPU or the SSD itself, crystal disk says it's good.

Any guess?

EDIT:

This MB have 4 memory modules maximum of 16GB. If I use 8GB (4x2) on A1 and A2 It can extract the file without freezing. If I use it in B1 and B2 it can extract the file without freezing, but if I use all of them it can't, it freezes °\{• >•/° What's the explanation for that? Memory trimmings are set correctly for this module (Kingston KVR16N11/4). They are all the same. Just one thing is different, 2 RAMS the chips are Elpida made and the other 2's are Kingston made (Both of them are from the same Manufacturer at the end, just the company that provides some chips to Kingston)

EDIT2

Yeah, apparently the problem is Always happening when all 4 sticks of RAM are in use. Just to take notes, I already raised the CPU Vcore to fix it and nothing. Set the memory voltage to manual (1.5v) and nothing.

Memory trimmings are in Auto but one thing concerns me... The CL thinking is at 9 (it works fine with 2x4). The manufacturer says it works in a range of 9-11, could it possibly be that this CL time doesn't work for 4x RAM? I'm gonna reset the CMOS and put all of them back to see if something changes

EDIT3

dude, how long should I wait after removing the CMOS battery? I waited for 5 minutes and it didn't reset (PSU wasn't connected and I pressed power button several times). I changed the times manually, I'm doing the extraction now, gonna see if it works. If it don't I'll reset the CMOS and wait at least 30 minutes.

EDIT4

I'm perplex, it worked °\{• >•/° I just changed the CL timing that was at 9, to 11 and now it's extracting. I'm gonna reset the CMOS just to make sure it was the timing and run it again to confirm.

Update.

It was extracting but failing OCCT CPU+RAM test single core test (5 minutes on each core) It failed with 06 seconds on CPU#0 5 errors. I decreased the MHZ of the memory from 1600mhz to 1333mhz. The test went from 06 seconds to 6 minutes with CPU#0 1 Error.

So returned the memory to its base clack 1600mhz but bumped a little bit o voltage, and it went from 1.5 to 1.55V.

Now it failed at 40min whit the same CPU#0 error

One thing that I really really would like to share is one information in the manual of that MB that states the follow information:

"If you adopt DDR3 1600 memory modules on this Motherboard, it is recommended to install them on DDR3_A2 and DDR3_B2 slots"

I was aware that some Ryzen CPU does not work well with 4x RAM and that's related to the CPU being used at its MAX capacity (I don't know) If someone could explain that to me, I would love to search more about it. I think if I bump the voltage to 1.6v it's gonna fix the problem, but I'm afraid in damaging it :/. I'm gonna test it with 1333mhz at 1.55v (Or mess up with the multiplier decreasing it to like... 10x? And increasing the FSB until the memory gets to 1600, will it work? Don't know). Either waiy, the FX8350 is working at stock GHZ with a 1.288V, does increasing Vcore will have any impact over the memory? Guess I'm gonna rise it to 1.3 and test it out with 1600mhz at 1.55

 

[EarthDog]

What's the explanation for that?

Memory isnt stable.

High level - more capacity/sticks puts additional stress on the IMC (integrated memory controller). You're doing right in adding voltage and lowering speeds to achieve stability... but it's a process.

 

[Alexsander]

Memory isnt stable.

High level - more capacity/sticks puts additional stress on the IMC (integrated memory controller). You're doing right in adding voltage and lowering speeds to achieve stability... but it's a process.

Yeah, now it's stable with 1.55v 11-11-11 at 1333mhz.

I'm gonna try to run it again but now with 1.300 Vcore CPU. 4ghz and 1.165 NB 2200mhz

That's really interesting I don't remember having this issues with a Phenom II X4 945 OC to 3.5 or a Core 2 Quad 945.

Guess that at the end I was unable to notice the problem of RAM instability because I was with high VRM's Temp (that was fixed due to the fan adaptation) and it was crashing the system as well. Now that temps are fixed, I was able to see the RAM instability.

OCCT completed an hour with no erros c: now I guess that's almost the END of that troubled history.

I was unable to find any info on the internet about this config or more detailed information about this MB. Think that's gonna help if anyone for some reason have the same issue/config.

EDIT1

Yeah, it worked raising the NB voltage from 1.15 to 1.16v. Could you explain to me why does raising the RAM voltage helped to maintain it stable as well? If the problem was the IMC, why raising the RAM voltage helped with it? Wouldn't raising only the Vcore/NB voltage be enough?Why's it like that ; -; I'm confused.

EDIT2

At 1.185NB Voltage and 1.5V RAM 1600mhz it freezed in 53min. The only stable one is at 1333mhz 1.55V :/ Guess I'll only be able to get all 4 channels to 1600mhz if I raise the RAM voltage to 1.6 °\{• >•/° That's maybe enough to fry them, but I'm not sure. Gonna raise the Vcore as well further to see if something changes. But I think that's it.


EDIT3

Yep, with 1.315 Vcore

NB - 1.12

And RAM at 1.55 I wasn't able to maintain 1600mhz.


I'll buy a thermometer to see how hot does it get at 1.6 and gets it out.

Gues that's it. The most stable apparently that I could get with 4x RAM on this MB was 1333mhz °\{• >•/° Kinda sad, but I'll definitely try to raise it in the future. I'm gonna let Aida64 running for at least 4h just to make sure everything is ok RN, and see what more I can check to state that's 100% stable.

~ -~ Dude... I really wanna put that thing at 1600mhz, is it too much a bump of only 0.1? I've seen a dude OC this thing to 2500mhz stable with 2v ; -; I know that they can't pass 85C and there's a "common sense" to not pass 1.65. don't know if I wait until I get a thermometer or just bump it up. It's a raise of 0.05 only.

EDIT4

It's not stuck at 3% on Windows Extended memory diagnostic, it's running fine RN °\{• >•/° I guess that's it. Problem solved, but I'll definitely see if I can run all these 4 sticks at were they were supposed to be 1600mhz. Just gonna gather more information about this MB and buy a thermometer, if I can safely give it a bump (at least on its specs its says that 1.575 is fine, guess 1.6 won't be a problem).


EDIT5

Nope, there must be some limitations on that MB, I'm really unable to use 1600mhz on all 4 slots.

I tried increasing the NBv to 1.2750 and RAM to 1.55V and could run OCCT for 23 minutes. Increasing the voltage to 1.6 just made it worse, running only form 3/6 minutes. For some reason at 1.2750 I cannot Increase it's clock to 2400mhz, SO start to stutter and my monitor keeps turning off (as if I disconnected and reconnected my GPU). Maybe pushing it forward I could make some progress but I just don't think it's worth it anymore. For everyone that have the Same MB, u can easily go for 1600mhz using 2x8 sticks on the A2 and B2 DIMM's. But the same config using 4x4 °\{• >•/° U just can't get it stable at all, at least I couldn't. (I'n bot saying it's impossible, but maybe it's not worth it, 1600 to 1333 is not a problem at all, you could just OC the CPU normally instead of trying it)

Stable config BIOS using 4x4 1600mhz at 1333mhz

Spread Spectrum - Auto
Turbo Core - Disabled
C-State - auto
Cool 'n quiet - Disabled
Virtualization - Disabled
CPU Multiplier - x21(4.1GHZ) (I'm gonna OC it further, but RN I don't need more than that)
Vcore - 1.285 up to 1.312
CPU NB - x20 (2200)
Voltage - 1.175 up to 1.187
Memory Clock - 1333mhz
Voltage - 1.55
Timing - 9-9-9-24 (That was the most I could get without losing patience testing and that's what the Auto MB config sets automatically, but u can use like... 7-8-8 or 8-8-8 if u have the patience and the luckyness

Hope this helps someone out there, I'll keep testing it out. If anyone out there need a help with that config or something related, or even if already tried 4x4 on that MB I'm all hands and ears.