[SOLVED] MSI mobo cpu red light, if gpu power is connected

May 23, 2020
4
0
10
Hello,

I have built this system:
Psu : FSP 700-60AHBC
Mobo : MSI B450 Gamin Plus Max
CPU : Ryzen 2600
GPU : MSI GTX1650 Super XS OC
RAM : CORSAIR Vengeance LPX DDR4 1x16GB 3000 Mhz

Firstly, i did not flash the newest bios (used whatever it was) and assembled that way. When i powered on, i could not pass from CPU post. CPU Error Ez Debug Led remained red without blink. Then i dissassembled all parts. I only connected CPU and Mobo connections of PSU(no cpu, ram, gpu etc.) as instructed in the flash bios button section of the motherboard manual. I reset CMOS by extracting and replacing CMOS battery. I connected a usb drive (which i formatted to FAT32 and in which only file was MSI.ROM) to Flash Bios Usb Port of mobo. I pressed the flash bios button. The led starting blinking a few seconds later and remained blinking a few minutes. Then the system reset itself. I did and observed these as the manual instructed. The BIOS version i used is 7B86vH6 on site: https://www.msi.com/Motherboard/support/B450-GAMING-PLUS-MAX#down-bios

Then I assembled again all parts. Now when the gpu power connector(6 pin psu connecter on my GPU) is disconnected, when i press the power button, first cpu red led is lit. 1-2 seconds later it is not lit anymore. A few seconds later VGA led is lit and it stays that way. GPU is not powered on(no led or fan) as expected, i think. From this, i understand that flashing made an effect, because i could never pass CPU red light error before.

The problem is when everything is assembled + gpu 6 pin power connecter is also connected. When i press the power button, the CPU red light is lit and stays that way without blinking. I think, GPU is not powered on again since no led or fan. The GPU was firstly in PCIE 3.0 x16 port. I also tried the other PCIE 2.0 x4 port (other plastic long PCIE connector below) and the results are same.

Okay, I thought if it was about GPU, why mobo does not lit VGA red light? The only way i could think was the PSU is faulty, so when GPU is also demanding power, the PSU cannot supply (by the way i believe 700 watt is way more than enough for this system), voltage levels drop and it cannot even supply CPU properly, hence the CPU red light. In order to test this. I measured voltage levels with voltmeter of GPU and CPU power inputs, when all mobo, cpu, and gpu power inputs are connected; and system is powered on. The measured voltage was 12.1 +- 1V (my voltmeter is not calibrated). I supposed the problem is not about PSU. (by the way the PSU came with a coil whine problem)

In all tests, i never modified the system when power was on.

In short:
The problem is told by mobo. Mobo says "problem is about cpu". Cpu says "i am fine if you do not power GPU". Problems happen only when gpu power is connected. Power supply says "it is not my problem". The bad thing is that i ordered parts separately. So if a part is faulty, i have to debug and find it quickly.

The last thing i can do is to continue flashing all 6 previous bios versions published on MSI website, to check if the system can be happy with one of them.

Lastly. Very seldomly even when GPU power is not connected something happens about CPU red light. It starts lit and not lit (as always so far, but after this point it should pass to VGA red light); but then it continues to lit and not lit again and again.

I tried to give as much detail as possible in order not to miss anything. Thank you if you could read.. :)
 
Solution
Okay, i found the reason of this problem. It is about PSU. But it is not that it can't supply demanded power. It is the Power good control signal(also called PWR_OK) on the ATX mobo connector. As I understood, when PSU is demanded power by mobo (through PRW_ON line, when front panel button is pressed), the PSU first checks itself if its output lines are well regulated and voltage level tolerances are lower than some threshold values. If the result of this check is positive, then it asserts 5V at the Power good line on ATX connector. This way the mobo will know that it is safe to continue system operation and will boot. If the psu self test fails, then it cannot assert 5V (i.e. 0V at voltmeter). This way the mobo will know that, it is...
May 23, 2020
4
0
10
I also noticed this:

If i power the gpu, cpu and motherboard, and connect cpu fan to mobo, but don't connect the case fan (there is only one) to mobo; then again system boot hangs on cpu red light. However, this time i also disconnected CPU fan. This way i could enter bios screen. This is always consistent. Therefore, i suspected again, if PSU also cannot supply power as declared(700 Watts). However, my system demand is calculated definitely not more than 400W with online calculators. I don't know if it is possible that a 700W declared FSP PSU will not be able to supply power as low as 400W . Also, i know that fans demand power as low as 5-20W. I don't know if PSU is at the edge of its degraded limits because of a faulty condition; and it is like one-zero situation at this edge. Moreover, If a PSU cannot supply power, I expect to see voltage drop at its 12V rail. However when i measure with multimeter, the measured voltage is ~12V, when all cables connected and it cannot post. I think i should at least see 11.5V or lower. Am i wrong? I need some experienced help please?

Again the thing is that i have to find the problematic one with %100 certainty as soon as possible, in order to demand payback for that one. The PSU will go for sure, even if its only problem is coil whine. The obvious thing to do is to debug the system with a power supply from a friend. But that won't happen before 4 days. Also i want to know if my system is drawing more power than 400W (if there is another problem in my system). Because; 1. I might need to replace another unit. 2. I have to know my system's exact demand to choose the output power of my next psu.
 
Last edited:
May 23, 2020
4
0
10
Okay, i found the reason of this problem. It is about PSU. But it is not that it can't supply demanded power. It is the Power good control signal(also called PWR_OK) on the ATX mobo connector. As I understood, when PSU is demanded power by mobo (through PRW_ON line, when front panel button is pressed), the PSU first checks itself if its output lines are well regulated and voltage level tolerances are lower than some threshold values. If the result of this check is positive, then it asserts 5V at the Power good line on ATX connector. This way the mobo will know that it is safe to continue system operation and will boot. If the psu self test fails, then it cannot assert 5V (i.e. 0V at voltmeter). This way the mobo will know that, it is not safe to continue system operation and will not boot.

The mobo i have does not have a seperate PSU debug led for this purpose (i don't know if other PSUs have it). Therefore such a problem is connected to CPU debug led. This is what is happening in my system. I verified it with voltmeter. If only cpu, gpu and ATX power connectors are connected, the power good line is measured 5V. However if the extra cpu fan is connected to mobo, power good line is 0V. Somehow that extra fan disturbs the regulation characteristics of the psu.
 
Solution