Z620 cannot boot anything more than a E5-2620 CPU. PSU issue?

[Wodz]

We have around 50 Z620 systems in our firm and once in a while, a system will suddenly crap out. We have a system that was running 2xE5-2650s /w 64GB. One day the system refuses to work anymore. On boot, the fans just spin up. No red LED.

So we remove the riser and the system boots without issue. So the problem is obviously with the riser. Let's check the CPU. Swapped CPU1 from the riser to the mainboard, boots without issue. Replaced the mainboard RAM with that from the riser. Boots without issue. So the RAM and CPU are verified as working correctly. But wait..

(Riser removed), we tossed a working 2690 v2 3ghz into CPU0 and the system has the same issue as with the riser. Fans spin up, blue LED for power. Never any beeps or issues. Cleared CMOS using the button on the mainboard. No change.

Absolutely appears to be a power issue. With the 2650 installed there is no issue with boot. All fans spin without issue. With any other setup, the fan on the GTX 1060 spins stops, spins, is very confused as to what it should be trying to do. This makes me think there is a voltage flux happening from the PSU. We have a spare PSU and spare Mobo (likely v1) in inventory that we can try. Before doing that I wanted to ping the community on this because this is clearly a common issue. I do not want to bother swapping out the PSU if the issue is actually faulty molex connections.

Although the CMOS clear did not fix the issue, it certainly happened. Once we put the 2650 back into the system and booted, we received the CMOS complaint that settings were reset.

In every configuration, there is a blue LED on the mainboard. Even when the system boots correctly, without issue, using the CPO0 2620, the LED is blue.

We also successfully flashed bios from 3.82 to 3.96 to no avail.

This is the third system in three years this has happened to. Each time we just remove the riser and let the user operate in "limp mode" with a single CPU and half their RAM. In one case, we simply replaced the entire system. In the second incident, we found that the LGA socket on CPU0 was not remotely tight enough, causing instant overheating issues.

System is currently operating without issue running the 2650.

Our offices are currently closed due to the endless quarantine period, so going into the office to bring the spare PSU and or mainboard back to our workshop is the last resort. 

Appreciate any and all assistance on this! Thanks

[savvy2]

replace the RTC coin cell it is due, below 2.9vdc is the cliff of death on this

this is always step any PC over 5 years old. never skip this or the BIOS goes nuts.

if the fans act odd, id think BIOS is nuts. first. and battery next is  1nano sec.

I am faster than the speed of dark..

[DGroves]

hp z620 have easily replaceable power supplies i recommend you do a "BIST" test and even if the supply passes the built in self test swap it out with another

the z620 power supply is a "multirail" design not a common rail one, so it's possible that a single rail in the supply is borderline in it's current output/voltage regulation

[Brian1965]

What is the boot block date of your current motherboard as listed in the BIOS?

[Wodz]

Savvy2 - Thanks for the input but this cannot be the cause. The CMOS battery would not stop a 2690v2 from functioning plus none of the other batteries have been changes in years, yet 49 other systems are working without issue. Did replace the battery to no avail.

Savvy2 -

[Wodz]

Ended up swapping to a different (rev1) mainboard and PSU, same issues. So now we are running a different PSU and the problem persists. Agree that it still seems like a power issue. System is currently working fine with a 2650. 2690 or the riser with another 2650 cause fans to just spin. 

Ran the BIST on BOTH PSUs and got a green light. When the system is running fine the green light inside the PSU is also on. Checking to see if the green light is on with a different CPU in there.

We tested 5 CPUs on this board:

2650 from mainboard - boots without issue, PSU is green

2650 (CPU1) from riser - boots without issue, PSU is green

2650 (CPU1) from different riser - boots without issue, PSU is green

2680 pulled from different system - does not boot, fans spin up, PSU is green

2690 pulled from different system - does not boot, fans spin up, PSU is green

Any 2650 as CPU0 and then we add a riser with any other 2650 for CPU1, does not boot, fans spin up

[Wodz]

Original mainboard was running BIOS 3.82 and we had upgraded to 3.96. Put in a different mainboard, also rev1, this one is running 3.94. Boot Block Date is 12/28/2011.  Thanks

[Brian1965 · Accepted solution]

A 2011 Boot Block date indicates you have a V1 motherboard which will NOT run any v2 CPU's.  A 2013 Boot Block date indicates a v2 motherboard.

[Wodz]

Confirmed that the 2650s in question are V1 CPUs. CPU-Z also notes that the board is Sandy Bridge. So Brian1965 is 100% correct that this Rev1 board does not support V2 CPUs. Thanks a lot for that! Explains why the 2680 and 2690 will not boot.

Does not explain why the riser w/ CPU1 2650 suddenly stopped working. But if replacing the board to Rev2 fixes EVERYTHING..then that is good enough!

[Wodz]

Found another thread that helped us identify which boards are which. Found a listing for a rev2 board on ebay. Having the company purchase that and we will test that board when it arrives. Honestly, if tossing out rev1 boards and putting in rev2 boards will fix riser issue AND allow the board to use CPUs above a 2650 v2 then that is a win!

We just audited the office systems and found that all systems running V2 CPUs were Ivy Chipset and all other systems were Sandy Bridge. More importantly, we found that in all 3 situations where the risers failed, those systems were Sandy Bridge.

Brian's insight is huge as there are a number of systems we were looking to upgrade but would have ran into this same issue (having Rev1 Sandy Bridge boards). Now having a full understanding, we will simply upgrade those boards, then the CPU and RAM. SUPER helpful!

Going forward, if and when a riser fails, it will likely be on a Sandy Bride chipset, we will simply replace the board with Ivy and be all set.

Thanks to everyone that contributed! Stay safe and healthy.