cancel
Showing results for 
Search instead for 
Did you mean: 
Affende
New member
5 4 0 0
Message 1 of 13
756
Flag Post

2nd CPU riser power input plug melted

HP Recommended
Z620
Microsoft Windows 7 (64-bit)

I bought the Z620 used, about a year ago, for CAD work and folding (F@H).  After folding for approximately 6 months on the CPUs, the system locked up, black screen, fans screaming (i assumed a thermal overload protection mode or similar).

 

It took me a few days to figure out the issue, but the second CPU riser seems to be the culprit ... specifically the black power input plug.  

 

The system has been running fine for the last 6-8 motnhs without he 2nd CPU riser card installed.

 

The input plug, from both sides (cable harness side and CPU riser side) are both melted.  I would like to get this problem resolved but i dont want to fry another input plug / CPU riser board.  

 

I am unsure if the only problem with the riser card is the input power plug or if something in the card has caused the plug to melt.

 

Without having another sytem to plug the card into, how can I start tracking down the root cause of this problem?  

12 REPLIES 12
DGroves
Level 11
4,491 4,474 408 937
Message 2 of 13
Flag Post
HP Recommended

since this system is out of warranty, do you want to risk killing the motherboard by reinstalling this known defective part?

 

the power plug  and wiring harness could have melted due to poor mating between the  connections, a loose connection will cause increased heat, which causes more resistance which causes more heat and so on

 

your programs running placed max load on the power delivery systems and as noted above, a poor connection became a 

failed connection over time

 

these connectors are not made for constant connect/disconnects, as they tend to loosen up over each connect/disconnect

and as noted above, a loose connection can cause problems

 

buy a new riser and new cable harness, and make sure they are fully seated and you should be good to go

0 Kudos
Affende
Author
New member
5 4 0 0
Message 3 of 13
Flag Post
HP Recommended

Thanks for your repsonse.   I did not plan on reinstalling the defective board.  My plan is to replace the 2nd CPU riser board (and max out the DIMMs while im at it) before i upgrade the graphics from a Quadro 400 to Dual Quadro k5000 (certified SLI compatibility).

 

My question lies in what can i do, before installing a new riser board and power harness, to ensure the problem will not occure again.

 

There is a good possibility a faulty connection was at risk, but i would  like to test (somehow) to make sure the same probelm doesnt arise again and to make sure there isnt another underlying problem (over voltage, for  example)

 

From my research, there is a history of the 2nd CPU riser having power problems. 

 

The CPU riser cards can be found, but a power harness isnt showing up;  where can these be sourced?

0 Kudos
DGroves
Level 11
4,491 4,474 408 937
Message 4 of 13
Flag Post
HP Recommended

if you are so inclined get a Infrared digital thermal tempture probe, the type where you simply point it at something and it returns a temp reading then use it on the computers power connections and surounding area

 

https://www.amazon.com/Infrared-Thermometer-Contact-Temperature-Measurement/dp/B002OD0NCG

0 Kudos
BambiBoomZ
Level 7
713 685 45 192
Message 5 of 13
Flag Post
HP Recommended

Affende,

 

Here are some guesses and speculations:

 

This is a serious event and in my view the risk of damaging the riser, CPU, RAM. and etc. means that the solution should include measurement of the current draw at the power input. Those components are made of high quality materials that can operate at temperatures well above those of normal operation, so melting is extreme.  Also, the power supply should be thoroughly tested for conformity to specification. There is nonvolatile memory on the riser and a severe power surge may affect it- worth reviewing.  Fan screaming may be a symptom of firmware problems.

 

I may be mistaken, but I thought that the riser board power arrived throught the principle socket and the additional onboard connections were for the CPU and memory fans. Is the riser CPU fan rotating freely? A locked up motor can create an extreme current draw.

 

It would be preferable to spending time with a DVM or something on a technician's time than having to replace components.

 

Let us know what you find out.

 

BambiBoomZ

 

 

0 Kudos
Affende
Author
New member
5 4 0 0
Message 6 of 13
Flag Post
HP Recommended
Thanks, DGroves, this is a good start for sure.  Luckily, being in the manufacturing repair business i have a few of these lying around.  Ill take some measurements and post the numbers so others can compare in the future if needed.



@DGroves wrote:

if you are so inclined get a Infrared digital thermal tempture probe, the type where you simply point it at something and it returns a temp reading then use it on the computers power connections and surounding area

 

https://www.amazon.com/Infrared-Thermometer-Contact-Temperature-Measurement/dp/B002OD0NCG


 

0 Kudos
Affende
Author
New member
5 4 0 0
Message 7 of 13
Flag Post
HP Recommended

@BambiBoomZ wrote:

Affende,

 

Here are some guesses and speculations:

 

This is a serious event and in my view the risk of damaging the riser, CPU, RAM. and etc. means that the solution should include measurement of the current draw at the power input. Those components are made of high quality materials that can operate at temperatures well above those of normal operation, so melting is extreme.  Also, the power supply should be thoroughly tested for conformity to specification. There is nonvolatile memory on the riser and a severe power surge may affect it- worth reviewing.  Fan screaming may be a symptom of firmware problems.

 

I may be mistaken, but I thought that the riser board power arrived throught the principle socket and the additional onboard connections were for the CPU and memory fans. Is the riser CPU fan rotating freely? A locked up motor can create an extreme current draw.

 

It would be preferable to spending time with a DVM or something on a technician's time than having to replace components.

 

Let us know what you find out.

 

BambiBoomZ

 

 


BB,

 

Thanks for chiming in.  The final decision to purchase a Z620  with (2) E5-2590s was actually influenced heavily by your reccomendation (on another board) late 2016 early 2017 in response my query.  Other than the 2nd CPU riser problem, I have been very happy with the system .... so a belated THANKS! to you.   .... upgrades are still planned and coming to increase performace in the gfx and stroage performance arenas. 

 

I do agree that the current situation is enough for me to be very weary of attempting a part swap.  I hope that the issue with with the (now toast) CPU riser.  That would be a very straight forward and simple fix.  Poking and testing with a DVM i believe will be the route i take.  I do have a few professional CPU repair places in town that could takle more extensive testing of the PSU, which is a good idea.

 

I hope that DGroves is correct is speculating that a loose connection, and the resulting heat, is to blame .... but I do prefer the test method over the guess method.

 

Do you know where i could find a pinout for the 2nd CPU riser and specs on the PSU output?

 

From my initial tracing and edumicated guestimating ... it seems to my eyes that the 2nd CPU riser receives main power from  the black (now somewhat melted) 12 pin plug closest to the contral panel of the tower. 

 

With the 2nd CPU riser installed (only tried once, months ago, to narrowdownt he possible problems) the machine goes into full fan power and no boot / no BIOS ...nada .... just fans screaming away with lights flashing.  I did at one point track down the flashing power indicator code ... and have since forgotten which code i was getting.  

 

After removing the 2nd CPU riser, the machine works flawlessly .... im posting on it now.  

 

 

0 Kudos
BambiBoomZ
Level 7
713 685 45 192
Message 8 of 13
Flag Post
HP Recommended

Affende,

 

Yes, I was very fond of z620_1 having the two E5-2690's.  While it was fantastic for CPU rendering, having 32 Threads at an all-core speed of 3.4GHz, eventually it was z620_1, that taught me the greater importance of the single thread performance in combination with the clock cycles per unit time rate for anything 3D. In my view it's better to have 16 threads with a Passmark CPU rating of 17100 and Single Thread Mark of 2354 than  32 threads with a CPU of 17100 and a STM = 1950. Besides the best mutli-threading efficiency occurs at 5-6 cores for all but a few programs, and the memory is more efficent over fewer modules, and there is slight loss in the parity check for multiple processors, for rendering, the clock cycles/time is the same and importantly, the better single thread perofrmance moves those polygons. I can see in Passmark results that the system rating is weighted towards the single thread performance as all other rates being equal, ther system with a higher single thread rating will be experienced as faster in games- and in 3D CAD. Anyway, with rendering, one sets the rendering in action and walks away, whereas waiting for a regen in a big 3D drawing means the real time performance is more critical.

 

As for your system, I've been looking for the pinout diagrams I'd seen about a year ago- this was quite comprehensive as it showed the pinout of every connector.  That was at the time I was adding liquid cooling to z620_2 and was concerned about the fan control /PWM. When I find those diagrams, I'll let you know. Our forum friend Brian1965 I think put me onto those as he designed a special fan controller for his high-rise cooling tower. He runs a  E5-1680 v2 at 4.8ghz - in the league of a top- category gaming i7.

 

PSU: When I had my corrupted BIOS episode with z620_2, I bought a replacement PSU ($70) just in case. As it happens, I'm reasonably certain that the PSU- which was only 14 Months old was not at fault or damaged, but I would not replace the original without testing it. Those are very high quality power supplies.

 

Diversion: Not to divert attention from the issue at hand,  but as, at a minum, a new riser is required and possibly more, consider the advanages of changing to a 2013 bootblock motherboard and running a single E5-1660 v2 or E5-1680 v2, with a slight overclock- to 4.1Ghz on all cores using Intel Extreme Tuning Utility (thanks MtothaJ!).  The cost of the V2 motherboard is the similar to replacemnt riser board, an E5-1680 v2 is perhaps $100 more than than the value of a pair of E5-2690's, so a change is not terribly expensive.  A new motherboard could be appraoched extremely carefully in term of upgrading- a fresh start.  The current motherboard as it is functioning properly ahould have some cost offset- if it is isn't damaged in some way. Yes, the E5-1680 's 16 threads as compared to 32, but a potential CPU mark of17000+ is not a million miles from the 2X E5-2690 average of 20005 (mine was 22035) and the comparitive average single thread ratings of 2193 as against 1831 mean that in every application except CPU rendering or server use the E5-1680 v2 will work more quickly. It depends on the proportional uses. If your system is moving huge blocks of data, or a dozen VM's the core count may be the point of focus.

 

First, the best thing is to identify the fault and study the PSU + current motherboard to avoid future troubles.

 

BambiBoomZ

 

 

 

 

 

0 Kudos
Affende
Author
New member
5 4 0 0
Message 9 of 13
Flag Post
HP Recommended

Thanks BB,

 

I am considering an upgrade on the mainboard front for the V2 processors, but it is below the need to upgrade graphics and storage, first.

 

  PSU side power plugPSU side power plugCPU riser plugCPU riser plugBack of CPU riser plugBack of CPU riser plugTop of CPU riser board where power plug connectsTop of CPU riser board where power plug connects

0 Kudos
BambiBoomZ
Level 7
713 685 45 192
Message 10 of 13
Flag Post
HP Recommended

Affende,

 

Thanks for the views of horror!

 

Can you find any distressed capacitors on the riser board?

 

I'm still looking for the pinout diagrams and PSU spec. 

 

As the system appears to working properly, the PSU is less likely the problem.

 

Die Zukunft is jetzt> If I might suggest, although it involves extra expense for shipping, at the time of changing to V2 processors, consider buying a working V2 system of low specification that includes the riser board. There is some reason for a loose motherboard and/or riser to be available.

 

If I might ask, what is the current specification of your z620 and what are the important applications?

 

BambiBoomZ

 

 

0 Kudos
Warning Be alert for scammers posting fake support phone numbers and/or email addresses on the community. If you think you have received a fake HP Support message, please report it to us by clicking on "Flag Post".
† The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation