N36L MicroServer died - replacement motherboard?

So my N36L finally gave up the ghost - seems to be the motherboard.
I have tested all other components, PSU is fine, RAM good, HDD good; when the unit is plugged into power the HP logo light seems to cycle-flash very quickly between the blue/red/orange lights - cannot seem to find any detail on this problem so assume the motherboard is knackered.
The unit doesn't power on at all (no fans, POST, nada). PSU was tested on another mobo and powered fine.

Anyways, I'm not prepared to pay US$400+ for a replacement HP board and the unit is out of warranty; so what other boards out there could I use in the MicroServer chassis - I don't mind doing a bit of back panel cosmetic cutting if needed...

Although I've searched, I can't seem to find much info on the original board - I'm assuming from the size that the N36L motherboard factor is Mini-ITX?
Has anyone replaced the standard mobo with a different make, if so which one? Is there a replacement with the SAS (or 6x SATA) ports and up to 8GB memory? I'll be using Win 7 OS. I've looked at various options so far without deciding - Zotac, ECS, Biostar, Gigabyte, etc. all seem to have mini-itx boards but they all seem to have cpu cooling fans which I assume would not fit into the current N36 mobo height space.

Any suggestions welcome, however looking at the lowest cost possible for the replacement mobo. I'm not interested in buying the N54L, but rather using what I have...

It's not very likely you'll get a replacement board for less than the cost of a new server. That's why everyone buys the microservers, they're cheaper than just an HTPC board (or they used to be). Also even if you do find one for cheap, it won't have the connectors the HP one has, especially for storage.

I believe it's a proprietary form factor. The replacement PSU also costs a ridiculous amount

Before trashing mobo, I would try two things:
- Disconnect all connecors, expansion cards, except LED indicators, especially any ATA devices and watch if any changes in behaviour.
- BIOS flash recovery.

If doesn't help, I would like to look at mobo internal voltage regulators.

Also try taking out the CMOS battery and leaving it out for like an hour.

NomNom said:

Also try taking out the CMOS battery and leaving it out for like an hour.

Click to expand...

Oh, yes... It should be done before trying BIOS recovery procedure. Just make sure that power cord is unplugged, it is obvious.

Thanks for the comments everyone. I tried the following so far:

- disconnected everything and tried to power up, nothing; the rapidly flashing lights on the HP logo continue.
- already removed the CMOS battery and shorted the J19 jumper (power cord unplugged) - no change
- tried the J26 BIOS recovery jumper, no change - is there another process to recover the BIOS when the system/mobo doesn't power on?

@ sajunky - what are the mobo internal voltage regulators?

There are usually three power FETs for the CPU power supply on the board. They should be close together and all have the same markings. Just check with a multimeter that they all behave exactly the same when testing across any two identical pins (change the polarity during the tests). Sometimes these can either short or go open circuit.

howardb said:

- tried the J26 BIOS recovery jumper, no change - is there another process to recover the BIOS when the system/mobo doesn't power on?

Click to expand...

This requires preparation of botable floppy/USB/DVD with BIOS image file and after making preparation set a jumper. Depends on motherboard, for dual BIOS it is different, read HP documentation.
BIOS recovery procedure: http://h30499.www3.hp.com/t5/ProLia...covery-Proliant-Microserver-N36L/td-p/4726687
Re: regulators, not so simple as described. Checking FETS is not so easy as bipolar transistors, also need to check the main CPU VRM chopper.

Have you tried to leave the machine on for several hours to see if it will boot? Have you tried to reseat the Ram and boot the motherboard?

Here is datasheet for a typical 3 phase controller which supplies voltage to the processor. I think that your motherboard might use something similar (see if you can identify the relevant chip near the processor and look for its datasheet). If you look at the schematic you should see a set of three identical drive circuits each with 2 FETs. With the circuit board disconnected from all power, as a rough check you could try to test between the gates, sources and drains of each transistor (with them still soldered in place on the PCB) using a simple multimeter and compare resistance readings (both polarities) between each of the two sets of three identical devices. The DC resistance values should be very close between each device unless one of them has malfunctioned. You could also try to measure the Vcore voltage provided that you can work out where to measure it safely. If there isn't an obvious fault on the board (has anything exploded, swelled, changed colour, melted etc.?) such as a faulty transistor, capacitor or inductor I don't think that there is much point in spending time on doing a lot of troubleshooting.
http://www.st.com/web/en/resource/technical/document/datasheet/CD00040795.pdf

There are a number of cheap (~R500-R600) mini-ITX motherboards based on the AMD C-60, E-350 and Intel Celeron 847 which should perform better than the AMD Turion and include an HDMI port. Take a look on Amazon and Newegg for some ideas. The MSI E-350 board was available locally online a few months back. You could also look at one of the locally available CFi mini-ITX cases if the MB does not fit in the HPMS.

Check for bulging dead capacitors, Ive fixed dozens of dead motherboards because of that.

Yotch said:

The DC resistance values should be very close between each device unless one of them has malfunctioned. You could also try to measure the Vcore voltage provided that you can work out where to measure it safely.

Click to expand...

Oh, this is wrong. It will be always the same, even one FET has failed. A reason - all FETs are only separated by inductor (like in PDF you are refering to), they will show the same resistance.
Yes, this example shows that for safety reason leave inspection of VRM/VID controllers to the trained personel. I agree with.

@Yotch. I am right or wrong? In this case I can be either right or wrong. What do you say?

Posting like this is not helping at all, unless it is clear at the end which advice is correct and which one is wrong.

Edit: Silence says all.

Thanks all - sorry was away for a while...

@ sajunky - I tried the BIOS recovery / image file method but the machine doesn't power on to do the recovery - when the power cord is plugged into the PSU, the only activity is the cyclic flashing HP logo leds - pressing the power button does nothing and the normal orange power light is not lit. I have left it with the power cord in for 60+ mins but nothing else happens. Also tried removing the BIOS battery for a few days, no luck.

@ Yotch - thanks for the tech info, however it's way above my level - I can do basic replacement soldering of caps (done previously on other boards with success), however none of them look like they're blown/bulging in this case (@FaSMaN) - is there another simple way to test if a capacitor is functioning correctly in order for me to check them all as a last resort.
Will also look at the mini-ITX boards and see what I can find.

Pics of the board (click to see larger size):

@sajunky: Several years ago I installed a Socket 754 Sempron CPU in a new Foxconn 760GX motherboard inside a cheap case with a generic PSU. The moment I switched on the machine the PSU started to make a sizzling sound and some white smoke wafted out of the vents. I replaced the PSU only to find that there appeared to be a short circuit somewhere on the motherboard as the new PSU would just shut down immediately. After looking at the L6711 datasheet that I posted above and comparing the resistance values in both directions between the gates, sources and drains of each of the six FETs in the Vcore power supply I noticed that one of them appeared to have shorted out and was giving different resistance readings to the other two identical FETs. There is a picture in the second post in the link below which shows the six FETs in a row on a typical motherboard. You can see the CPU socket nearby.

At the time I needed the machine rather urgently and I was able to source a cheap replacement MB so I did not complete the repair as I would have probably needed to order a replacement from eBay as it was not available locally. I was also not 100% sure that the fault was limited to the FET as the controller might also have been damaged or was faulty to begin with. In this case though it may be quite expensive to replace the whole Microserver so it seems worthwhile just to do a few basic tests before scrapping the board.

http://www.benchtec.co.uk/forums/th...nt-information-on-measuring-your-cpus-voltage

@Howard: Looking at the photos that you posted I would guess that the Turion CPU is underneath the big heatsink near the centre of the motherboard. I don’t see a lineup of six FETs as shown above so the Vcore PSU seems a bit different. I would try to look for the Vcore PSU controller chip which should be near to the CPU. Just write down the numbers of all of the nearby chips and post them in this thread or use Google to try to find the datasheet. Once you have this information you should be able to work out where to measure the Vcore voltage or where to look for a possible faulty FET etc. Unfortunately it seems that you don’t have access to any of the Turion CPU pins as they are hidden so you need to look at the PSU instead.

Do any of the heatsinks get warm when you turn on the machine? Do any of the fans, hard drives, flash drives etc. start up? Could the power supply be shutting down due to a short circuit on the motherboard? Some of the lights are still turning on though?

Yotch said:

comparing the resistance values in both directions between the gates, sources and drains of each of the six FETs in the Vcore power supply I noticed that one of them appeared to have shorted out and was giving different resistance readings to the other two identical FETs.

Click to expand...

Only the shortened gate would give different indication. It can happen, but I have challenged you, as it is not a way to test this circuit. Not with multimeter.

I also do not advice generic public to meassure CPU voltage and the link provided confirm this. You need to have some (at least basic) electronic background to perform certain operation and interpret results.
@Howard. There is nothing from the photos. If you are traveling William Nicol near Diepsloot, you can drop this motherboard here for few days and I can check it when time allows. No charges, no promises, as if fault is in the chipset (the most common) or surface mount controlles, I will not be able to replace it. Just checking basic components. PM to my Email address.

My N36L also gave up the ghost last night. Used as my main media center.

Except mine is completely dead, no light, and bugger all happens when you push the on button, when I disconnect the power cable, I can hear the sparks on the contacts.

My guess is power supply?

Is it standard ATX-24-pin PSU?

Just some info for you guys. It is definitely a lot cheaper to just get an entirely new MicroServer than to replace the motherboard.


620826-001 (for N36L) R3979.94 ex vat ETA 7 – 10

661787-001 (for N40L) R3435.77 ex vat ETA 7 – 10


You must be really desperate to fork out this cash for this motherboard.

TJ99 said:

It's not very likely you'll get a replacement board for less than the cost of a new server. That's why everyone buys the microservers, they're cheaper than just an HTPC board (or they used to be). Also even if you do find one for cheap, it won't have the connectors the HP one has, especially for storage.

Click to expand...

DrJohnZoidberg said:

Just some info for you guys. It is definitely a lot cheaper to just get an entirely new MicroServer than to replace the motherboard.


620826-001 (for N36L) R3979.94 ex vat ETA 7 – 10

661787-001 (for N40L) R3435.77 ex vat ETA 7 – 10


You must be really desperate to fork out this cash for this motherboard.

Click to expand...

Q.E.D.