Intel Atom C2000 Failures

[ericloewe]

Hi everyone! greetings from Spain.
I have a DELL S4048-ON switch crashed after shutdown and reboot. I think I'm affected by this Clockgate from Intel, I cannot get any data via serial console and switch try to boot in a loop. I search all the web and I saw same problem on this model (2015 factory date).
I was wondering if anyone has S4048-ON's schematics so I could find LPC_CLKOUT on the mainboard. I see a connection pinout near microprocessor (pic attached), probably one of those pins is LPC_CLKOUT.

These switches have an ASPeed BMC, don't they? Look for a daughterboard with an American Megatrands [sic] MegaRAC sticker. The BMC connects via LPC (and USB and PCIe) to the host, so that would at least narrow down your search a bit.

[pat0te]

Stick a scope probe to it?

Unfortunaly, I don't have any scope.

[pat0te]

Hi everyone! greetings from Spain.
I have a DELL S4048-ON switch crashed after shutdown and reboot. I think I'm affected by this Clockgate from Intel, I cannot get any data via serial console and switch try to boot in a loop. I search all the web and I saw same problem on this model (2015 factory date).
I was wondering if anyone has S4048-ON's schematics so I could find LPC_CLKOUT on the mainboard. I see a connection pinout near microprocessor (pic attached), probably one of those pins is LPC_CLKOUT.

These switches have an ASPeed BMC, don't they? Look for a daughterboard with an American Megatrands [sic] MegaRAC sticker. The BMC connects via LPC (and USB and PCIe) to the host, so that would at least narrow down your search a bit.

I think this switch has not BMC, but I'm going to take a look. Thanks for your suggestion.

[pat0te]

Hi everyone! greetings from Spain.
I have a DELL S4048-ON switch crashed after shutdown and reboot. I think I'm affected by this Clockgate from Intel, I cannot get any data via serial console and switch try to boot in a loop. I search all the web and I saw same problem on this model (2015 factory date).
I was wondering if anyone has S4048-ON's schematics so I could find LPC_CLKOUT on the mainboard. I see a connection pinout near microprocessor (pic attached), probably one of those pins is LPC_CLKOUT.

These switches have an ASPeed BMC, don't they? Look for a daughterboard with an American Megatrands [sic] MegaRAC sticker. The BMC connects via LPC (and USB and PCIe) to the host, so that would at least narrow down your search a bit.

Zero chips or stickers found with an American Megatrends / MegaRAC ::sadface::

[ericloewe]

Ok, further digging and a series of fortuitously timed posts by a few people led to further useful info about Supermicro A1SAi/A1SRi boards, and likely other vendors' board too:

Well, the gist of it is that there are two clock lines, one feeds the TPM header and is involved in the boot ROM selection; the other runs to the BMC. This is why fixing the boot failures often still left the BMC unresponsive to the LPC bus, it just wasn't getting any clocks.

Supermicro's fix for this is a pair of 150 Ohm bodge resistors, one to pull up each of these two clock lines to 3.3 V. The fix is actually pretty neat on these boards and could pass casual inspection, depending on how well the soldering turns out.

[Sacodepatatas]

Unfortunaly, I don't have any scope.

You can buy a DIY 50MHz Frequency counter for less than 5€ from AliExpress/eBay. Or you can build a temporary one by yourself if you've got a PIC16F628A and some led displays in your spare box.

Edit: Or one of those EZ-USB FX2LP boards that can be programmed as a logic tracer and costs about the same price.

[wraper]

Unfortunaly, I don't have any scope.

You can buy a DIY 50MHz Frequency counter for less than 5€ from AliExpress/eBay. Or you can build a temporary one by yourself if you've got a PIC16F628A and some led displays in your spare box.

Edit: Or one of those EZ-USB FX2LP boards that can be programmed as a logic tracer and costs about the same price.

And they will be totally useless for the job.

[Sacodepatatas]

Why? I have used such frequency meter for monitoring the stability of 32MHz clock signals, I don't see why a 25MHz Signal can't be counted with these. Maybe related to the logic voltaje levels, but nothing that couldn't be solved quickly. The most simple manner that I think for watching if there is a clock signal at any port, without any equipment at all but a multimeter, is just a high pass filter followed by a peak detector (a schottky, a capacitor  and a high value resistor) and then measure the voltage level at the output.

[wraper]

Why? I have used such frequency meter for monitoring the stability of 32MHz clock signals, I don't see why a 25MHz Signal can't be counted with these. Maybe related to the logic voltaje levels, but nothing that couldn't be solved quickly. The most simple manner that I think for watching if there is a clock signal at any port, without any equipment at all but a multimeter, is just a high pass filter followed by a peak detector (a schottky, a capacitor  and a high value resistor) and then measure the voltage level at the output.

Because there is no signal at all or it's extremely weak once it fails.

[pat0te]

Ok, further digging and a series of fortuitously timed posts by a few people led to further useful info about Supermicro A1SAi/A1SRi boards, and likely other vendors' board too:

Well, the gist of it is that there are two clock lines, one feeds the TPM header and is involved in the boot ROM selection; the other runs to the BMC. This is why fixing the boot failures often still left the BMC unresponsive to the LPC bus, it just wasn't getting any clocks.

Supermicro's fix for this is a pair of 150 Ohm bodge resistors, one to pull up each of these two clock lines to 3.3 V. The fix is actually pretty neat on these boards and could pass casual inspection, depending on how well the soldering turns out.

I saw this repair on Internet, but Dell S4048 motherboard is very different. There is TPM header near microprocessor but only with 1 line of pins.

Video for Supermicros:

[ericloewe]

Ok, further digging and a series of fortuitously timed posts by a few people led to further useful info about Supermicro A1SAi/A1SRi boards, and likely other vendors' board too:

Well, the gist of it is that there are two clock lines, one feeds the TPM header and is involved in the boot ROM selection; the other runs to the BMC. This is why fixing the boot failures often still left the BMC unresponsive to the LPC bus, it just wasn't getting any clocks.

Supermicro's fix for this is a pair of 150 Ohm bodge resistors, one to pull up each of these two clock lines to 3.3 V. The fix is actually pretty neat on these boards and could pass casual inspection, depending on how well the soldering turns out.

I saw this repair on Internet, but Dell S4048 motherboard is very different. There is TPM header near microprocessor but only with 1 line of pins.

Are you sure it's for a TPM? Do you have any documents saying that?

If so, my suggestion is that you figure out which of the pins are not power pins. From there, try pulling up each of them to 3.3 V (one at a time!) until the thing boots. Once it boots, you'll have identified your clock pin and can do a more permanent fix.

Pictures are also welcome for the next person to come along with such a dead switch.

[pat0te]

Ok, further digging and a series of fortuitously timed posts by a few people led to further useful info about Supermicro A1SAi/A1SRi boards, and likely other vendors' board too:

Well, the gist of it is that there are two clock lines, one feeds the TPM header and is involved in the boot ROM selection; the other runs to the BMC. This is why fixing the boot failures often still left the BMC unresponsive to the LPC bus, it just wasn't getting any clocks.

Supermicro's fix for this is a pair of 150 Ohm bodge resistors, one to pull up each of these two clock lines to 3.3 V. The fix is actually pretty neat on these boards and could pass casual inspection, depending on how well the soldering turns out.

I saw this repair on Internet, but Dell S4048 motherboard is very different. There is TPM header near microprocessor but only with 1 line of pins.

Are you sure it's for a TPM? Do you have any documents saying that?

If so, my suggestion is that you figure out which of the pins are not power pins. From there, try pulling up each of them to 3.3 V (one at a time!) until the thing boots. Once it boots, you'll have identified your clock pin and can do a more permanent fix.

Pictures are also welcome for the next person to come along with such a dead switch.

Well, no, I'm not sure about TPM. I don´t have any documentation, that's the problem. I posted this pic before regarding TPM style connector:
https://www.eevblog.com/forum/index.php?action=dlattach;topic=95943.0;attach=1713377;image

[charly]

Just registered (a programmer here, my electronics skills are below zero ) to say thanks for this thread and keep it going for anybody still around (this hardware is way more than what many NAS boxes offer today).
I've had a 2550F since 2014 and it died on me like a year ago (using it with FreeNAS at home). Tried everything, it just turned some lights on the motherboard, but no video, no nothing. I even distrusted the power supply. Now this bridge that I made with audio friends help made it back to life! I didn't want to throw away this beauty even if it just works for a few extra months.
Thanks a lot!!

[Gil]

Where is the pin located on the 251+ ??  many thanks

[JakubPL]

Hey. I managed to (at least I think I did) revive a Dell S4048-ON. It took some effort and I had to sacrifice one motherboard.

But to the point:

You need to solder a wire to the R389 resistor and add a resistor (I used a 330 Ohm one) and connect it to the pins near the CPU where there's 3.3V.

So far, I've been testing it by repeatedly unplugging it and resetting it, and it's working fine.

I hope this helps someone.