Intel R1000/R2000 GZ/GL Fan

[Adraen]

Hi everyone,

 

As some of you might have seen I'm working on a build around an Intel S2600GL motherboard (https://linustechtips.com/main/topic/605605-ebayinator-ii/), I spent some time this weekend trying to figure out how the fan headers are wired but without much success ... I got the obvious +12V and GND as well as the status LED (pinout from the motherboard datasheet) but I'm struggling with the PWM and Tachometers that don't seem to follow the same norm as typical 4 pins fans as well the present pin which I'm not sure if it's an input or output ...

 

Anyway I was wondering if anybody here on the forum would have fan for such a server to do a quick inspection on how the pins are connected? If anybody has access to such fan but doesn't know how to go about it I would be more than happy to help!

 

If you are wondering the connectors for the fans are the small inline 10 pins connectors below the RAM slots and the matching plug I got is: http://uk.rs-online.com/web/p/pcb-connector-housings/2799207

 

Cheers

 

[leadeater]

Try this, page 104. http://www.intel.com/content/dam/support/us/en/documents/motherboards/server/sb/s2600gzgl_tps_r2_4.pdf

 

I have a slightly older Intel server board, S5520HC but this uses standard 4 pin connectors.

[Adraen]

41 minutes ago, leadeater said:

Try this, page 104. http://www.intel.com/content/dam/support/us/en/documents/motherboards/server/sb/s2600gzgl_tps_r2_4.pdf

 

I have a slightly older Intel server board, S5520HC but this uses standard 4 pin connectors.

Thanks, that's actually what I already had a look at, I assumed the pinout was different but the voltage levels the same but it looks like it's a bit more complicated than this unfortunately.

 

For instance I measured the PWM levels at 12V which is much higher than the max allowed by traditional PWM fans http://www.formfactors.org/developer\specs\rev1_2_public.pdf and also it doesn't give much detail on what the "present" pin is about (I tried to pull it low with a 10k resistot but didn't change a thing) or what the voltage levels are (well it's a manual not really a technical datasheet)

[leadeater]

30 minutes ago, Adraen said:

Thanks, that's actually what I already had a look at, I assumed the pinout was different but the voltage levels the same but it looks like it's a bit more complicated than this unfortunately.

 

For instance I measured the PWM levels at 12V which is much higher than the max allowed by traditional PWM fans http://www.formfactors.org/developer\specs\rev1_2_public.pdf and also it doesn't give much detail on what the "present" pin is about (I tried to pull it low with a 10k resistot but didn't change a thing) or what the voltage levels are (well it's a manual not really a technical datasheet)

The present pin and LED pins should be OEM only for the server management IPMI/BMC, total assumption of course, but this information can be used for active airflow management and whether or not it will even allow the server to power on in some cases.

 

Going by this likely awful assumption it looks like the PWM pin is shared between two fans, could explain the higher voltage if there was also some kind of voltage divider where the fans actually plug in. Usually those plugs go off to another board mounted in the front of the rackmount case where the fans plugs in using standard 4 pin.

[Adraen]

2 minutes ago, leadeater said:

The present pin and LED pins should be OEM only for the server management IPMI/BMC, total assumption of course, but this information can be used for active airflow management and whether or not it will even allow the server to power on in some cases.

 

I do think so too, as for this board the fans are actually managed by the BMC (which means that you need to reflash the BMC to update the profile). I measured the present pin and got 3.3V when the machine is on, but from the manual it looks like it's active-low (the N at the end of the pin name). My assumption was that it was an input pin tied to ground by the fan to indicate a fan was plugged (so you can know if the fan is faulty if the fan is present but the tachometer isn't ticking).

 

2 minutes ago, leadeater said:

Going by this likely awful assumption it looks like the PWM pin is shared between two fans, could explain the higher voltage if there was also some kind of voltage divider where the fans actually plug in. Usually those plugs go off to another board mounted in the front of the rackmount case where the fans plug in using standard 4 pin.

I'm pretty sure you're right about the PWM pin being shared, AFAIK the fan on the R1000 are double bladed fan (to have very high static pressure) and each fan has it's own tachometer (from the datasheet) but a single PWM is used to control both fans. I will get an oscilloscope and check that out see if the PWM signal is just a standard carrier but at 12V instead of 5, if that's the case a simple voltage divider would do the job. Still have to figure out the tachometer

 

Thanks for the help!

 

[leadeater]

6 minutes ago, Adraen said:

-snip-

Are the fans hot swap like in HP servers or is the cable actually attached to the fan? The fan replacement steps in the manual makes me think it's attached. Seems Intel does things a little different to HP, Dell, IBM/Lenovo etc.

[Adraen]

3 minutes ago, leadeater said:

Are the fans hot swap like in HP servers or is the cable actually attached to the fan? The fan replacement steps in the manual makes me think it's attached. Seems Intel does things a little different to HP, Dell, IBM/Lenovo etc.

I'm not sure at all, In the manual (page 58) there is "Hot-swap fan support" 

 

Actually going through the manual again I got

Quote

Each fan is associated with a fan speed sensor that detects fan failure and may also be associated with a fan presence sensor for hot-swap support

 

Based on http://www.intel.com/content/dam/support/us/en/documents/motherboards/server/sb/r2000gzgl_serviceguide.pdf (page 50) it looks like you're right there is a connector/front-panel to which the fan connects to, I'm not sure if the board actually does something or not.

 

Would be nice to see someone on the LTT forum with access to one of those chassis

 

Cheers

[leadeater]

3 minutes ago, Adraen said:

I'm not sure at all, In the manual (page 58) there is "Hot-swap fan support" 

 

Actually going through the manual again I got

 

Based on http://www.intel.com/content/dam/support/us/en/documents/motherboards/server/sb/r2000gzgl_serviceguide.pdf (page 50) it looks like you're right there is a connector/front-panel to which the fan connects to, I'm not sure if the board actually does something or not.

 

Would be nice to see someone on the LTT forum with access to one of those chassis

 

Cheers

I'll have a look at a couple of different generation HP servers at work and see if they also have the same 10 pin connector, if it does I'll see if I can have a look at where it goes etc.

[Adraen]

2 minutes ago, leadeater said:

I'll have a look at a couple of different generation HP servers at work and see if they also have the same 10 pin connector, if it does I'll see if I can have a look at where it goes etc.

Thanks, much appreciated :), I will have a look on my end as well but pretty sure none of the server we have are Intel. I will get a scope home tonight and start investigating a bit as well when I have some time and update with the details either here on the build log in case some people are interested.

[Adraen]

Update: So I had a look at the servers I had around, all of them use traditional 4 pin connectors, so no luck.

 

I brought the oscilloscope and had a quick look around, at first glance it looks like PWM, TACHO1 and TACHO2 are pulled-up to 12V and the present pin is pulled up to 3.3V. Not much more, the PWM line doesn't oscillate (I guess because there is no tacho reading)

 

Anyway I will keep digging! Worst case I could just use the 12V line from the fan connector but what's the fun in that, having PWM and tacho reading would be much nicer.

[Adraen]

(I hope it's okay to post after one (two) of my posts?

 

Did a bit more debugging today, looks like I got the tacho reading working, the 12V is fine (but I think I killed one of the fan because of the PWM level at 12V instead of 5 ...)

 

 

The sensor reading from the BMC a bit lower than it should as the arctic F12 should be up to 1350 RPM but it's a detail. Regarding the Critical Status I would guess it's because of the present pin not being asserted (but I still don't know how to deal with this one). Later on I will try to plug a second fan on tach1 of the same port, might be because it needs both fans to be operation (1A and 1B)

 

If anyone is interested here is the tacho reading:

 

EDIT day after: Looks like it was simply requiring both FAN1A and FAN1B, I hooked a second fan and it works, at least for the sensor and fan status  still need to look into the PWM

 

[Adraen]

Me again, looks like there is little interest here but anyway it becomes a struggle-log

 

Came across an image of the FAN for the system and the fan connector also has 10 pins which reassure me quite a bit about the fan housing not doing much (or anything at all)

 

I also found a patent (https://www.google.com/patents/US20140211353) which describes the PRSNT_N pin and it uses the voltage drop across the diode to detect if the fan is present or not. I grounded PRSNT_N hoping to get something on the PWM line but still nothing just a straight 12V signal ...

[leadeater]

That should help, I'll get someone I know who is a registered electrician to have a look at that and see what he says.