cancel
Showing results for 
Search instead for 
Did you mean: 

Do the 2 CPU and AIO fan headers share the same power rail on X670E Crosshair Hero

Castillan
Level 9

As per the attached diagram.  The manual says that the 3 headers support 1A/12W each.

I have a 2-pump water-cooling system.  Each pump can draw up to 20W and they have 4-pin fan header style power connectors.  I have one pump plugged into the 3A Water Pump header and the other pump plugged into a fan splitter that draws power from a SATA plug on the PSU.

I am trying to clean up the cabling in my case.  Ideally I'd like to plug the 2nd pump into a motherboard fan header, and it would be nice if I could use the AIO header.

This got me to wondering if each header is really limited to 1A each, or if it's more like the 3 headers sharing a common 3A power rail, and if for convenience Asus just documents them as 1A each?

I do not have any fans plugged into the CPU fan headers, so I'm wondering if the AIO pump header can support at least up to 2A if the CPU headers are not in use?

I appreciate that this is a strange request for information, but most modern PSU's have their power connectors all share a common power rail even if each individual connector is only rated for so many amps, and it got me to wondering if Asus had done the same sort of thing with the 3 fan headers there.

1,654 Views
1 ACCEPTED SOLUTION

Accepted Solutions

Castillan
Level 9

I broke out the multi-meter.  For what it's worth, the 3 fan headers (CPU/CPU-OPT/AIO) do share a common ground rail, but they do not share a common power rail (meaning a it's not a single 3A power rail shared between the 3 x 1A headers).

That information does answer my question and it's a definite that using a single header is a bad idea.

View solution in original post

7 REPLIES 7

Nate152
Moderator

Hello Castillan,

According to the manual, the W-Pump header is the only header that supports 3 amps (36 watts), all other fan headers including the AIO_Pump header support up to 1 amp.

Therefore, if your pumps draw up to 20 watts, you can't connect the second pump to the cpu fan header or you will burn it out.

I personally would run both pumps off of sata/molex power from the psu.

x670e hero headers.png

 

 

@Nate152 wrote:

I personally would run both pumps off of sata/molex power from the psu.


Thank you for the reply.  It didn't really address my question, but I suppose I was asking about the practicalities of doing something not specifically supported in the documentation.  It's crossed my mind to just manually make up a cable with 3x4-pin headers doing a daisy chain from the power pins of the 3 headers and using just the RPM/PWM pins from just one of the headers and using that to drive a single 4-pin male connector that I can plug the pump into.  Basically like a 3-way splitter, but in reverse.

Now I'm concerned though as to why you are recommending to not use the 3A/36W header with a 1.6A/20W power draw.  Have there been issues reported with using more than 1A on the 3A water pump header?

Castillan
Level 9

I broke out the multi-meter.  For what it's worth, the 3 fan headers (CPU/CPU-OPT/AIO) do share a common ground rail, but they do not share a common power rail (meaning a it's not a single 3A power rail shared between the 3 x 1A headers).

That information does answer my question and it's a definite that using a single header is a bad idea.

Castillan
Level 9

To add a final update on this.  I run my pumps with a 53% PWM signal.  (53% and not 50% because 53 is a prime number, which reduces the chance of harmonics.  The pumps actually run quieter, despite spinning faster, with a 53% PWM signal, rather than a 50% signal).

I measured each pump power draw at 53% PWM to be just 8W at the wall, so likely more like 6.5W or so at the actual headers.  I fashioned up a reverse splitter with RPM + PWR + GND from one header to PWM + RPM + PWR + GND to a 2nd header to a 4-pin connector that the pump plugs into.  This allows the motherboard to provide up to 24W via the headers, which is enough to start the pump up and run it at full speed for the 20s or so until the PWM signal kicks in from the BIOS.  The pump RPM can be read from either header, while the PWM signal is driven by the 2nd header.  I tested with a fan first, to ensure everything was working fine and that each header is supplying power.

All in all, it seems to be working fine.  Touching the wiring, headers, and motherboard around the area of the headers even when the pump is at full speed doesn't indicate any significant warmth, so it appears to be holding up.  In general operation, as mentioned, the pump is only drawing a little over 3W from each header, or ~0.25A, so well within spec and no more than a typical CPU heatsink fan would draw.

Two words for you to consider: inrush current.  You can't measure it easily with a standard meter, and a water pump is one of the cases which will have a higher inrush current due to the inertia of the water.  Normal centrifugal pumps (e.g. a D5 or similar) won't be too bad, compared to other pump types which don't allow the rotor to so easily slip with respect to the water speed, but there's still going to be a high load at startup.  It's unlikely to be the case here, but pumps can draw 5 to 6x their nominal full load current when starting, and initial spikes of >20x nominal are not out of the question.

It just seems like unnecessary risk to the board's power circuits to me, when you can easily avoid the risk by taking the power from a SATA or PCIe power connector.  I can't see any significant downside to just supplying the power from a properly rated circuit and just using the motherboard headers for the PWM and tachometer signals.


@Murph_9000 wrote:

Two words for you to consider: inrush current.  You can't measure it easily with a standard meter, and a water pump is one of the cases which will have a higher inrush current due to the inertia of the water.  Normal centrifugal pumps (e.g. a D5 or similar) won't be too bad, compared to other pump types which don't allow the rotor to so easily slip with respect to the water speed, but there's still going to be a high load at startup.  It's unlikely to be the case here, but pumps can draw 5 to 6x their nominal full load current when starting, and initial spikes of >20x nominal are not out of the question.

It just seems like unnecessary risk to the board's power circuits to me, when you can easily avoid the risk by taking the power from a SATA or PCIe power connector.  I can't see any significant downside to just supplying the power from a properly rated circuit and just using the motherboard headers for the PWM and tachometer signals.


Thank you for the reply.  Yes, what you mention is true of positive displacement or high pressure head pumps, but that isn't the scenario here.  What you write here does come across as unnecessarily alarmist.  If this was a huge issue in the PC water-cooling space, we'd have reports everywhere of people's 2A/3A headers burning out.

As for "easily avoiding the risk", after much searching, I found there to be precious few PSU powered (whether 4-pin molex / SATA or PCI-E) hub style solutions out there which are rated for more than 1A per header.  A couple are rated for 2A/header, and that's still not really any better than my makeshift solution.

There is this Noctua product here that has an adapter RPM/PWM header + SATA power into single 4-pin PWM fan header as one of its accessories.  That would be the ideal for this scenario given the in-rush handling capabilities you describe, but that adapter is not sold as a separate product by Noctua.

One could pick up any number of SATA to 4-pin fan header adapters, and modify those to add on an RPM/PWM to motherboard plug, much like the Noctua adapter does above, but again, this isn't a "buy off the shelf" solution.

Still, after reading your response, I did take it to heart and moved back to my prior hub solution to power the second pump.  I still left the first pump on the motherboard's 3A header.  I re-did the cable routing inside my case to clean everything up and it looks much cleaner.  I wouldn't want to plug both pumps into that hub though, as that may present an overload hazard.  There are some reviews for that hub that mention them catching on fire when overloaded, so 2 pumps into a single point of failure fire-risk isn't ideal.  For the price of that hub though, it's not much of difference to buy the Noctua product and toss everything but the adapter.

The fact of the matter is that there really are precious few easily available off-the-shelf solutions that are properly rated for handling 12W+ PWM pumps once we start factoring in the sorts of in-rush current considerations you describe.

I'm not trying to be alarmist, I just have concerns about running motors with a combined nominal power rating that significantly exceeds the rating of the motherboard header.  I think it's reasonable to assume that the ASUS engineers have designed the headers to cope with the typical inrush current from fans or centrifugal pumps at the nominal rating, so I'm not worried about a single 20W pump of that type on the 3A/36W header, but two of them could be pushing a bit too far into failure territory.  Motherboards are expensive compared to a few connectors and a little time to assemble something that doesn't exceed the rating.

In the absence of finding a suitable off the shelf cable / adapter, I'd just make one for it.  The Noctua NA-AC4 cable included in the NA-FC1 kit would be just the thing, but I agree that it doesn't look like they sell that on its own.  The inline SATA power connector could be problematic to obtain, although you could possibly get one with cables attached and cut the other end off.  Alternatively, the old 4 PIN Molex drive power connectors or PCIe 6 pin power connectors are not too difficult to find.  I would then remove the two power pins from the pump's 4 pin header connector, leaving just PWM and tachometer signals connected to the motherboard, and splice the two power cables into a suitable power connector.  Add appropriate sleeving, heatshrink, cable length, etc to tidy up the solution as you desire.  Essentially convert the pump's wiring to be similar to the EK PWM pumps.

The Akasa FP3S could possibly be an off the shelf solution for running both pumps off a single header setup (with SATA power).