There is no risk to putting the pump on a PWM header. You have the option of having FanXpert run the pump at a constant speed (100% or otherwise) or having the pump respond to changes in temp. Swiftech is the clear industry leader in the "All-in-one" cooler category and they provide an 8 pin PWM Splitter and IIRC pump speed varies from 1200 to 3000 rpm.
Yes, it's certainly not necessary to control the pump speed, but there are wear and noise advantages to doing so. You are also losing one of the great features of ya MoBo which is the ability to control this to your liking. Most modern Z87 MoBos come with a PWM CPU and CPU_1 header and with Z97, every board I have seen has PWM available on all chassis and CPU headers. Ideally ya wanna control the fans and pump separately.
While the 105 pump is a bit weak, certainly no harm and some advantage to using speed control on it. The higher the speed the faster the wear rate and the rate is not linear, it's exponential. Noise is an issue also
Here's how I am using Asus Fan Xpert (4 Channels)
Channel 1
CPU Header => Pump No. 1 (Swiftech 35X) .... rpm range 1500 - 4500 (0.75 to 2.25 gpm total w/two pumps)
CPU-1 Header => Pump No. 2 (Swiftech 35X).... rpm range 1500 - 4500
Channels 2 - 4
CHA_1 => Fan Hub 1 => 6 Fans on 420mm Radiator ... rpm range 325 - 1250 (never breaks 850)
CHA_2 => Fan Hub 2 => 4 Fans on 280mm Radiator
CHA_23 => Fan Hub 3 => 5 Fans on Case
I used 6 temp sensors and 2 flow meters to set up the loop but this is very easy do.
Option 1 - It is not necessary to have variable speed control but you will certainly benefit from running at a fixed speed lower than max. FanXpert has both fixed and variable speed options so if ya wanna run at fixed speed. Run Intel XTU and monitor Core temps. In order to balance noise / wear versus performance, in FanXpert, click the radio button for fixed rpm and set speed to max. Drop the speed in increments of 100 rpm (or whatever seems comfy) and let the system stabilize. Record temps versus rpm, both of which are visible in the AISuite window when running FanXpert. Those are not CPU core temps tho, that's the CPU "package temp" which is much lower. Once done, suggest using HWiNFO64 for core temp measurements and make sure you are comfy w/ those.
Rinse and repeat till ya get down to about 30% of full speed (that's typical for WC pumps, not sure if it's of any benefit with Corsairs stuff to go below 50%) . Obviously stop if ya see any temps ya don't like. What you will observe generally is that there is no change in thermal performance (or even flow rate if measuring) for the first series of drops.....on my pumps, 70% and 100% give same performance. So, there is no reason whatsoever for me to run at above 70% as the noise difference is observable and I get no real benefit. So if I was to run at fixed rpm, I would set the pump to always run at 70% of full speed w/ FanXpert.
Option 2 - After completing the above, you can leave your machine at idle and do the same ... observe temps versus speed. Mine worked fine at 30% and I would not go below that in any case.
Of course using a water temperature sensor would be far more useful in this regard.
I should note that under CPU testings my fans barely move, I have to add Furmark to increase water temps more than 2C...even than Delta T (water - ambient) stays under 10C and rpms never break 850 rpm. The entire PC is dead silent, you can not tell the system is on even with side panels off. However at 4500 rpm I can hear the pump....so far it has never broken 3100 rpm and it's inaudible at that speed.
Now the thing with the H series from Corsair with the 2700 rpm fans is that the fans are so loud (equivalent to a 1950s style vacuum cleaner) they drown out any pump noise. So I do see techy's point in that reducing pump noise may have no real advantage there. However, the wear and tear issue is something to consider.
I wish you would have had an opportunity to check out the H220-X before purchasing the H105....for an extra $15, would see a world of difference..... Corsair also provides a 8 pim PWM fan splitter hub ($11) in the kit to address your issue, making the cost the same.
The H220-x tops every ever CLC on the market and has numerous advantages:
-Can open loop and add blocks
-All copper components, versus Corsair's aluminum
-All custom water loop grade / quality components
-1/6th as loud as the H100i / H105
http://www.hitechlegion.com/reviews/cooling/liquid/40870-swiftech-h220-x-open-loop-240mm-cpu-cooler-review?showall=&start=3
However, with a CLC you are limited to only what it is, and only the performance it can deliver as assembled. With that in mind, the H220X makes infinitely more sense as a starting point into liquid cooling. It is ready to go out of the box and requires no maintenance, but it is also fully expandable when you are ready to take the next step. Even if you aren’t looking to take the next step, the H220X outperforms every CLC on the market, and does it at more than 20 dB quieter.....
There is not a 240mm CLC or air cooler that can beat it, and it does it at 20+ dB quieter than the competing CLCs. To sum it up; the H220X offers better performance, lower noise, better aesthetics, flawless design and build, better components and the option of expandability when compared to a CLC. Putting it gently, choosing any CLC over the H220X would be doing yourself a huge disservice.
I should also note than FanXpert has an option to shut off case fans below a certain point (recommended) and to ramp up / ramp down fans at faster / slower rates. Experiment with these as when subject to varying loads, the fans can get caught "chasing their tail" so to speak and in some instances results in whiring sounds as fans repeatedly spin up faster than down slower...... reminds me of a new driver learning to use a clutch / stick shift.