Overclock.net banner
1 - 6 of 6 Posts

Call me VSG
12,870 Posts
Best bang for the buck watercooling pump

review by geggeg

Announced at CES 2014, this pump was something I was looking forward to a lot- especially coming from the MCP35X I have been using so far. So when I started my big build and realized a single MCP35X wouldn't necessarily be able to handle what I planned for the loop I got in touch with Bryan and he helped me tremendously by sponsoring the build with 2 MCP50X pumps. This review is a result of that, mostly because I was curious to see how these handled in comparison with the other pumps I since procured- another MCP35X and a Laing D5 Vario.






The pumps come in a glossy black box, in contrast to the usual Laing simple white boxes. A sticker on the side gives the specs of the pump. If you notice, the pump has advertized head pressure higher than the MCP35X but at a lower max flow rate. But what's inside is what matters so let's take a look:



Inside is the pump itself in a bubble wrap pouch for protection, 2 mounting screws, 2 neoprene mounting pads and an instruction manual. Let's take a look at the pump now:



The first thing I noticed was how "solid" it is compared to the MCP35X- weighs more and feels sturdier as well.





G 1/4" threads tapped in for inlet at the top and outlet coming off the horizontal side.


The "Warranty void if removed" sticker. But, as we will see soon, this time the sticker is actually there for a reason- warranty will indeed be void if one opens up the pump for any reason unfortunately.


SATA for power! If everyone replaced molex with SATA for power, the world would be a much better place



Now I realize there likely won't be any 3rd party tops but never say never. The pump body sans top is pretty much 1" tall, and the entire unit is 2.4" x 2.4" x 1.8" for those looking for compact pumps.



In fact one can see the pump is only just taller than the MCP35X unit (2.4" x 2.4" x 1.35"). But, if claims of this pump not needing a heatsink/fan are true, then things change drastically:



This is even before the heatsink feet come in. The heatsink incidentally is 3.15" x 3.15" x 0.71" (Again before the feet) so the "compact" MCP35X isn't necessarily so. Speaking of a not compact pump,



Now I did have a Bitspower delrin D5 mod top on but there was something wrong with that top resulting in big performance degrading. So for the tests coming up, I used the stock top as seen above.


I must note that the disassembly was done post testing to make sure there was no issues arising from possibly incorrect reassembly. First up, the warranty sticker and the 4 philips head screws in the bottom:



Inside we see something vastly different from the Laing DDC:


The magnetic impeller/rotor unit is a totally new part made by Swiftech for this.







We see a gasket/O-ring that by touch and bounce characteristics suggests is made of Buna-N. I could be mistaken here but suffice to say that it does the job.


Unlike the Laing DDC, there is no ceramic ball bearing here. I am not familiar enough with these parts to claim what this is, but the ball bearing was the one piece prone to wear and tear in the DDC pumps. Swiftech claims the same 50,000 hours of MTBF for both the MCP35X and MCP50X. Now let's remove the motor unit out:





No, that's not water. It is in fact a glue of sorts holding the PCB and motor tight. I also opened up an MCP35X for a side by side comparison (MCP50X part on left, MCP35X part on right):








Now here's why Swiftech has decided to enforce voiding warranty if the MCP50X is opened:


These two small washers are part of the impeller unit and hold it in place on the pump motor. Here's a look at how the unit looks without the washers:


and how it looks with them:


Needless to say, if you aren't looking for these then it can be very easy to lose them. The thicker, outer washer also is shaped to fit in perfectly:



So needless to say, it took careful and steady hands to put these back together. While I don't really see how the washers would come out when removing the pump top (assuming upright position), I can understand the reasoning behind the warranty being void. There are plans for another top from Swiftech but if that comes to fruition then it will likely only be sold pre-installed with the pump for this very reason.

Performance testing

Since I had 2 each of the MCP50X and MCP35X as well as an MCP35X2 top, I decided to go ahead and make 2 test loops- one close to what I will eventually have in my GPU loop and another more representative of a typical user's loop. The components are listed below:

Reservoir: FrozenQ Liquid Fusion 400 mL cylindrical reservoir
Controller/readout: Aquacomputer Aquaero 6 XT
Pressure drop readout: Dwyer 490-1 wet-wet manometer (with Koolance NPT 1/8" male to G 1/4" female adapters)
Flow meter: Aquacomputer High Flow non-USB previously calibrated from a King Instruments rotameter
Sound meter: Extech AN10 sound probe previously calibrated in a sound lab
Temperature: Rosewill IR thermometer

High restriction loop (components between the high and low pressure ports on the manometer): Watercool Heatkiller Mo-Ra3 420 Pro 9x140, 4 Koolance QD3s, EK FC 780 GTX Classy water block and 2 Bitspower 1/2" x 3/4" compression fittings w/Primochill Advanced LRT Clear 1/2" x 3/4" tubing

Low restriction loop (components between the manometer ports): EK Supremacy Evo CPU block, Swiftech R9-LE full cover water block, XSPC AX480 radiator with 6 Bitspower 1/2" x 3/4" compression fittings w/Primochill Advanced LRT Clear 1/2" x 3/4" tubing

By "2x MCP50X" or "2x MCP35X", I am referring to connecting them together using fittings and not a dual top (as in the case of the MCP35X2).

Each loop was set up inside an isolation box for a controlled ambient temperature (20 ?C +/- 0.25 C) and ambient noise level. Due to the box not being personal property, I wasn't able to decouple any pumps by using any double sided foam/neoprene so the sound levels measured may well be higher than what people will notice in general. Each pump was connected to the reservoir using an XSPC rotary male-male adapter except in the case of the D5 where more compression fittings/tubing had to be used with the stock barbs. Measurements were taken after the loop was bled of air and a further 5 hours had passed for steady state to be achieved. In the case of multiple pumps, average pump noise and temperature is an average of the two pumps together. Each measurement set was repeated 5 times for statistical accuracy. Finally, the MCP35X pumps were used with the heatsinks on because the isolation box did not have the best air flow and I didn't want to take the chance of getting them too hot. So temperatures there may well be better of worse compared to your situation depending on air flow, use of heatsink, fan etc.

PWM Response


Anyone familiar with the PWM controlled Laing DDC is aware of the short duty cycle- typically between 20-60% PWM signal. So the near linear response spanning the entire PWM range for the MCP50X was impressive to me. The same trend holds with multiple pumps:


Part 1: High restriction loop

The pressure drop from the components listed above reached a max of > 7 psi at the max flowrates achieved! Let's take a look at the flowrate behavior with PWM control, and also RPM:


As was the case with the PWM response curve, flowrates can be controlled near linearly with the MCP50X which also has a larger range of flowrates provided- 1.3 GPM to 0.15 GPM. A similar scenario is seen when the multiple pumps are added in:


Looking at this, one would be tempted to think the MCP35X is a better performer. But remembering how the PWM response is for the pumps, and plotting the flowrate vs RPM curve (Where the D5 vario comes in as well), things look different:


Both the MCP50X pump samples (randomly picked, with distant serial numbers) performed near identical, which is good to see. The production of these pumps doesn't seem to have any effect of variability then. At the same pump RPM, the MCP50X puts out a higher average loop flowrate compared to the MCP35X and also the D5 (which to be fair isn't in it's strong zone here anyway). When adding in the multiple pumps, things don't change much:


The one noteworthy point here is the effect of a dual top- especially one that is arguably the best performing top on the market. The MCP35X2 is a better raw performer here and I can only imagine what an MCP50X2 dual pump top would do here if Swiftech were to ever make one.

But what about pump noise?


The MCP50X at full speed is quite loud. There's no denying it here- both pump exhibit near identical sound profiles. But when toned down to 70% or below, they get a lot quieter- I can't emphasize enough how big of a change it is going from 100% to 70% to 50%. As seen above, they get into D5 noise territory which is remarkable. When looking at multiple pumps as well:


It is a similar story, except with the MCP35X2 top again being exceptional. I was skeptical of how it was that much quieter than single pumps at the same flowrates so I set up the loop outside at home- no difference.

Next up, pump temperatures (remember that the MCP35X pumps were with the passive heatsink on, and the D5 is cooled by the loop):


The D5 was the coolest, and it makes sense too. Even without a heatsink on, the MCP50X pumps were cooler than the MCP35X w/HS so that's impressive again. I have heard of way too many dead DDCs because of overheating so this is a good piece of news. When adding in multiple pumps, the same story comes up- a good dual pump top is better than 2 pumps connected together:


There is quite an appreciable increase in temperatures in the case of 2x MCP35X. I must note here that it was the 2nd pump (not directly connected to reservoir) that was the hotter of the two. Heatsink mounting was not a factor here either.

Part 2: Low/medium restriction loop

The maximum pressure drop measured was a little over 3 psi here. I only ran single pumps here- given the flowrates achieved with single pumps alone the need for dual pumps is more for redundancy here than performance in my opinion. Let's see how the flowrates change with PWM signal and then also RPM:



The maximum flowrates achieved are quite higher here than in the previous loop which again agrees with what was expected. The D5 here performs much better than before also but the MCP50X and MCP35X continue to right up there at the top. Once more there is a larger range of control with the MCP50X vs MCP35X.


When look at flowrate vs noise, a similar case as with the high restriction loop emerges. The MCP50X is still the loudest pump at full speed/near full speeds, quietens down a lot at 70% PWM signal or lower and is as loud (quiet) as a D5 afterwards. For people looking at 1 GPM flow or thereabouts (whether this number is anything more than a suggested rule of thumb or otherwise), the D5 is the quietest followed by MCP50X and then the MCP35X. When properly decoupled, I expect the trend to continue but at lower absolute numbers. There was no pump vibration in any case to add to noise.


Here's one of the more telling graphs. At low/med restrictions and tuned down via PWM, the MCP35X pumps get hot really quick even with a passive heatsink on and a bit of airflow in the chamber. The D5 and MCP50X pumps in comparison run much cooler.


The MCP50X is available for sale in the USA at an msrp of $69.95 with reseller pricing going lower with discount codes if applicable. The MCP35X has an msrp of $99.95 but can be found for $85 or lower new. A stock D5 vario, such as the Swiftech MCP655, runs at an msrp of $108.95 but identical D5 vario pumps can be bought for $79 or lower. The MCP35X heatsink is a further $14.95 and a decent D5 top runs another $30+ on average. MCP50X definitely is the most bang of your buck pump here.

In terms of size and connectivity, it is the most compact pump (excluding the MCP35X without HS) and has a SATA connector for power. On the flipside, it only comes with PWM regulation so and not all 4 pin "PWM" headers are actual PWM controllers as seen here: http://www.swiftech.com/pwmcontrollers.aspx

When comparing MCP50X vs MCP35X (or any Laing DDC for that matter), if PWM control is not a hindrance then go with MCP50X. It is louder at full speed but gets quieter quickly, offers more head pressure, has a "fuller" PWM response curve and runs cooler. It is also a lot less expensive. There have been no long term tests on the pump yet, for obvious reasons of availability only being a couple of months now, but so far so good.

When comparing MCP50X vs Laing D5 (unfortunately I do not have a PWM controlled D5 nor a D5 Strong), the former is a no brainer in a high restriction loop. But in a more typical low/med restriction loop, I would probably go with a D5 if space or cost is not a concern. If it is, then at ~1/2 the cost of a D5 with a 3rd party top the MCP50X is a good choice. Irrespective of what you go with, Swiftech has done a good job here. I would have liked to see possibly longer cables (sleeved cables are a nice bonus as well) and really would love to see more tops- a plexi version and also a dual pump top for those having a different aesthetics preference or wanting redundancy/more performance. The only reservoir that mates directly with the pump is Swiftech's own MCP35X-Res but a simple male-male adapter will connect just about any reservoir out there.

Price, performance, temperatureNoisy at full speed, lack of aftermarket tops/reservoir mounting options


1,959 Posts
Best bang for the buck watercooling pump

review by geggeg
wow. ive felt first hand how hot a mcp35x can get within seconds in a low restricted loop, so seeing your results is what i find most impressive. fantastic job, swiftech. welp, you've convinced me. im definitely ordering one now to pair with my mcp35x (separate loops). awesome review!
1 - 6 of 6 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.