Archive for April 3, 2011

Controlling fans manually has been available via standard fan controllers for some time, you turn a knob up or down and it adjusts the voltage to the fan and resulting speed/noise.  There are many flavors of this all manual control type, but what if you wanted to automate this process similar to current motherboard 4 pin CPU fans such that speeds remain low(low noise) while loads are light, but they are increased when the loads increase(high noise)?  There are several options to do this, but most options typically consist of a very advanced AND expensive fan controllers.

I personally have used a few manual fan controllers.  My first was one from Thermaltake, unfortunately it didn’t take long for that fan controller to burn out.  Then I bought a sunbeam rheobus controller with four channels similar to this.  I used that fan controller for over two years and to this day it still works fine although my loads were always kept fairly light.  It was cheap, but it seemed to hold up and the only obvious difference was the heatsinks that it had.

Soo…..when I heard Sunbeam was coming out with the Rheosmart to convert PWM into analog voltage, I was instantly very interested.  When they hit the shelves at Newegg, I added it to another hardware order I had planned…after all it was only $25 shipped.  It sat in the box for a while and eventually I got around to using it and ultimately doing some testing and this review.

General Photos & Information

Specs say 30 watts per channel max

PWM IN, Fan out, Fan out, Fan out, Molex In

In general the controller looks pretty good from the front.  I do like the mesh finish for some cases and the LED lights are not overly bright like my old model.  The turning knobs however feel cheap because the socket pot flexes quite a bit, but in normal operation it should work well enough.  The PCB parts and heat sinks look ok, although the sinks were not very evenly glued in place.  For the low cost the looks and quality is ok.

Operation is fairly simple.  There is a button below each knob and an LED above each knob.  When the button is pushed in the LED turns red which means it will operate under PWM.  When it is out, the LED turns green which means the knob and manual voltage controls.

Manual control range is good, giving you the ability to control from 0V to the maximum which I found to be around 11.2-11.4V due to losses in the controller while supplying it 12.1V+-.

Setup

The controller comes with a very good manual that should help you right through the process, but I’ll attempt to at least give you an idea how things work.  There are 5 ports on the controller.  Two input and three output.

  • The 4 pin molex connector is what feeds the controller and fans coming straight from the power supply
  • The 4 pin PWM fan connector is actually serving several functions.  One is a pass-through to a pump or CPU fan, in my case I used a pump.  The second function is that it sends the PWM signal and ground to the fan controller.
  • The three out connections are simply power and ground outputs to the fans.  If you use the RPM splitter cable provided  (you only get one of these), you can also send the fan RPM signal to the motherboard.

Here is a diagram of the schematic in picture form:

So, the controller takes the PWM signal generated by the motherboard and converts it to analog voltage which any normal 3 pin fan can use to speed up and slow down.  In my long term test I used the PWM pass through to also regulate the speed of my Swiftech MCP-35X which has it’s own PWM speed controller.  This way PWM is increasing and decreasing speeds of my fans using the controller as well as my pump using it’s controller.

This is what speed fan looks like watching RPM logging:

Testing

For testing, I tried two load scenarios, one light, one heavy:

  • 2 Each Fans (Yate Loon D12SL12) = Approximately a 4 Watt Load
  • 1 Each Pump (Swiftech MCP35X) PWM unplugged = Aproximately a 20 Watt Load

Fans (Light Load) PWM vs Voltage

I have been using the unit to control four yate loon fans on my CPU loop for a couple of weeks now using PWM control with very good success, so I had high expectations the testing would go very well for the light load fans test.

Unfortunately the user manual didn’t provide any sort of correlation between PWM and voltage, so I wanted to see for myself what that would be and plot that relationship. I simply used Speedfan to regulate PWM at 5% increments and a multimeter to read voltage at the fan plug.  This is that result:

Light Fan Load scaled fairly well with PWM, not quite linear, but still a good range. Max Voltage = 11.43V

In my own long term test I have been running the fans in between about 30% to 70% which means the fans have been running between 6.5V and up to 10V.  Note that maximum voltage with this lighter load is 11.43V, and my power supply is providing roughly 12.1V.  Losses within the fan controller are responsible for this voltage loss and it is fairly typical for normal fan controllers to do this unless they have a transformer to boost voltage.

Pump (Heavy Load) PWM vs Voltage

Since the controller was rated to 30 watts, I had hoped the controller would also be capable of running a standard DDC type pump via PWM>Analog control.  If this worked, it could technically turn an ordinary DDC pump into a PWM “like” pump.  I chose to use my Swiftech MCP-35X, purposely disconnecting the pump’s PWM connector so it would function like a normal DDC pump at full speed, and instead connected the 4 pin molex connector to a 4-3pin adapter then to the fan controller to allow it to voltage control.  Previous testing indicated the pump could voltage control from about 7V to 12V fairly well (not as good as it’s own PWM controller, but still some voltage control range). I then followed the same test of turning PWM down from 100% in 5% increments and recorded the voltage:

No good, the pump was too much at 65% or lower.

Unfortunately the test didn’t go well, the pump would repeatedly quit when the PWM percentage was turned to 65%.  Max voltage at 100% PWM was 11.21V.  Min voltage at 70% PWM was 10.02V, so the range was only 1.2V.  I’m assuming there just wasn’t enough current to keep the pump running or some sort of voltage instability that caused the pump to shut down.  I’m not entirely sure, but it wasn’t something I wanted to try long term considering the range was so small.

HEAT TEST

As part of the testing, I also did a manual voltage control on both the fans as well as the pump at two voltage levels.  Using manual controls I was able to dial the pump down further, although the heatsink very quickly became hot.  With my finger I could manage to touch and hold the sinks just barely at lower 40s, but more than that is too hot to touch and hold.  I think some folks understand this, but a fan controller actually has the biggest load when the voltage is lowest or furthest from the source voltage, and that is also apparent in the heat results below:

Heat sink temp using a laser thermometer

With the lighter fans load the aluminum heat sinks kept the temperature at a reasonable lower 40s or about 16-18C higher than ambient.  The pump test however produced fairly high temperatures particularly the 8V scenario which was reaching into the mid 60s.  Perhaps that’s still within specification, but I personally wouldn’t feel comfortable running one this hot especially if you had a valuable hard drive directly above and little air flow.

Conclusion

I am very happy with the controller for my own purpose of controlling around 4 watts per channel worth of fans.  It has given me dynamic loading PWM capabilities to those fans that I would otherwise not have.  My fans now throttle via CPU temperature such that 99% of the time they are a whispery 900RPM, and when I load hard such as rendering a video, playing games, or benching..they turn up to 1300RPM. I have the controller doing my fans, and the Swiftech MCP-35X using it’s own PWM controller, both are using speedfan to regulate.  Between the fans and pump reductions, PWM dynamic loading has cut my ambient noise level from 42dbA to about 35dbA which is fairly dramatic…and I’m happy.  The controller also does allow the user to switch between PWM and manual controls via a push button, so you do have some flexibility there to use a few channels manual and others PWM. In addition the PWM connector includes a daisy chain connector so I can utilize the PWM of my pump.  The pump isn’t capable of running through the controller, but the PWM signal is split to the two components with success.

I would not however recommend this controller for use on pumps or really high fan loads if the intent is to control them via PWM.  My heavy load (approx. 20 watt) test scenario failed to keep the pump operating when PWM was lowered to 65% or lower.  Heat sink temperatures also reached 64C when run passively in an open air test which is getting really hot for my own preferences.  Having the controller buried in a case front where little airflow may be available, could lead to even higher temperatures.

Overall, a great and low cost product to give you PWM like fan control of lighter loads.  I would recommend it for 5W per channel for full control. You could probably do more, but I was only successful in my light 4W test load in making use of the full PWM conversion feature and heat was also fairly comfortable at that lighter level.

Cheers!
Martin

Probably one of two most popular pumps in all of watercooling, the Koolance PMP-450 is a D5 Vario pump and packs a very strong amount of pumping power while retaining it’s built in variable speed controller.  There are multiple flavors of this pump and I have personally used the D5 series ever since my first loop watercooling several years ago.  Koolance has taken this very popular variable speed pump and added what it has been missing for far too long…an RPM sensing wire.

I would like to give special thanks to Tim from Koolance for sponsoring this powerful pump:

Overview

Koolance offers two flavors of the PMP-450 pumps.  One is the subject PMP-450 pump with variable speed control and the second is the PMP-450S pump which is fixed and designed to run at high voltage and higher speeds.

The pump comes in a factory box as a “Bare Pump” type product.

Box indicates it is a Laing D5-38/810 vario with 1/2″ barbs

Well packaged bare pump, no accessories

Both flavors of the pump share the exact same pump housing with the exception of the rear cup in which the PMP-450 has a hole where the variable speed control protrudes.

Variable speed controller and a BLUE RPM WIRE!!!!!

What’s different about the Koolance Brand pump?

There is one thing that I had always very much missed with various other D5 Vario pumps, and that’s the RPM sensor wire.  I’ve have several non Koolance brand variable speed D5 pumps now and none of them ever came with this very valuable feature.  As far as I know, Koolance is the only one to offer a variable speed Laing D5 that comes factory with the blue wire 3 pin RPM sensing wire.

Why is RPM sensing important?

RPM readout provides two important bits of information:

  • Pump and Loop Health Indicator – Having the ability to read RPM is a good way to see the pump is functioning as it was designed.  Sudden changes in RPM are indicators that either something has changed in the loop (IE a block is plugging), or that the pump is experiencing problems.  Without the ability to monitor via RPM, you are left with very little indication.
  • Pump Failure Shutdown – RPM is likely the easiest method of setting up an emergency pump shutdown routine.  Most motherboards and bios tools have some ability to set a minimum RPM level for the CPU fan header.  While this was originally intended to serve as a failsafe for CPU heat sink fans, it also works for pumps that have RPM sensing abilities.  While the D5 series may very well be the most trusted pump in all of watercooling, it’s always good practice to have a failsafe.  Running two pumps in series can give you redundancy, but what happens if you have a single pump and the pump fails…bad things can happen.  I personally have had an instance where I was working on my case and accidentally bumped a loose molex connector only to have the pump quit working.  My 8800GTX video card loop actually melted the acetal in my VGA block and the tubing had deformed to the point that it was nearly ready to burst.  Luckily I was right there looking and noticed the water boiling in my loop after getting a sense of some odd smell.  Had the tubing burst, surely there would have been disaster.  Having had the pump on a shutdown routine, would have prevented that.

Soo…I’m extremely happy that Koolance has now provided us with a D5 Vario WITH RPM sensor!  That’s awesome!

With that, let’s look around the pump.  It does come with a nice steel base which lifts the pump off the ground.  This metal base is ideal for sitting on a decoupling material since there will be no issue with heat.

1/2″ barbs come factory, no need to install a top for larger tubing

Now, let’s have a look on the inside, first and overall parts picture:

Tool-less disassembly reveals the goods, Ceramic/Carbon ball bearing, spiral volute

Another nice features of the PMP-450 is the completely tool-less ability to take the pump apart.  The pump is held in place by the large ribbed collar which simply unscrews with the twist of the hand.  The large o-ring you see sits in the volute housing and seals the metal pump housing to the thermoplastic volute.

At the heart of the pump and common to all D5 & DDC pumps is the very desirable ceramic ball bearing which mates up with a graphite impeller bearing cap.  The one point is the only point of contact and wear and makes for an extremely long life.  I have yet to see one wear out unless someone accidentally ran one dry.

The other perhaps not so obvious feature that makes a very large impact on the pump is the metal pump housing (Canned Spherical Motor).  There are two benefits to this.   One is the cooling capability and heat transfer that the metal housing provides.  Unlike it’s brother DDC series which uses a plastic housing and resulting heat buildup, the PMP-450 and it’s metal house serves extremely well to watercool the pump.  This does lead to more heat entering the water, however the cooling ability is beneficial to the pump electronics in keeping it cool.  The other benefit to this canned housing is how the metal canning creates a water tight seal around the motor housing.  While the pump is mounted any failure in the o-ring or other possible leak will generally have a very difficult time ever finding it’s way into the electronics of the pump.  I think it’s the above two reasons that make this pump one of the most durable water cooling pumps on the market.  They are water cooled, and have built in leak protection.

Easy assembly

So the pump overall has some real durability enhancing features, it has an RPM sensing wire, and very easy to take apart and clean for maintenance needs.

About the only downside I can think of is the larger size and lack of decoupling pad.  Some folks also have noted that the barbs are slightly over-sized and take a little more force to install tubing on.  I consider oversized barbs a big benefit in general because it generally leads to much lower chance of leaks.  Also decoupling material such as a piece of egg crate works perfectly fine.  I just wouldn’t recommend bolting the metal stand to the case if possible.

12V Test Results

Detail (Retest Done 12-15-12)

Koolance-PMP450-PQdetail12V

Following my usual pressure vs. flow rate testing, I came up with the following family of curves at 12V.  Generally settings 4 and 5 would be good options for average to higher restriction loops and settings 2 and 3 for very low restriction loops.  Setting 1 is really a bit too underpowered to maintain acceptable flow rates, although I would encourage anyone to try.  Note that my setting 5 is actually max and setting 1 is min.  I figure anyone that is operating at 5 likely has the knob turned to the max which is very slightly more than 5, etc.

Setting 1 through 5. Pressure is the solid lines, Watts is the dashed lines.

12V vs 24V

You may have noticed that the pump is perfectly capable of operating at higher voltage up to 24V.  This may lead you to believe there would be a significant performance difference between 12V and 24V.  I tested that below:

Setting 5 12V vs 24V, very very minor benefit with low restriction loops

Unfortunately there is not much benefit to using 24V on the PMP-450 pump.  Up to about 1.5GPM there really was no measurable benefit, on the contrary because it was consuming about 1-2 watts more.  I would not recommend purchasing a controller to operate this pump beyond 12V, it’s just not enough difference to bother with.

Performance PMP-450 vs PMP-400 + Top

While tops don’t help much on the PMP-450 pump because the factory top is already very good, they do help a lot on the PMP-400.  Soo…many have folks choose the PMP-400 for it’s slight performance advantage.  Here is that comparison:

The difference here will not add up to much temperature difference, but the PMP-400 with top is a slightly stronger pump for our more restrictive water cooling loops.

Size

One thing you should also consider with this pump is it’s relative size.  It is generally bigger than the PMP-400 or other DDC pump at least until you put on a top and lift the PMP-400 to provide more cooling.  Then they are comparable.

Size Comparison

On the left is actually the 450S model, but both (450 & 450S) are the same size so I reused the picture.  Without the lifting base, the PMP-400 is a fair amount more compact.  Size is something you’ll want to consider.  Also note that the inlet port and outlet ports are reversed between the two pumps.  The PMP-400 with top accepts the in from the top, where the PMP-450 accepts the in from the side.  Depending on your tubing configuration, you might have a preference one way or the other.

Noise

The pump is extremely quiet, particularly when installed in an acetal aftermarket top.  Check out the noise data in my pump noise round 1 piece where I tested both the stock top and after market top.

Efficiency

Not a huge deal as I think the differences are relatively small, but the PMP-400 when coupled with an aftermarket top will produce slightly more pumping power per watt than the PMP-450, but that is only after the PMP-400 has had the factory top (with an inlet elbow) removed.

Not quite as efficient as a PMP-400 with top

This is pretty minor when you’re talking about 20 watts worth of heat, but something to consider if you’re looking at running something extreme like triple pumps in series.  The PMP-400 with top is a bit more efficient by a few watts depending on the restriction.

Conclusion

Pros
  • Extremely reliable long lasting pump
  • Canned metal housing protects electronics from leak damage
  • Canned metal housing cools the pump motor very well
  • Factory Top performs very well, no inlet elbow
  • Factory Top comes with 1/2″ barbs
  • Koolance brand includes an RPM sensor wire, yes!!
  • Factory speed controller built in (no need for voltage controller to reduce speed)
  • Easy tool less entry
  • Cost – When compared to a PMP-400 plus top
  • Very Quiet
Cons
  • Larger in size
  • Not quite as powerful as a PMP-400 with top
  • Not quite as efficient as a PMP-400 with top
  • No decoupling pad or accessories (bare pump)

So there are some give and takes when compared to the PMP-400 series, but you’ll find the user base very much split out there.  I believe the durability, long life history, and cool operation are all very desirable features many prioritize highly.  I like this pump very much, and particularly like it now that Koolance has provided us with the RPM sensor.  You really can’t go wrong with either the PMP-450 or PMP-400 pumps, I use them both myself and can’t really pick a favorite because I see benefits in both models.  The nice thing about the PMP-450 is that you get a speed controller and a good 1/2″ compatible top factory out of the box.  You also get a pump that runs very cool and has a long history of reliability.  These are all very good qualities and I highly recommend it.

Cheers!
Martin