Hello All Just wondering what speed I should be runnig my Swiftech pump? It is the variable one so I can choose to have it running very slow, very fast or somewhere in the middle. What is better for cooling? Should the water be running slow enough so the radiator can do its work, or should it just be running the water around as fast as possible? Suggestions please. Thanks DJ
That is true to a certain extent however if it ran at a billion ltrs a second it wouldnt be any better. Water has certain thermal properties. As soon as the reach the wall of this property you will see no difference and in some cases a decrease in performance as the water moves to fast to remove the heat from the block. I would increase it in stages and run it for the day. Take note of min and max temps along with averages. Then look over the week to see which is best and go with it. Andy
its subjective to the loop its in. If you have loads of 90 degree fittings, then the flow rate would die off, so to compensate for it, an increase in speed of the pump would raise the flow rate again. Radiators are the biggest constriction of speed anyway. If the pump is making noise at full speed then its either the following cavatation, aeration, or vibration. the deisgn of the them is so that is generates mininiumal amount of noise when in perfect operation. Speed control is a pointless effort as it would only increase the pump life by a small amount.
For me, I run mine (MCP655) at 5 all the time because I can't hear it anyways (no really, it's silent - I can stop all my fans and mess with my pump settings and I can't tell the difference) and that much more flow isn't a bad thing.
Radiators have the highest flow rates of any loop component, barring reservoirs of course. Blocks will ALWAYS be the most restrictive part of a loop. From what i've seen/heard the MCP655/Laing D5 Vario usually reaches a performance wall around setting 3, unless you have a particularly restrictive loop.
Errr.... incorrect. Read the sticky, but fundamentally while you get diminishing returns with flow (i.e. as flow rates increase, the improvement you get for every 1 l/sec additional flow decreases) increased flow rates will ALWAYS give improved cooling - although you may be unable to measure it.
Contradicting much? More flow DOESN'T always give better cooling, in fact most of the time the added pump heat dump totally NEGATES the added flow. Check out Martin's Liquid Lab results, from his testing it's pretty much widely accepted that anything over 1gpm is wasted, except for a couple of especially restrictive blocks (Koolance KL-350 in particular).
Not really - most people measure chip temperatures using a cruddy thermistor embedded inside the CPU. For higher flows, you really need a Platinum Resistance Thermometer, fairly high quality heating block and a decent resistance meter to measure the effect of flow. The point about pump heat is a valid one (IIRC Cathar actually did most of the original testing, although it may well have been replicated since). I should have explicitly stated that I was referring to the water-chip ΔT, rather than the system power. For high power pumps (and this is more true for old-style pond pumps than for modern watercooling pumps), then the ideal flow rate will depend on the radiator and waterblock combination. What I was addressing here was the prevalent myth that water needs somehow to slow down and be in contact for a period of time with the block to do any good. That's tripe.
Nope, very much correct! Water will eventually heat up regardless of whether or not it is taking heat away from something else. Energy transfer from water particle collisions will ensure this. Can't change the physics, it's inevitable. Andy
You want to watch your phrasing there - I had to read that three times before I realised you weren't trying to violate the Zeroth Law! There are frictional losses in a water cooled system - there have to be, or the net power of the pump would be zero. However, "water particle collisions" is a very, very bad way to think of it for a liquid or high pressure gas - the losses are through turbulence and boundary layer entrainment, which is rather more complex than simple collisions. It must be said, however, that you're nowhere near what will happen in a real system for the type of pumps typically used in water cooling systems. These have powers in the range of 5-10W, and therefore represent less than 10% of the total heat load on a system. Since the air-water ΔT is typically in the range 3-5°C for such a system (if well designed - if it's more than this, you want to look at increasing radiator area or air flow) then even doubling the pump power represents a temperature increase in the water (and hence by the Zeroth law the CPU) of significantly under a degree. On the other hand, temperature differences between the CPU and water are typically in the range 10-20°C. Hence, either the improved cooling caused by increasing flow needs to be very small, the pump power way in excess of that commonly used (old-style pond pumps will act like this) or the radiator needs to be very poor (e.g. passive radiators) to see the effect you are describing.
