pa120.3 shroud + fan
- Thread starter pete
- Start date
I don't use a shroud on my 120.3
Marci from over-clock.co.uk is definately the person to speak to about these rads. Loads of great information on there regarding these radiators.
I'm currently using Yate Loon 120mm fans connected to a fan controller for quietness, but did have Sharkoon 120mm 2000's on there.
Marci from over-clock.co.uk is definately the person to speak to about these rads. Loads of great information on there regarding these radiators.
I'm currently using Yate Loon 120mm fans connected to a fan controller for quietness, but did have Sharkoon 120mm 2000's on there.
name='pete' said:
I use a shroud and a PA120.3 rad. My fans operate in 'push' mode. The shroud reduces the noise of the airflow as it hits the rad and it also eliminates the deadspot i.e. the area of reduced airflow around the centre of the fan. As far as I know the shroud makes minimal impact on the points I've mentioned above where the fans are in 'pull' mode.
name='Mullet' said:
im sure Marci will shed some light,
i found this on XS,
http://www.xtremesystems.org/forums/showthread.php?t=141129&highlight=push+pull+shroud
kemp plz remove if if not allowed
Read the "suck or blow" thread in this forum... should explain the situations where a shroud is beneficial, and those where it isn't...
Blindly getting Yate-Loons without checking your heatload that you need to cool could land you with undesirable temps. See http://www.over-clock.com/ivb/index.php?showtopic=20277
Blindly getting Yate-Loons without checking your heatload that you need to cool could land you with undesirable temps. See http://www.over-clock.com/ivb/index.php?showtopic=20277
hi,
thks for the links but have already have read the 2 threads marci.
the options i think that i have are that most w/c have yate loons with similar systems specs as myself.
or that i go for more expensive, high cfm fans i.e like panaflo to give me higher degree of play temp wise and use a fan controller.
buy 2 shrouds and try for myself to see which works best for me
any info is appreciated
thks for the links but have already have read the 2 threads marci.
the options i think that i have are that most w/c have yate loons with similar systems specs as myself.
or that i go for more expensive, high cfm fans i.e like panaflo to give me higher degree of play temp wise and use a fan controller.
buy 2 shrouds and try for myself to see which works best for me
any info is appreciated
There is no other info available... anything else is conjecture or opinion. The links I gave you take into account all testing done on our PA Series rads from all sources.
Your two choices are, go with what is scientifically tested to be roughly correct (ie: the outcome of the links I gave you)... or go with "the trend" of slapping Yates on cos everyone else has, regardless of whether it'll cool the desired heatload satisfactorily or not, and then deal with the arising issues (high temps for watercooling) later.
Ie: Calculate what you need, buy that, and have a semblance of a guarantee of performance levels... or buy what you want, slap it on, and hope for the best with no guarantees other than "that's what everyone else is using".
Bear in mind when doing the calcs, if you're using both rads on a single loop then divide total heatload in half when working out what fans you need on the rads as the heatload will effectively be split between the two rads. (It won't be in reality, but that's the easiest way to approximate it for the sake of the calculations.)
The shrouds will make very little noticeable difference to things in terms of performance, other than reducing noise in some configs... the only reason we released them for sale is because we were asked to by a large percentage of the market as their use had been popularised in the US due to the common use of shrouds on heatercores (but that was because the core area was larger than that of the fan, similar to the PA160's necessity for using it's shroud), not because of any performance requirement. Ask yourself...
1 - Are they worth the additional cost?
2 - Are they worth the additional radiator depth thus making them very impractical to mount?
Personally, I'd disregard the shrouds altogether.
Mr. Smith
New member
name='Marci' said:
That was a very interesting read, apparently my OC'd q6600 and OC'd 8800 gts + NB have 750w of heat that needs dissipating.
My flow rate is approx 1.75 GPM.
Using the PA120.3 chart it indicates I need 100+ cfm fans to adequately deal with the load!
I happen to have 2 140cfm delta fans and one s-flex low cfm fan (no clearance for my other delta) on the 120.3. I have the fans on a controller and keep the deltas turned down to the min rpm the controller will allow, from 3200rpm to ~2000rpm.
Just from monitoring temps with fans on full vs fans on min I can see no benefit, perhaps 1deg after hours of prime. I attributed this negligible variance to the efficiency of the rad.
I planned to move down to some 'regular' 100x25mm 60-100cfm fans for a bit more peace and quiet.
I was leaning more towards the 60cfm due to the results from my crude 'testing' (monitoring temps during prime with diff fan speeds). Now I am unsure...
....maintain coolant temps within 10 deg C of ambient...
Lesser fans will simply mean coolant temp is greater than 10 deg C + ambient, but the radiator will still work fine. That's the key... no matter what you do, the radiator will cope fine, but temps will rise with less airflow (obviously), and will rise noticeably when airflow drops below that which is required for the hardware.
Now, with the whole heatload thing... The above linked method is still an approximation. Power consumption = Thermal Output is only a loose fit. In actuality, it's an over-fit.
IE: In reality, power consumption is always more than thermal output. (It's physically impossible to determine the ACTUAL thermal output of a PC component as you can't measure the temp in the correct place to the required amount of accuracy. Even Intel and AMD have the same problem. They use a TTV - basically a PC Simulator - to qualify heatsinks and cooling methods, not "real-world". The only people that refer to "real-world" testing tend to be end-users - to refer to "real-world" as a manufacturer effectively means guesswork and assumptions as the heatload can't be quantified. With a load simulator such as the TTV, the heatload CAN be quantified)
No PC is 100% efficient with it's use of power etc, so the actual heatload will always be a good site less than the calculated power consumption, but if you aim to cool the calculated then you end up with a good amount of headroom, which is more or less what you're seeing there...
Don't forget that 1 degree temp change equals more than it seems... generally a drop in coolant temps of 1 degree by increasing airflow can be seen as the cooling system being capable of achieving a further 50w of thermal capacity. A 1 degree rise in coolant temps can be loosely seen as the system loosing 50w of thermal capacity.
These are all just approximations at the end of the day, but the basis is solid testing data. I prefer folks overshoot when it comes to the heatload vs airflow than end up under and thus seeing temps not much better than good air.
If you want to have fun abandon Prime, and use TAT to inflict powervirus load (altho tricky for quadcore owners as the quadcore version of TAT was withdrawn from public availability). THEN you'd see some larger temp differences between fans at low rpm and fans at fullblast...
Also, when loading under Prime, you're only loading the CPU... not everything that you have watercooled. It's rare to have EVERYTHING under full load at once, and quite difficult to achieve... however, ATITool with it's 3D Display running will put both CPU and GPU under full load if you're an ATI card user. Thus, using Prime to indicate the cooling system performance is a bit pointless as it only exhibits the effects of a single item in your loop being at full load. Even still, that doesn't put the northbridge / southbridge under full load...
So, your power consumption may be 750w max with EVERYTHING at full load, but in reality this is never achieved. THIS is why "real-world" tests produce different results largely to empyrical testing. With a rig with 750w of power consumption, there's probably only around 500w AT MOST being produced as thermal output by the items that you actually have cooled at any given moment in time, despite the rig potentially being able to create 750w of thermal output if it could all be put under 100% load simultaneously.
Mr. Smith
New member
Thanks for taking the time to reply.
I overlooked the coolant temp - doh. I simply monitored core temp and did not even go so far as to take coolant temps (i said i was being crude!).
I can try tat or prime again with rthdribl to load the cpu and gfx and use coretemp and riva to monitor the temp diff with fans min/max and see the diff. Anyone know of any 80cfm fans?
P.S Apologies Pete for hi-jacking your thread
I overlooked the coolant temp - doh. I simply monitored core temp and did not even go so far as to take coolant temps (i said i was being crude!).
I can try tat or prime again with rthdribl to load the cpu and gfx and use coretemp and riva to monitor the temp diff with fans min/max and see the diff. Anyone know of any 80cfm fans?
P.S Apologies Pete for hi-jacking your thread
