Case airflow/coolant temperature in a Switch 810

muffles

muffles

muffles

Associate
Joined
21 Jun 2012
Posts
158
Hi guys - so I've got the setup as shown here: http://forums.overclockers.co.uk/showthread.php?t=18437222

The front fans (2x AP-15s) are intakes, on push into the front rad, and the bottom fan is also an intake (just to get a bit more air into the case).

The top fans (3x AP-15s) are exhaust, and on pull on the top rad, with the rear fan also an exhaust.

I've got an in-line coolant temperature sensor installed (on the input to the res, if it matters) and have been messing about running GPU stress tests to see what temperatures I get. Based on the startup figure I've seen I think my ambient temperature is ~25C give or take a few Cs.

Normally it is hovering around the 29-30C mark but this is not under load, this is minimal use (browsing etc). When I run the stress tests I can get it up to ~34C, and have even seen 38C with FurMark (seems to stress it out more than Heaven? But I am no expert on this stuff). This is with 15 minutes on each, and I wouldn't rely on me having the "right" settings to correctly test (although I tried!).

So with the usual aim for a 10C delta-T I am having a look at what might be causing this. My actual question therefore is - am I causing a problem with my current fan directions? The top rad may not be getting as cool air as it could. I'm thinking of reversing the rear fan so it's also an intake.

I know I could just try this out but the way it's installed means I have to dig around with the cable management to turn it round, so a bit fiddly. Hoping to get some feedback here first!
 
Interesting, I can actually change that more easily :)

But will that be OK with just a single exhaust? I presume that having that many intake fans causes the hot air inside the case to flow in the direction of least resistance - i.e. the single exhaust fan, and any other vents etc in the case?
 
Temperature wise they seem OK, I've done the GPUs (crossfire) more and they peak at 57/51 on FurMark and got to something like 53/48 on Heaven (both 15 minute runs).

CPU I can't remember exactly (haven't done much on that, just a short IBT run), think it was peaking at something like 55 - it's a standard 2600K (for the moment, would like to overclock but I'm still learning).

So those seem OK, it's just the water temp that is confusing (worrying?) me.
 
Just tried the 3 top AP-15s in push - to be honest, results seem similar. 15 minutes of Heaven and the GPUs hit 51/47, with a water temp that is a little higher - 36, touching 37 (water temp at beginning of test was 30 so a couple of degrees higher than previous test, which could explain it).

Does anyone else have a water temp sensor and can confirm the readings/specs they have? I'm a bit confused as my GPU temps seem normal, it's just that my water temp seems to exceed this target 10C delta...

I have an Aquacomputer inline sensor, it says it's compatible with many fan controllers albeit does not list mine (Aerocool Touch 1000). Can't imagine that's the problem though, surely the sensor readings are a general standard.

Also I have my D5 on setting ~2.5 out of 5, from what I read that is suitable but could also look to adjust that I guess.
 
Here's a thought: I have 3 other temperature inputs on my fan controller, and 3 "tape on" style temperature sensors. Actually, I have all these plugged in now (it's annoying - without the temp sensors plugged in, the fan controller won't adjust fan speed :().

Can these be used to measure air temperature (unsure if it only works when on to something)? If so, where is the best place to locate it to measure the air temperature?

Interesting point: the 3 tape on style sensors are currently just bunched up behind my fan controller. Not sure if this shows anything, but they are reading around the same temp as the water (31-32C depending on the sensor, water is currently reading 31C). Odd? Maybe it's hot behind there?
 
If you're happy with your cpu/gpu temps then I wouldn't worry too much about water temp. You are running 2 gpu's afterall so 10c delta is probably more difficult to achieve.

Are you running the fans full speed? And your pump?
 
Last edited:
Yeah, I guess - the bit I'm confused about is why my water temp would rise so much, but the GPU/CPU temp doesn't. Would imagine one should affect the other, and if my water temp was high that would indicate high GPU/CPU temps should be present too.

I think I'm looking for some WC guru to step in and say "it's fine, that's a normal water temp and we all get it"... :D

The GPUs are only standard 6950s (albeit MSI ones so a bit overclocked from the factory), and I've got a 240 and 360 rad (80mm thick) so I thought I was well within limits for keeping them cool.

I've got my fans running at 1000rpm (Scythe GT 1850s) and my pump is a D5 at ~2.5 (out of 5 - not sure what RPM it equates to). I assume that gives me reasonable flow rate but I'm slightly regretting not shelling out for the flow sensor as well now :(
 
P.s. in looking into the line of thought that the sensor might be at fault, I've been trying to identify the spec of the sensor I have, and what I require. Through the process of elimination ;) I have determined the sensor I have has a 10k Ohm thermistor (it states it can be used on a variety of fan controllers as mentioned, and one of those mentions that spec). I can't tell what voltage it is running at though, which I understand is a factor (5v or 12v).

I've sent an email to Aerocool to ask them to confirm the spec required to connect a temp sensor to the fan controller, will see what they say.
 
spixelspixel said:
You should try setting fans + pump to max and retesting :)
Click to expand...

Is that how you would normally test your temps? Don't mind doing it but probably tomorrow - I have to dig in the case a bit to change the pump speed (fan speed from the controller of course). Will update on that test tomorrow :)
 
I got bored, so did the test now :p

With full fans and pump, and 15 minutes of Heaven, I get 48/44 on the GPUs and a water temp of 32, so definitely better (that's about a 7C delta T although I don't know the exact ambient temp). That's a couple of C lower than before (and more on the GPU temps itself).

Should I always have been doing this testing on full fans/pump then? If so, noobish error on my part then :D
 
muffles said:
I got bored, so did the test now :p

With full fans and pump, and 15 minutes of Heaven, I get 48/44 on the GPUs and a water temp of 32, so definitely better (that's about a 7C delta T although I don't know the exact ambient temp). That's a couple of C lower than before (and more on the GPU temps itself).

Should I always have been doing this testing on full fans/pump then? If so, noobish error on my part then :D
Click to expand...

Now do them individually to see if its the fans or pump increase which is giving the most benefit :D My guess would be that pump setting was a bit low considering the amount of blocks/radage you have and I didn't notice much difference in my cpu temps with my fan on 1000rpm vs 2000.
 
51/47 GPUs and 36 water temp with just full pump...I think the 36 water temp is perhaps because my ambient is up a degree or two. Compared to full pump & fans the GPU temps are up ~3C or so.

Going to try with pump on 3/5 (it wasn't 2.5, it was 3 all the time btw) and fans on full now :)
 
Full fans... 48/44 and 32 water temp....so I think it's actually the fans that make the difference :)

My full results (for Heaven) are then:
* Low fans/low pump: 53/48 GPU and 34 water temp
* Low fans/full pump: 51/47 GPU and 36 water temp
* Full fans/low pump: 48/44 GPU and 32 water temp
* Full fans/full pump: 48/44 GPU and 32 water temp

These are all with my top fans in pull, looking back at my results I saw 51/47 GPU and 36 water temp from putting the top fans in push. May investigate that later as that was with low fans/low pump and seemed to gain a couple of degrees on the GPU temps (albeit with a couple of degrees up on the water temp!).

Either way I think I need to do some repeated tests to make sure the results I'm getting are consistent.
 
You must log in or register to reply here.
Back
Top Bottom