Water temperature sensors

M

mizer357

Limp Gawd
Joined
May 12, 2014
Messages
136
The whirring up/down of my fans based on CPU temps has been bugging me, and I'd like to switch to using water temperature to control fan speed. I'm looking at G1/4 fittings, either in-line in the loop, or swapping one of my stop fittings for one with a temp sensor. My questions are:

1) Where in the loop are these sensors ideally placed? Just to keep things simple and if it's all the same, I'd like to put the stop fitting in my reservoir.

2) Is an in-line sensor 'better' than one integrated into a stop fitting, or vice-versa, or does it not matter?

Examples:
In-line:
71170_0.jpg

Stop:
71006_0.jpg


Thanks.
 
I would think the inline would be best, since you're always getting flow past it. With an inline sensor, placement isn't as important; the whole loop equalizes to within a degree or two. I've got one of those phobyas you pictured and tried what you're talking about, and found some things.

1.) Your water never gets as hot as your components. The temperature delta on your coolant from idle to full load is fairly narrow, so your temperature control range will be tight.

2.) To use one of those, you'll either need a temperature sensor header on your motherboard or a fan controller with the same.

3.) If you use a motherboard header, you may need to use software like SpeedFan to get it to control the fans based on the input from that sensor. (You'll probably want to use SpeedFan anyway, because even if your mobo can control fans based on temperature sensor input, it's likely built around an air-cooling control range.)

These things are a hassle. What I eventually did was just go to SpeedFan and still base fan speeds on the CPU, but with a much less aggressive fan profile. Fan speed on rads has less effect than you might think on cooling, so you can keep them pretty low most of the time (unless you're chasing blistering overclock performance.)

Hope this helps!
 
I use an older NZXT fan controller that has a temp sensor for each fan output. I placed those sensors in the radiator fins near the fan it goes with. As the surface temp of the radiator increases, so does the fan speed. Fans tend to stay below 40%, only occasionally rising to 50%.
 

Thank you so much! That was insanely helpful.

You brought up something I'm concerned about regarding the temperature delta with water temps. I'm currently using an Aquaero 6 as my fan controller, and I'm comfortable with setting fan speeds based on min/max temperature variables for CPU and GPU. I would imagine a 1 degree rise in water temperature could mean an exponential increase in CPU/GPU temps, and was curious what typical min/max temperature values were for safely gauging water temperatures. Are there generally accepted rules, or is this typically just a trial, check and re-check thing?

Based on your advice, I'm also wondering if I should skip this exercise altogether, and keep the fan speed CPU/GPU temperature dependent and just mess with my fan curves as you did.

In any case, thanks again.
 
mizer357 said:
Thank you so much! That was insanely helpful.

You brought up something I'm concerned about regarding the temperature delta with water temps. I'm currently using an Aquaero 6 as my fan controller, and I'm comfortable with setting fan speeds based on min/max temperature variables for CPU and GPU. I would imagine a 1 degree rise in water temperature could mean an exponential increase in CPU/GPU temps, and was curious what typical min/max temperature values were for safely gauging water temperatures. Are there generally accepted rules, or is this typically just a trial, check and re-check thing?

Based on your advice, I'm also wondering if I should skip this exercise altogether, and keep the fan speed CPU/GPU temperature dependent and just mess with my fan curves as you did.

In any case, thanks again.
Click to expand...
If there's any mathematically reliable correlation between component and water temperatures, I didn't find it; I was looking for the same thing. Bear in mind I'm kind a newbie to watercooling, so take my advice for what it's worth; I just tried to do the same thing so I thought I'd chip in. =)

If you go the SpeedFan route, I think there's a neat thing you can do: you can get your fans to react to the highest of the two temperatures between your CPU and GPU. I only think you can do that; I'm still tweaking my setup and learning it. SpeedFan is powerful, but not very user friendly. I highly recommend JayzTwoCents' video on configuring SpeedFan (and, incidentally, his channel in general for watercooling.)

Let us know how it turns out. =)
 
VanGoghComplex said:
If there's any mathematically reliable correlation between component and water temperatures, I didn't find it; I was looking for the same thing. Bear in mind I'm kind a newbie to watercooling, so take my advice for what it's worth; I just tried to do the same thing so I thought I'd chip in. =)

If you go the SpeedFan route, I think there's a neat thing you can do: you can get your fans to react to the highest of the two temperatures between your CPU and GPU. I only think you can do that; I'm still tweaking my setup and learning it. SpeedFan is powerful, but not very user friendly. I highly recommend JayzTwoCents' video on configuring SpeedFan (and, incidentally, his channel in general for watercooling.)

Let us know how it turns out. =)
Click to expand...

There is a mathematical relationship, but it's complex and you need to know a lot about the specific heat transfer characteristics of the components, your loop, the ambient temp, and you need to use differential equations to track the two.

You're correct in your earlier statement about the control bands. In a decent loop, CPU temps might range from 25-60C, but over that range the water temp would only go from maybe 25-30C.

mizer357 said:
Thank you so much! That was insanely helpful.

You brought up something I'm concerned about regarding the temperature delta with water temps. I'm currently using an Aquaero 6 as my fan controller, and I'm comfortable with setting fan speeds based on min/max temperature variables for CPU and GPU. I would imagine a 1 degree rise in water temperature could mean an exponential increase in CPU/GPU temps, and was curious what typical min/max temperature values were for safely gauging water temperatures. Are there generally accepted rules, or is this typically just a trial, check and re-check thing?

Based on your advice, I'm also wondering if I should skip this exercise altogether, and keep the fan speed CPU/GPU temperature dependent and just mess with my fan curves as you did.

In any case, thanks again.
Click to expand...

If you're already dumped the $180 into an AQ6, then you might as well pick up a couple sensors. If I were you, I'd put them in and start running some tests. Make a couple tables and see how your water temps compare to the CPU/GPU temps. Once you have that data, you can tweak your fan curves to compensate. There's no easy way to tell in advance what you're going to see in your particular system, but since you have the sensors, you can easily collect the real-world data and adjust based on that.
 
Thanks for that guys. I imagine system-specific variables like water volume would probably play a bigger role in temperature variances than other, more traditional kinds of setups, and would make hard/fast rules hard to come by.

Do you recommend more than one sensor in the loop?
 
mizer357 said:
Thanks for that guys. I imagine system-specific variables like water volume would probably play a bigger role in temperature variances than other, more traditional kinds of setups, and would make hard/fast rules hard to come by.

Do you recommend more than one sensor in the loop?
Click to expand...
From what I've read, if you're sensing coolant temperature, one sensor is plenty for a simple serial loop; your coolant gets to a homogenous temperature pretty quick unless you've got really low flow.
 
I have a stop/plug type in my reservoir (using a single sensor). The highest delta I've measured from coolest to hottest water temp is right around 9.5°; idle delta T of .8° with pump and fans at max RPM and max of 10.3° with CPU and GPU at full load after two hours with pump and fans at max RPM. Typical use case fluctuates from about 4.5-6.3°.
 
mizer357 said:
Thanks for that guys. I imagine system-specific variables like water volume would probably play a bigger role in temperature variances than other, more traditional kinds of setups, and would make hard/fast rules hard to come by.

Do you recommend more than one sensor in the loop?
Click to expand...

If you've got an AQ6, I'd get two sensors unless cost is a factor. One "hot" and one "cold", so put one at the hottest part of the loop (exit of the waterblocks), and one at the coldest part (exit of the radiators). That'll give you the most "useful" data, but don't expect to see much of a delta unless you're in a very low-flow loop. I can't imagine a useful reason to have more than 2.
 
I'm looking through available fittings, and was wondering if you guys had thoughts on the following:

Is there any reason this style of fitting, with the extended sensor:
41L2IBljD3L.jpg


would be better than these, shorter versions that look more like traditional stop fittings?
51KPHuny4uL._SY450_.jpg


Ideally, I'd love to just replace a couple of stop fittings to get water temps and the second model seems it would be less restrictive on the loop.

Thanks!
 
I'm using the Monsoon plug that is similar to the second one you have posted and have no complaints.
 
The way you want to set it up is a temp sensor in the water and another of ambient air. Aquasuite lets you set up a "virtual sensor" that will reflect this delta temp.

I have the inline aquacomputer ones, they look just like the phobyas.

I have a sensor at the outlet of each of my rads. One rad feeds the gpu and the other feeds the cpu. The deltas for each (between air and water) control the fans for each rad.
 
Last edited:
sprout2287 said:
The way you want to set it up is a temp sensor in the water and another of ambient air. Aquasuite lets you set up a "virtual sensor" that will reflect this delta temp.

I have the inline aquacomputer ones, they look just like the phobyas.

I have a sensor at the outlet of each of my rads. One rad feeds the gpu and the other feeds the cpu. The deltas for each (between air and water) control the fans for each rad.
Click to expand...

Thanks! Measuring the air temp as a baseline completely makes sense. I really appreciate the tip. As timing would have it, I got my sensors (Monsoon) this morning and just got them in the loop.

Sensor 1, at the bottom of my reservoir:
sensor1.jpg


Sensor 2, on the second GPU block, above the outlet:
sensor2.jpg


I'll set up a third ambient sensor this evening and do some testing.

Thanks all!
 
I run my ambient air temp sensor at my towers intake. The water temp sensor is at the end of the loop in my reservoir.

 
I wanted to report some observations so far:

Water temp sensor 1, pre-blocks, and water temp sensor 2, post-blocks, are pretty much the same. Sometimes there is a fluctuation of 0.1C, up to 0.3C at the most). If anyone is thinking of trying this with a single loop, you should only need a single water sensor.

Under extended load, there is a difference of about 12-13C between ambient temperature and water temperature.

I have my fans at my preferred min power for a delta value of 0, and as high as I want to go at a delta value of 10. There is a slight curve to my fan speed profile. All this seems to have addressed my initial issue of my fans constantly whirring up and down when based on CPU temps. I'm probably going to be tweaking settings as I go, but looking good so far.

Thank you all for your help!
 
good deal. I see a 2-3 c delta between my rads (2x120 and 3x120). One feeds the gpu block and the other the cpu. Pretty close to what you are getting. I'm running 56mm thick rads too fwiw.

I like to use the rad output to judge the efficiency at the given fan speed (what I can control). Block outlets will work too, just different from what I picked.

12-13c is a decent delta under load. I think I have mine capped at 15 for max speed (about 50% on the fans). I haven't been through a summer yet with it. We'll see.

If you haven't read it yet, there's a super long thread over at overclock.net. I recommend either using it as reference (searching) or reading it in your spare time. I read the whole thing, and got a ton out of it.
 
Last edited:
sprout2287 said:
good deal. I see a 2-3 c delta between my rads (2x120 and 3x120). One feeds the gpu block and the other the cpu. Pretty close to what you are getting. I'm running 56mm thick rads too fwiw.

I like to use the rad output to judge the efficiency at the given fan speed (what I can control). Block outlets will work too, just different from what I picked.

12-13c is a decent delta under load. I think I have mine capped at 15 for max speed (about 50% on the fans).

If you haven't read it yet, there's a super long thread over at overclock.net. I recommend either using it as reference (searching) or reading it in your spare time. I read the whole thing, and got a ton out of it.
Click to expand...

Awesome. Thanks for your observations and the tips.
 
You must log in or register to reply here.
Back
Top