Question about H100 function

[trents]

What should I expect from my Corsair H100?

System basics: 2500k oc'd to 4.7 on about 1.45 vcore under load. Core temps are hitting mid 80s c.

When I feel the hoses on the H100, neither one feels warm as I would expect if one was carrying coolant from the pump that was mounted on a processor whose core temps were exceeding 80c. My infrared thermometer reports one of the hoses to be just under 28c and the other about 3-4c cooler. I've checked the TIM contact pattern and there is essentially 100% contact between pump block and CPU lid.

Is this normal behavior for a water cooler of this caliber?

 

[EarthDog]

AIO's are not like car rads where there is a thermostat there essentially separating the hot and cooler liquid. It is a free flowing and small loop. If you can notice a big temperature difference between hoses, there is likely something wrong with the system (slow flow rate?).

As far as performance goes, there are a couple tests I recall with H100 and that CPU around...

Short answer, yes.

 

[trents]

AIO's are not like car rads where there is a thermostat there essentially separating the hot and cooler liquid. It is a free flowing and small loop. If you can notice a big temperature difference between hoses, there is likely something wrong with the system (slow flow rate?).

As far as performance goes, there are a couple tests I recall with H100 and that CPU around...

Short answer, yes.

Ah! No thermostat. That makes sense now.

 

[GTXJackBauer]

Plus these AIOs come with weak pumps. I remember someone doing a mod and showing us the flow of his AIO which I believe was a H100. It was a trickle flow effect. Very slow moving.

 

[RnRollie]

H100 *should* perform better than that
however, since it's not really automatic speed control... you should play around with the pump/fan speed - there are 3 settings

Put you finger lightly on the top/side of the block, you should feel some (faint) vibration... if there isn't, then there is a problem... like pump not running.

It its "bangin' away" then its probably airlocked, move the rad/case around abit

Then again, maybe its working as it should, but maybe you should ease back on the O/C and/or voltage... the H100 barely outperforms a good Air HS.

 

[trents]

H100 *should* perform better than that
however, since it's not really automatic speed control... you should play around with the pump/fan speed - there are 3 settings

Put you finger lightly on the top/side of the block, you should feel some (faint) vibration... if there isn't, then there is a problem... like pump not running.

It its "bangin' away" then its probably airlocked, move the rad/case around abit

Then again, maybe its working as it should, but maybe you should ease back on the O/C and/or voltage... the H100 barely outperforms a good Air HS.

Changing the button on top of the pump from high to low seems to make about 2c difference in temps but it's hard to know if that's due to an actual change in pumping force or fan speed since that also quiets the fans down. There is vibration when I touch the pump but there is also vibration when I touch a lot of other components such as the ram or a sink. This could be from the the fans attached to the test bed frame.

As you can see, the flat surface of the radiator is currently in a vertical position and slightly higher than the pump. I think I will try changing the attachment so that the radiator is more overhead and flat. It could be air locking in its present orientation.

Thanks for the input.

 

[EarthDog]

It shouldn't be 'air locking' like that (id imagine only with the in/out up top that could happen?). I had that same exact orientation inside a HAF XB with no issues.

 

[Tech Tweaker]

This sounds like normal operation to me.

Back when I had my H100 its tubes never got noticeably warm either. I would always just feel a slight vibration on the tubes, that would tell me there was some kind of flow going through them.

Also, the control switch on the top of the pump is only for fan speed. The pump speed should remain constant no matter what setting the switch is at (low, medium, high). Pump speed should be fairly constant at somewhere between 2000-2300RPM.

 

[GTXJackBauer]

Finally found the video from someone's mod. Made a thread about the flow of the H100 a while back.

Thread

 

[EarthDog]

That looks, ehh, broken.

 

[trents]

That video is weird! Looks like he's done away totally with the radiator and is using "open air" water cooling. If the point of that video is to demonstrate that the Corsair hydro pump is weak I'm not sure it accomplishes that purpose. The outlet orifice looks way bigger than the hose diameter and that would "hose" (excuse the expression) the pressure such that you only get a trickle.

 

[Johan45]

As others have said Trents that temp differential seems normal to me. The only way I managed to get my AIO to perform well was to strap 4 really good fans to it.

 

[EarthDog]

Well, flow rate is flow rate no matter how much pressure (which is all the size of the tubing would change). So long as it is above 1GPM, then there are only minor improvements to be had regardless.

 

[trents]

So here's a pic after a stress test. Just wanted to confirm these temps are appropriate with given voltages and given cooling solution.

 

[EarthDog]

Seems about right assuming appropriate ambient temps.

 

[trents]

Ambient temps are 21-22c.

 

[Mandrake4565]

Trents, If memory serves me correct using my True Ultra 120 with 110 cfm fans on it, the temps on my 2500k @ 4.7 were in the mid 80's at 1.43. I know it's not apples to apples but those temps look about right. My ambient temps are around the same as yours.

 

[sobe]

The vcore is what's killing you IMO without going for a custom loop, TEC, or a phase change unit. Can try lowering that to 1.40 and see if there's a temp change. I would see if you could get away with a lot lower vcore running at 4.5GHz, just to see. I didn't run a 2500k, but my 2600k only needed 1.33v @ 4.5GHz, 1.45v seems a bit high to me....

I saw 70c under Prime95 on a Coolermaster Hyper 212 Plus on that 2600k, which is why I say see about dropping the vcore.

 

[EarthDog]

I wouldn't go any higher for 24/7, that is for sure...try lowering it, certainly and see if it works. Same game here with Intel, lowest stable voltage.

 

[trents]

The vcore is what's killing you IMO without going for a custom loop, TEC, or a phase change unit. Can try lowering that to 1.40 and see if there's a temp change. I would see if you could get away with a lot lower vcore running at 4.5GHz, just to see. I didn't run a 2500k, but my 2600k only needed 1.33v @ 4.5GHz, 1.45v seems a bit high to me....

I saw 70c under Prime95 on a Coolermaster Hyper 212 Plus on that 2600k, which is why I say see about dropping the vcore.

Check the pic again, it's not 1.45 it's 1.432v for the max. That is the lowest stable voltage. If I lower the offset one more .005 increment in crashes before completing one iteration. Got a 3570k coming in a few days. We'll see what I can get away with on that one. Should run cooler since the fab process is smaller. Hoping for about 4.6 on the 3570k.