My Thermalright HR-01 Experience (so far)

[Ken5d]

I bought the HR-01 (aka HR-01-K8), since it was a bit cheaper at jab-tech instead of the HR-01-775. I was told by Thermalright that it was the same heatsink as the HR-01-775, without the 775-specific mounting system. (That's the above-board, pin-based mounting system.) Since I also bought the LGA775 Bolt-Thru Kit (for under-board mounting), I had no need for the pin mounting system.

However, as it turns out, there is a difference between the HR-01 & HR-01-775. The heatsinks are identical, but the HR-01-775 version comes with a shorter fan duct (orange), vs the -K8 (purple). The purple one was very difficult to install in my SOLO, and I'm a bit afraid of the lateral pressure it must be applying to the heatsink. Though by using the bolt-thru kit (which was very easy to install, by the way), it sure won't be lifting up off of the cpu!

So, how are the temps? It's a E4300 (running at stock right now) in a SOLO case. I've got the fan duct installed (which lined up perfectly in the SOLO), attached to the chassis exhaust fan (a 120mm, 4-pin JMC PWM fan, running at about 630 RPM). There's no fan attached to the HR-01 at this time. I've also got 2, 92mm Yate Loons as intakes.

My ambient temp is 20C. At idle, I'm getting +7C (i.e., 7C degrees above ambient). At 100% load (orthos), it shoots up to +23C. For fun (and info), I also tested load with the JMC at 100% (which is quite noisy). It dropped the CPU temp by 5C (so, +18C above ambient).

Also interesting to me was PWM cooling. I was quite worried about the effect the fan duct would have on PWM (as well as NB) cooling. Turns out (so far, anyway), I needn't have worried. At idle, PWM Max was +31C (again, that's an increase of 31C degrees over ambient). At 100% load, it went up another 10C to +41C. Here's the interesting part. Recall that with the JMC at 100%, the CPU temp dropped by 5C. Well, the PWM temp dropped by 13C degrees! So there's no doubt that there's a definite cooling effect from the exhaust fan sucking up air off the bottom of the board (through the HR-01).

It didn't seem to help the NB as much (assuming that the "system" temperature reported by my board is actually related to the NB temp). In fact, running Orthos didn't change the NB ("system") temps, either. It was +10C at idle and load, and it dropped one whole degree C with the JMC at 100%.

I hope that makes some sense.

[haywire]

Ken5d:

Thanks for the detailed report of your HR-01/e4300 test results.

How are you measuring "ambient" temperature? Is that the board temperature reported by AISuite?

By PWM cooling, are you referring to plugging the fan into the CPU_FAN header that controls fan speed thru the BIOS? (sorry, I have ASUS mb)

How did you conclude that the exhaust duct is causing fan to pick up air off the bottom of the board?

When not relying on PWM, how are you controlling fan speed? I'm kind of new at this fan tuning business.

I don't think my Centurion 5 is providing very good airflow at this time (stock configuration, not modded), so I'm anxious to see if switching over to Antec Solo will help with that issue.

Thanks,
-haywire

[Ken5d]

How are you measuring "ambient" temperature? Is that the board temperature reported by AISuite?

First, a tiny correction. AISuite is Asus' software (I believe). This is a Universal-abit board. They call their stuff uGuru (as in micro-guru, but I don't know how to type that character.) Sorry for the correction, how could you know what abit calls it?

Anyway, nope. I measured ambient by a thermometer that's in my room. uGuru reports a "system" temperature, which may be the board temp, or it may be the NB temp (though as I mentioned, it doesn't seem to change, so maybe the board temp is more likely).

By PWM cooling, are you referring to plugging the fan into the CPU_FAN header that controls fan speed thru the BIOS? (sorry, I have ASUS mb)

No, I was referring to the cooling of the PWM thingies (that's a technical term). There are 5 of them on the QuadGT, apparently related to the fact that it uses a 5-phase power design. uGuru gives the temperatures of each PWM separately. In quoting a PWM temp, I was using the max for all 5 PWMs. I don't know if the PWMs make up the VRM (voltage regulator module) or not, but they're almost certainly related to it. Maybe someone who really knows what they're talking about will jump in and help correct my terminology (though it's abit that calls them PWMs, not me!)

How did you conclude that the exhaust duct is causing fan to pick up air off the bottom of the board?

That was an assumption from the observation that the PWMs (which surround the CPU socket on 2 sides) cooled significantly by increasing the speed of the exhuast fan (ducted to the heat sink). I couldn't account for it any other way (in my very limited experience) other than it must have increased air flow through the PWMs.

When not relying on PWM, how are you controlling fan speed? I'm kind of new at this fan tuning business.

I'm sure something I wrote mis-led you on that question, sorry. I think part of the problem is that the fan is a PWM fan, and the motherboard has PWM components. They aren't related, other than they're both referring to pulse width modulation.

To answer your question as best I can, though. The exhaust fan is always controlled via PWM. The 2 intake fans are voltage controlled.

I don't think my Centurion 5 is providing very good airflow at this time (stock configuration, not modded), so I'm anxious to see if switching over to Antec Solo will help with that issue.

I can't help you with that one, because I don't have enough experience to compare this with other cases. I really loved working with the SOLO, but I wish it had used 120mm intake fans instead of 92. With the filter (not to mention the wire grids that I have yet to cut off), the 92s aren't providing enough intake to give me a positive airflow (even with the 92s at high speed and the exhaust at low speed). But the combination of the HR-01 and the fan duct appears to work well in this case, at least based on my early experiences so far.

[Ken5d]

I've now attempted some overclocking, and I was quite surprised at the results. However, they are somewhat tainted, as you'll see.

The best I've been able to achieve thus far is a disappointingly low 55% OC, to only 2.79 GHz. I hope to get higher, but right now, I'm stuck at 2790.

The surprising news to me was the delta of OC temps over stock temps was only +1C at idle, and only +3C at load. The PWM temp deltas were pretty much neutral (-1C at idle, +1C at load). All temp deltas account for the difference in ambient temperature, if any, between readings.

The tainted part is that the CPU fan under PWM control ramped up a bit as the CPU got hotter. From the normal low of about 630 RPM, it went as high as 900 RPM. I couldn't hear any increase in noise (but my room has a noisy computer in it, so the floor noise level is a bit high).

To be fair, I should re-run the test making sure the fan doesn't ramp up. But this test reflects how I'm actually going to use the system, and so is more useful to me than determining the theoretical increase under identical conditions.

[Ken5d]

I added a Yate Loon D12SL to the HR-01, as an experiment. (The HR-01 fan clips require an open-cornered fan, which the YL is not. So I tied the YL to the fan clips with a nylon tie. Not terribly secure, but good enough for my purposes.)

I ran it at 12v, because even at full speed, buried in the middle of the SOLO, I don't think it adds noticably to the noise level. So I figure any cooling I get is "free", noise-wise.

At stock idle, there was no change in the CPU temp, but the max PWM temp was lowered by 5 degrees. At stock load, however, there was even more improvement. CPU temp was 3 degrees cooler, and PWM max was 9 degrees cooler. In fact, my PWM max temp at load with the YL on the HR-01 is the same as the idle temp without the YL. That's a pretty noteworthy improvement, IMO.

[nick705]

Does the QuadGT have a ventilated I/O backplate (like the AW9D Max)? I wonder if you've increased the overall negative pressure in the case with the higher airflow CPU fan+duct, so more fresh air gets drawn in over the PWM heatsink.

The Abit "Silent OTES" idea always struck me as a bit of a gimmick, but from what you're saying it looks like it actually works as intended.

edit: not sure if "increased negative pressure" strictly makes sense, but you know what I mean....

[Ken5d]

Does the QuadGT have a ventilated I/O backplate (like the AW9D Max)? I wonder if you've increased the overall negative pressure in the case with the higher airflow CPU fan+duct, so more fresh air gets drawn in over the PWM heatsink.

The Abit "Silent OTES" idea always struck me as a bit of a gimmick, but from what you're saying it looks like it actually works as intended.

No, the QuadGT's backplate is solid. There are holes in the SOLO at the other end of the rear (the opposite end of the mobo from where the CPU sits), but if there is increased air flow into those holes, I couldn't tell. I tried with and without the added YL, and using the kleenex & hand methods, I couldn't feel any increase in air flow (either from the exhuast fan, or into the vent holes).

I do notice now, though, that with the YL on, there's an increased feel of air behind the YL, washing over the RAM.

The AW9D Max uses an even larger cooling "solution" (OTES2) than the QuadGT does. (The heatsinks are smaller, and there's only a single pipe connecting the NB and a PWM heatsink.) Like you, I'm still unsure whether these systems are actually helping or hurting.

[dfrost]

uGuru reports a "system" temperature, which may be the board temp, or it may be the NB temp (though as I mentioned, it doesn't seem to change, so maybe the board temp is more likely).

The uGuru-reported "system" temp does seem to be a board (or "case") temp. The Abit AI7 uses it as the controlling temp for their NB fan (a failure-prone little screamer) in FanEQ, but my measurements with an independent IR thermometer show that it doesn't change much with NB temp anyway. I've never figured out where the "system" temp sensor is located. Since I quickly went to passive NB cooling, I'm using the "system"-controlled fan header for my Nexus 120mm exhaust fan. Works great!

I don't know if the PWMs make up the VRM (voltage regulator module) or not, but they're almost certainly related to it. Maybe someone who really knows what they're talking about will jump in and help correct my terminology (though it's abit that calls them PWMs, not me!)

I wouldn't claim to be "someone who really knows what they are talking about," but I will jump in anyway. Abit's "PWM" is indeed their VRM temperature. On my Abit MB, the "PWM" sensor is located very near one of the VRM MOSFET's.

[Ken5d]

Thanks, dfrost, for the info. Very much appreciated!

[Ken5d]

Final? Update

I've run numerous tests, though if I could do it over again (no time), I think I could do a better job. Still, here are a few of the results (all taken with the CPU at 100% load [Orthos] for about 10 minutes).

Vent vs. no vent:
Based on my results, the fan vent that goes from the heat sink to the case exhaust fan can make a large difference, or very little difference. It depends on your setup.

With a 120mm JMC PWM fan @30% (~650 RPM) as the case exhaust fan, and no fan on the heat sink, using the vent resulted in an 8C degree drop in CPU temp, but a 2C rise in PWM (VRM) temp. (Note that I consider changes of 1-2C degrees to be noise, as I don't necessarily trust that the numbers are all that accurate.)

However, with the JMC fan mounted on the heatsink, and no exhaust fan, there was no change in CPU temp with or without the vent. (There was a 2C degree rise in VRM temp without the vent, which seems anomalous to me.)

Fan on HS vs. Fan on case (as exhaust):
As might be concluded from above, this also depends.

Using the vent, there was no change in the CPU temp whether the CPU was mounted on the heatsink or on the case. However, there was a 4C drop in VRM temps when mounting the fan on the heatsink.

Without the vent, there was an 8C drop in CPU temp when mounting the fan on the heatsink (but no change in VRM temps). This result seems pretty intuitive.

1 Fan vs. 2 Fans in series:
There are quite a few permutations on this, as you might imagine, especially since my second fan (a low speed Yate Loon) is different from the first. In fact, too many for me to count, let alone test. One conclusion seems to be, if you use 2 fans, then whether or not you use the vent doesn’t seem to matter. (However, in my particular configuration, where the PSU gets its fresh air supply from the air directly above the heat sink, it seems using the vent probably makes more sense, even if it doesn’t immediately help the measured temps.)

Without a vent, using 2 fans was clearly better than using 1. Sticking the YL on the case as an exhaust when the JMC fan was on the heatsink lowered the CPU temp by 4C, and the VRM temps by 11C. (The YL was running at 8v, ~1000RPM.) These were the best temps I recorded in all my tests. All this proves is if you add more airflow, you get better cooling (not always the case, but true here).

But what about with a vent? With the JMC fan on the heatsink, adding the YL lowered the CPU temp by 4C, and the VRM temps by 7C. But after removing the JMC fan from the heatsink, there was no change in CPU temps. The YL was able to cool the CPU just as well without the help of the JMC. But the VRM temps suffered, rising 8C. I tried raising the speed of the YL, varying up to 12v, but the best it could achieve was a 2C lowering of the VRM temps (still 6C higher than with the aid of the JMC).

So, my final configuration will most likely be to use the vent, along with the JMC on the heatsink, and the YL as an exhaust.