Ducted ORBS

[sitka]

Unbelievable,

This guy http://home.iprimus.com.au/mek mentioned it as did I on www.Hexus.net Reviews Forum. I tried it with paper along time ago but I didn't know what I was looking at and I was thinking of using a 120 fan but his pop bottle trick gave 10 degrees in about 3 minutes of work. Not sure if I believe it yet, to good to be true.

Oh yea, I'm Christening it Thermaltake "Tuber" Orb or TOrb

 

[BLoop]

Linkified.

Worth a shot. I'm gonna go see if I can find a bottle.

 

[sitka]

Yea I couldn't find one to fit so I slit the side and wrapped with an elastic.
Adjusting the airspace at the bottom is too fun and very effective.


Hmm, maybe it should be a concet, coonckat, cuncatenat,....concatenation?

Of Ducted and Orb

= Dorb, hee hee, (I got a million of 'em)

 

[Gmork1]

Hmmmm, I've got a ThermalTake Super Orb (2 Fans)

Think I'm gonna go find a bottle...

Gmork

 

[sitka]

Initial numbers are in. Pooh-ya!

30 minutes running CPU Stability TEST 6.0 In CPU warming mode

Duron 650@1017
with 1.7V
38deg C

Would easily have been 47deg C in pre "Tuber" config just a few hours ago.

 

[Tominator]

Great idea!

I'll bet a Pepsi bottle has better air flow and runs cooler than a Coke bottle does!

 

[sitka]

Tominator makes a good point

PLEASE INCLUDE SODY POP BRAND WITH POST

Believe it or no I used Guarna Flavoured GUTZ,
kind of like Black Cherry JOLT. If you don't know JOLT you just ain't trying.

 

[AMD4ME2]

We Need Pics!! I'm having trouble picturing this in my head and why it would cool so much better.

 

[Dundain]

Sweet Yea, please post pics and does anyone know of a way this could be done to square heatsinks? (say, FOP-32)

 

[Mikewarrior2]

Sitka, too bad I think both measurements may be off significantly. Your 37C full load temp at 1017 isn't really possible unless your AMBIENT CASE temp is around 11C.

likewise, the old reading may have been wrong, so that means that the actual core temp difference isn't really being portrayed fully.


Mike

 

[sitka]

Thanks, MW2. Lots of stuff on that link and I'll pour through it and see if I can understand it.

If I believe you and try to think why the readings would be incorrectly low. Maybe the ducting is forcing air at the thermo sensor in some unique way. This may not have been happening without the duct, thus the readings may be off.

Let me get back to the link now and put my thinking cap on.

 

[Viperoni]

On the guys page, how the heck can your ambient temp be higher than CPU temp without pelts???

 

[Mikewarrior2]

what's interesting is that both teh orb-superorb both tend to be read very low by mb's(10-15C too low), and almost consistently.

I personally believe that it could be the way air flows off the heatsink. THe "vortex" forces air under the socket, which really messes up the reading. Just an idea.


Mike

 

[sitka]

Viperion, didn't catch that the first read, seems funny or impossible.

Anyways stuck a thermal prob next to the die. Pretty inaccurate but maybe better than the back side sensor which could be anywhere.

Ducting gives a few degrees cooler on this measurement but then again there is definately a larger volume of air swirling near it with the duct. This can only be a bad thing if all sensors are being cooled better while the die itself is cooling worse. But I think one could say that there is an improvement in the near environment that the CPU operates in.

I'm going to set the voltage at 1.85 and see what happens...

 

[sitka]

Anybody with a Fop38, you now how the heatsink flares at the top?
Does the air go ripping through that wide part and hit the motherboard?

 

[Mikewarrior2]

it does on 2 sides, but not the entire socket.

 

[kponds]

I think this idea (at least the websites description) has gotten a few holes in it.

First of all, the guys says that when he puts the tube on the RPM speed of the fan goes up and assumes that its due to increased airflow. I don't think thats correct at all. Assuming that he's got the same voltage applied to the fan in both scenarios, the only reason the fan speed would increase is that theres less resistance to sucking air thru the fan blades which would mean the fan is not being as efficient in moving air - i.e. the blades are cavitating. The fan blade pitch is calculated for the amount of air that can be pulled into the cooler. By sealing the sides of the cooler, you limiting airflow into the cooler and pulling in less air, hence the blades will spin faster.

Secondly, he indicates that the temperature on the motherboard is at 24C and the CPU is at 18C. Unless their is active cooling (peltier or cold water applied) that would indicate that the ambient temperture in the room is at a maximum of 18C (64.4 degrees). I don't know what temperature this guys runs his AC at in his shouse but if its at 18 degrees C or lower and he's got 18 degrees C on his CPU, his motherboard temperature is realtively hot at 24C. I would really like find out what he's using to measure temperature and would like to see the tests done using a Thermal diode reading as I suspect the drop he's seeing is based increased airflow across a thermistor on the motherboard or on the heatsink.

Thirdly, the Ram Air Effect he being mentioned in one of these posts reminds me of the older (mid to late 60's muscle cars). The Ram Air intake was designed to force more air into the carbuerator of the car. The car essentially had a hood scoop which when the car was moving forward directed outside air thru ductwork directly to the carbuerator. The main benefit to this - increased air pressure going into the carb which resulting in more air (and gas) going thru the card and the result being better horsepower. The systems works great at 40 MHP or more but didn't provide any horswpower increase when the car was sitting still. Unless the guys PC is moving around on him, there is no RAM Air Effect going on here.

I simulated this setup using a piece of cardboard in one of my test systems and did see the same results as he did with the fan speed. I didn't get the same results with the CPU temperature (thermal diode temperature) - mine increased 4 degrees C with the tube bottemed out.

In my humble opinion the only way this modification could be a benefit is if the ThermalTake really screwed up on the design of the cooler and mismatched the the fan to the cooler. I sorta doubt they did that.

 

[sitka]

Kponds, I agree totally with what you say about the fan speed. And in fact when I ducted the Sorb the fan speed decreased a little which is what you would suspect. These fans may spin fast but they aren't very strong and any change in the surrounding airstream seems to have an effect. With the Voltage cranked to 1.85 there was a rise in the readings but not as high as I thought it should have been. This leads me to suspect that the probes are getting cooled out of wack with the die like mentioned. But this does not preclude a lack of a performance increase. The true way to get this worked out is to install the probe in the Orb and check both Ducted and Non-Ducted. (Too many case Mods already on tap to take this on today). All other things being equal that would tell if the near die heatsink material contains less heat, then those crazy excited electrons on die would have an easier time getting away from where we don't want 'em.

....no such thing as cold, only a lack of heat.
Yea right tell that to them in Michigan

 

[Viperoni]

The tube wouldn't have a ram air effect like he described, but by leading all the air in from the front, straight into the sink, that would drastically improve the amount of air coming through the sink.

-----------------------------------------
ORB<<<<<<<<<<<<<<<AIRFLOW<<<<<<<<<<<<<<<<
-----------------------------------------

Normally the airflow would just go in a circle from the bottom of the orb, back to the top, and down again

The fan would also be geting fresher (colder) air, which woudl also decrease temps

It's a good idea, but 2-3c is the max you'll lose on your temps

 

[sitka]

So we ran without a problem for weeks on end and then

Stardate 12312000

surfin

get about 10 IE windows opened up an a couple OE's, write some letters, the last one is a masterpiece on HTTP and we get IE (5.5 if it matters) error.

clear, redo

A OK

surfin the waves cosmic, 20 X (time units/60) later it happens again

WTF

Check MB temp, Higher than I've seen it with Ducted config

Observe::: duct is pushed down all the way

Acknowledge::: previous action &quot;I wanted to see what would happen&quot;

Evasive Action::: back off duct about .750 inch and get temp drop of 3 degrees in a few moments

Once again running without fault


Just saying...

 

[sitka]

Just a little bump, vintage stuff

 

[crash2much]

My system is ram cooled. I duct taped it to the ceiling fan with a funnel jammed through the side of the case. Runs a 12c but I get real dizzy playing Quake this way.