[Review] PrimoChill Typhoon III

[Diverge]

Well no, I was just saying stick a Y on the outlet of the pump top, it's not quite the same as all that As long as you have a res with multiple ports you only need to add 1 Y everything else is pretty much the same.

I'm going to try it with a DDC and see what happens, I have to order some parts first though.

Soon as you introduce a Y you're adding resistance, unless the bottom part of that Y, the |, is larger diameter then the 2 branches.

A good analogy would be a 4cylinder turbo charged exhaust system. From the turbo down, usually called a down pipe, you might run 3" pipe. Under the car you might have a Y, with the | being 3" in dia, and the branches being 2.5" to run dual exhaust out the back (for looks, or cause its hard to run a single full 3" out the back in some cars). Swap out your 3" downpipe, for a 2.5" one, and you just lost a lot of horsepower by adding more back pressure, and negating your dual exhaust. Or use a Y with 2.5" dia on each of the 3 ports, and you just killed some flow as well (or lost horsepower)

[bluehaze]

I battled back and forth on how the heck to properly test the Typhoon III in order to show the capability and performance that was unattainable before.

The thing is if you put a Y onto the inlet and outlet the restriction remains the same so the results will be the same as a single loop. To properly test you would need to compare multiport res to the T3 with a Y coming out of conventional pump top. This has been my thinking all along is that the res at the end of both loops removes the restriction at the end of the loop therefore both loops will be able to freely flow only restricted by the individual components of each loop and that is where the gains come from.

The T3 is innovative and convenient in that it combines the pump top, res and the Y into one making for a simpler loop and saving you money if your just starting out but I honestly think if you already have a multiport res and a pump regardless if it's d5 or ddc you will save alot of money and see the same gains in flow just by throwing a Y on the outlet of your pump top and running individual loops back into 2 different ports on your multiport res.

I am going to test it with a ddc as it would seem like the better the pump the bigger the gains, I have to order a Y and another flow meter though to properly test it. And I already understand where your coming from with the testing as it's going to cost time and money and I suspect I will only see similiar gains in flow which result in less than 1 degree of difference in temp so it's kind of painful ordering the parts just to test it

[alacheesu]

I started with these numbers:
Typhoon III
Parallel loop 1: 1.5 gpm
Parallel loop 2: 1.5 gpm
Total system flow rate: 3.0 gpm

Using the PQ-curves from this review, I calculated that a stock D5 running the same parallel loops would have a flow rate of somewhere between 2.8gpm and 3.0gpm, or between 1.4gpm and 1.5gpm for the parallel loops. So, a win for the Typhoon III, but I don't see why it couldn't be done with a stock D5 and Y's. My calculations did not include the added pressure drop from Y's (which is probably relevant at ~3.0gpm), but because it's using the stock D5 PQ-curve it should be accurate otherwise. From this we can also see that the Y's more than doubled the system flow rate and are clearly a major breakthrough in loop design... wait, what?

[SNiiPE_DoGG]

Typhoon III - CPU Concept by SNiiPE_DoGG

(fixed thumbnail with no imagshack advert)

[Utnorris]

Interesting block Snipe. Curious if you did something like the Stinger and had the two returns coming back to the two inlets on the T3 how that would work.

[BoxGods]

i think the simplest way to explain the T3 is its like giving the D5 a top that has 1" inlets and outlets, instead of .5" (and yes i know 1" opening is more than double a .5" opening, its just for thinking, not proven math)

BINGO. Your are essentially correct, but to continue your analogy, the line or path between the inlet and outlet ALSO need to be the 1" you mention (and no thats not the right scale lol)

Soon as you introduce a Y you're adding resistance, unless the bottom part of that Y, the |, is larger diameter then the 2 branches.

A good analogy would be a 4cylinder turbo charged exhaust system. From the turbo down, usually called a down pipe, you might run 3" pipe. Under the car you might have a Y, with the | being 3" in dia, and the branches being 2.5" to run dual exhaust out the back (for looks, or cause its hard to run a single full 3" out the back in some cars). Swap out your 3" downpipe, for a 2.5" one, and you just lost a lot of horsepower by adding more back pressure, and negating your dual exhaust. Or use a Y with 2.5" dia on each of the 3 ports, and you just killed some flow as well (or lost horsepower)

This is an awesome way to see it. the base leg of each Y in the "add Y's to a pump" method is the portion going through the pump and is the restriction. In T3 that leg is nearly doubled in throughput.

Typhoon III - CPU Concept by SNiiPE_DoGG

(fixed thumbnail with no imagshack advert)

I wondered how long it would be before somebody considered that blocks CAN be customized for more flow rate now. Monsoon, my CPU block design shows a 7 degree drop when hooked up to a T3 vs a regular D5 because it likes the higher flow.

[SNiiPE_DoGG]

I have a question about this test - why arent the two outlets on the pump tested individually in each loop? - shouldnt the performance be constant in a given loop if you just stop the flow completely through the other?

I wondered how long it would be before somebody considered that blocks CAN be customized for more flow rate now. Monsoon, my CPU block design shows a 7 degree drop when hooked up to a T3 vs a regular D5 because it likes the higher flow.

except I just realized that that design doesnt make any sense because the fluid velocity in the block would still be the same just split in two

[millertime359]

Great review Skinnee, thanks for getting it done so quick for us.

[Maxspam]

Some of the images arent working

[Manicdan]

Typhoon III - CPU Concept by SNiiPE_DoGG

(fixed thumbnail with no imagshack advert)

i would have put the barbs on a slight angle, give it a cool look than just strait down. and if you want to make it square so they can choose to have it wider or taller.

on thing i really dont like about cpu blocks is how small they look once you put on the tubing.

[bluehaze]

Soon as you introduce a Y you're adding resistance, unless the bottom part of that Y, the |, is larger diameter then the 2 branches.

No not exactly, a Y halves the restriction by converting 1 intake path into 2. There is a resistance at the inlet of the Y but as long as that inlet is the same size at whats feeding it it does not add any restriction at all.

Also as long as the restriction of the inlet of the Y is less than that of the combined outputs of the Y the Y will not add any restriction whatsoever.

In otherwords the only time a Y will pose a restriction (in a computer loop) is when turning 2 sources into one. (For instance turning both loops back into one before dumping back into the res) It will never pose a restriction when turning one source into 2 (At the beginning of a loop feeding two sepearate loops) as long as that restriction is greater than the restriction at the inlet of the Y.

[CrazyJoe]

...
looking at the results of test one, the other options returned 1.39 to 1.48 GPM, the T3 gave the top loop 1.49 GPM and the bottom loop 1.98 GPM. so we have a 1% to 42% gains depending on component and comparison
Math: (1.49/1.48=1%) and (1.98/1.39=42%)

Just a correction on your math -

(1.49-1.48)/1.48 = 0.7% increase in flow
(1.98-1.39)/1.39 = 42.4% increase in flow

[Diverge]

No not exactly, a Y halves the restriction by converting 1 intake path into 2. There is a resistance at the inlet of the Y but as long as that inlet is the same size at whats feeding it it does not add any restriction at all.

True, but in your case, your inlet is the size of the 1 fitting hole coming out of the pump. All you're doing is taking that flow and dividing it into 2 paths.

The T3 doesn't have that limit, since it has a big chamber, and 2 ports exiting from a chamber that was tuned to do so. And an even bigger chamber on the inlet side.

In otherwords the only time a Y will pose a restriction (in a computer loop) is when turning 2 sources into one. (For instance turning both loops back into one before dumping back into the res) It will never pose a restriction when turning one source into 2 (At the beginning of a loop feeding two sepearate loops) as long as that restriction is greater than the restriction at the inlet of the Y.

True again, but you're not gonna gain anything. Your flow was sorta already capped by the pump exit itself, and the diameter of the inlet to the Y.

Take your garden hose (one connected to a house), and put one of those Y splitters on it. Run one hose full open (no nozzle), then do them both at the same time. Whats gonna happen? Soon as you open the 2nd one, your gonna drop in pressure, and the water is gonna come out the hoses slower.

[bluehaze]

True again, but you're not gonna gain anything. Your flow was sorta already capped by the pump exit itself, and the diameter of the inlet to the Y.

That is where your wrong and what most people don't understand, pretty much any block or rad you throw in a loop is going to add more restriction than the inlet of the Y therefore the governing factor now becomes the blocks in the loop not the inlet of the Y. The inlet of the Y will always be able to flow more than the blocks because it has less restriction.

[DarthBeavis]

That is where your wrong and what most people don't understand, pretty much any block or rad you throw in a loop is going to add more restriction than the inlet of the Y therefore the governing factor now becomes the blocks in the loop not the inlet of the Y. The inlet of the Y will always be able to flow more than the blocks because it has less restriction.

25/3

[surfhick]

In the TIII, the inlet (water collection chamber) is more than twice as big as the two outlets(barbs) and the pump is more than capable of keeping it full of water. A tee will be limited by its inlet. The amount that comes out is <= to the amount that goes in. It can't be more. For a tee to do the same thing, its inlet would have to be twice as big as either of the outlets and the reverse would have to be true of the collection tee. I THINK that this is what the TIII accomplishes. When the water comes out of the pump volute it goes into a small reservoir (within the reservoir) that feeds the outlet barbs.

Since I see calls for a comparison between the TIII and tees. I wonder how having the D5 pump into a small res with two outlets would compare. Res > D5 > res > 2 loops > back to Res.

[shazza]

I'd suggest cubing the diameter of the inlet, then taking the logarithm of the inverse exponential sine wave to calculate the absolute impact on flowrates ...


Or, maybe I'll just keep using my Tyhoon III and not worry about this higher math stuff

(sorry, skinnee, for the useless post - all this intellectual debate is making my brain hurt)

[Bei Fei]

I'd suggest cubing the diameter of the inlet, then taking the logarithm of the inverse exponential sine wave to calculate the absolute impact on flowrates ...


Or, maybe I'll just keep using my Tyhoon III and not worry about this higher math stuff

(sorry, skinnee, for the useless post - all this intellectual debate is making my brain hurt)

I thought it went like this?

[JOCKTHEGLIDE]

Yes, there is a billet aluminum top coming that will be available in brushed black and silver anno, as well as polished to a chrome finish.



All kinds of cool possibilities



.

SOLD

[MomijiTMO]

Great review, skinnee. Now I'm definitely going to set up the second loop on my TIII.

It is a bit confusing showing the sum of the two loops as a separate bar on the chart, took me a while to realize what it was saying. And I'm still puzzling as to whether it's a fair representation - since none of the individual blocks have this combined flowrate going through them. (Not trying to take away from the performance of the TIII at all, just not wanting to give the cynics anything to snipe at).

I think this post sums up my opinions on the T3 and the review. It is very misleading and me being me skipped the text and went straight to the results and graphs. I was like HOW IS THAT POSSIBLE and then read through the article. I'd report on 2 flow rates and if the cpu flow is better than the alternatives then it's a winner. Stuff gpu flow rates [within reason].

[SNiiPE_DoGG]

I still want to see flow rates in parallel (per loop) vs flow rates of each loop on its own from the same port.....

[tenormadness]

Great review! Just a quick question though. For the flow rate tests what speed is the D5 set for?

[Jah]

Great review

Now could you pretty please include a graph with 0 flow in the 2:nd loop to the dual loop tests, just as reference. I know the data is already in the review and it possibly confuse some people but just make it grey and skinnee

[shazza]

I think this post sums up my opinions on the T3 and the review. It is very misleading and me being me skipped the text and went straight to the results and graphs. I was like HOW IS THAT POSSIBLE and then read through the article. I'd report on 2 flow rates and if the cpu flow is better than the alternatives then it's a winner. Stuff gpu flow rates [within reason].

To clarify my position; I think the review is great. I do think showing the bar graph with the combined flows can be misleading ... but this is entirely up to skinnee as to how he wants to show it. The important point is the Typhoon III works, and works well in both single and dual loop configs. I know I'm quite happy with mine.

My being silly was just because we sometimes like to make things very complicated

[Snyxxx]

I can appreciate the amount of time to perform this testing, but there are two things that bother me and I cannot let go.

1. As others have said, adding the flow from each loop is not fair. By halving the amount of pressure drop of the one big serial loop to attain two loops, the restriction is less and flow rates must increase.

2. I guess I will be the first to mention that the commingling of water within the pumps do not make this a "true" dual loop setup. The cooler water for say a CPU loop will be contaminated by the warmer water of the other loop averaging the water temperature. I really see the term "dual loop" thrown around a lot in this review, but from a temperature perspective, this is not true.

I have always been a fan of big inlets for these types of pumps, which is why so many of us modded both the D5 and DDC from their pathetic stock tops in the first place.