[Review] PrimoChill Typhoon III

Quote:

---

Originally Posted by Martinm210

Thanks for the review, great work as always.:up:

Sorry to hear all the grumbles, but I can understand everyone's perspective. This parallel stuff is hard to convey and creates alot of confusion because it's never really been talked about or testing much at all.

We've always talked "Flow Rate" as being one thing. This is true in a series system, where pump flow rate and block flow rate are equal and the same. There was never a need to indicate which one we're talking about.

Parallel systems are different though. In parallel, you have "Pump" flowrate and "Block or individual loop" flow rates. The purple bars represent "Pump" flow rate, it's what the pump produces. What I think some folks want to see is only "block" flow rate. I think they want to see that comparison only, because block flow rate is what translates to some sort of thermal performance gain.

I think they are both important.

Pump flow rate is important to know and understand when considering where the pump is operating. You don't really want a pump operating too far on the right hand side of the curve either. The more pump flow rate the more pump heat dump, the harder the pump works. I don't think we see too many problems in the context of watercooling and these smaller DC pumps, but I wouldn't dismiss it completely. I have had direct experience with this on a commerical pumping system I designed this last winter. Simply put, my pumps where too big for the amount of restriction I had and I learned that bigger is NOT better. In my case, I had two 5hp 2" pumps capable of about 200GPM and a pressure drop of only about 10' at 150GPM or so. This meant the pumps were operating at nearly free flow conditions. The result was that the pumps drew more current than I could legally have my electrician set the circuit breakers at. The pumps were operating too far on the right hand side of the curve, drawing too much current, overheating and eventually drawing more than 140% of the electrical pump motor rating (which is the limit we can set the breaker). My solution: add valves at the pump outlet to artificially introduce restriction to the system. My electrician thought I was nuts when I told him this, but sure enough it worked just fine. This in turn lowered the flow rates and allowed the pump to operate more appropriately in the middle of the curve. This is still a workaround, but it goes to show you pump size and selection is important and you don't want a pump that is too big(operating too far on the right) either or you can have other problems. Anyhow, long story short, pump flow rate is important too, but only for the pump, how much heat it produces, how much current draw, and how hard the electrical motor has to work (More is worse for heat dump and can be worse for the pump motor/efficiency. The optimal efficiency point is usually in the middle of the curve somewhere which is a good balance of restriction flow).

Block flow rate is what we are normally looking at and thinking about when we are after thermal gains. It's this flow rate that matters and translates to thermal performance and probably what everyone is really wanting to see here. These were all represented in the other bar colors (More is better).

If I had to vote on how to present flow rates. I'd suggest noting both, but distinguishing between and separating block flow rates and pump flow rates. They are both important, but both represent two different things. Pump flow rate is important for pump sizing, pump operation, and how much pump heat dump is to be expected. Block flow rate is important to know for thermal efficiency of blocks and radiators.
Pump flow rate - More is usually worse, Ideally you are in the middle of the curve or left if anything)
Block flow rate - More is always better

This parallel discussion and testing is all new stuff though, not something anyone has done much published testing on...so I'm not surprised we all have our own thoughts on what we want to see or what's the best way to compare things.

Great work, and interesting results!!:clap::up:

So much for the old "Series is always better" thought....never say always!!..Everything is only always a shade of gray!!!:eek::D

---

WoW. Martin just commented on a part I designed...may have to go wake the wife up to read that =)

I think the REAL advantage people are going to come to love about the T3 in parallel is that you now have a lot of options. If you want to run a restrictive type CPU block, your not going to take a hit on your radiator flow rates. Said another way, users can mix and match blocks and rads with a lot less worry now because you can configure your loops to work with each parts strengths.

And I know we are going to see new block designs that take advantage of the higher flow rates simply because they were restricted to what a lower cost pump (read D5 or DDC or even EIHM) was feeding them.

I hope Martin saw that I DID mention him/his comments as the inspiration for the T3's inlet design. I know that testers like Martin and Skinnee must often wonder if all the hard work they do REALLY has an impact. It does I assure you.

Quote:

---

Originally Posted by BoxGods

I think the REAL advantage people are going to come to love about the T3 in parallel is that you now have a lot of options. If you want to run a restrictive type CPU block, your not going to take a hit on your radiator flow rates. Said another way, users can mix and match blocks and rads with a lot less worry now because you can configure your loops to work with each parts strengths.

And I know we are going to see new block designs that take advantage of the higher flow rates simply because they were restricted to what a lower cost pump (read D5 or DDC or even EIHM) was feeding them.

I hope Martin saw that I DID mention him/his comments as the inspiration for the T3's inlet design. I know that testers like Martin and Skinnee must often wonder if all the hard work they do REALLY has an impact. It does I assure you.

---

I agree it provides options, but I disagree about the less worry part. Parallel takes alot more effort and planning or it can be worse!

Parallel "Can" be better, but it won't always be. In order for it to be a benefit over series, you need several things. A pump with a very flat flow oriented curve (The D5 fits that pretty well). A very high restriction loop and ability to BALANCE the pressure drop in both loops. And you also need ALOT of restriction to prevent overdriving the pump motor. Perhaps the D5 and DDC are so bullet proof these days that you need not worry. I'm not so sure though, I don't think it's any mystery why we had so many DDC2 failures back in the days...alot of those were likely tied to low restriction setups or dual pump setups that pushed the motors harder than optimal.

If you have a very unbalanced loop it's possible for the restrictive parallel loop to receive less flow than it would under a series setup. In addition, running in parallel has the chance of putting the pump under alot of unnecessary stress.

Here is an example of how the pressure drop and pump curves relate. In series weather it's a pump or the pressure drop, the curves are added vertically in the pressure Y direction. In parallel, the curves are added in the horizontal or flow rate X direction. The blue and red curve add up to the green curve.

If you setup a parallel set of loops where one loop has high restriction and the other is relatively low, you can easily do more harm than good.

Here is a theoretical presentation of where series will still be the better option. The parallel loop has three flow rates. Series has one.

http://img197.imageshack.us/img197/5373/paralr.png
Series
Pump/Rad/CPU/GPU Block flow rate = pump flow rate, all one flow rate = 1.3GPM

Parallel
Pump = 3GPM (Working well beyond the best efficiency point) This will prematurely wear the pump out and decrease life of the motor.
CPU & Radiator = 1.0 GPM
GPU = 2 GPM

So in this particular setup, the only benefit would be for the GPU, the CPU and Radiator would suffer, and the pump motor would suffer.

This is all just presentation on how the curves interact. You would need to work with actual pressure drop curves to pin the optimal piece down, but I'm pretty sure series will be the best solution for many situations as well.

Bottom line: Parallel "Can" be a benefit, but you better know what your doing or you can cause more harm than good. Series setups is the safer route if you either don't want to bother with the check or don't have enough information to check. Unfortunately we have ALOT of blocks and various products out there that don't provide pressure drop information.

How can you make sure when you don't have pressure drop information that parallel is the optimal solution??:shrug:

Anyhow, I just want people to understand neither parallel nor series is going to ALWAYS be the optimal solution and there are some details and conditions that would favor one over the other. If you don't understand it pretty well, series is still a very very good direction to flow...:D


In general I think parallel might be a good option if:
-You know what you're doing and have pressure drop information for everything.
-A running a flatter high flow pump like the D5 (I doubt the DDC will show much benefit)
-You have ALOT of restriction, enough that the combined pressure drop curve is not overly stressing the pump motor.
-You can balance the pressure drop of the two loops relatively equally.


Anyhow, I like the options present with parallel, but I would caution people to do a bit more planning if it's something they want to do...:up:

will there be some sort of calculator?

What I meant by less worry was more that if you wanted to add a new part to parts you already had then you had options, or that you could try different options and were not limited to just one serial loop, or just parallel loops, and that you can try different combinations.

For new users who want to start with just a CPU block a single loop is fine, and if they want to add a second rad and some GPU blocks down the road then they can move to parallel loops if that offers better performance without having to buy a new pump--the more options you have the better your chances are for a good result.

Yes there is a configuration utility coming that will allow users to plug in different blocks, radiators etc in different combinations to more easily determine whats ideal for their specific set up. The Loopilator lol. All the data and the database portion are ready, just working on the front end now.

That's some very interesting and good food for thought..

With the T3 out there now, I had been thinking of doing a new build with these two loops off of a T3:

CPU block -> Chipset block -> Rad_1 -> T3

3x GPU blocks (tri-SLI) -> Rad_2 -> T3

My only concern was determining if this was too much load for a single D5...

Quote:

---

Originally Posted by Ketzer7

That's some very interesting and good food for thought..

With the T3 out there now, I had been thinking of doing a new build with these two loops off of a T3:

CPU block -> Chipset block -> Rad_1 -> T3

3x GPU blocks (tri-SLI) -> Rad_2 -> T3

My only concern was determining if this was too much load for a single D5...

---

Regardless of the merits or caveats of the T3, I'm not sure this is really practical. GPU's are generally capable of running at much higher temps than a CPU, and sharing the same reservoir isn't going to keep your CPU temps down without some serious radiators... I think running two separate loops would be preferable.

Thanks Martin for the insight. I can see where there could be an issue for some setups, I know for me personally with my setup I saw no real gains from a temp standpoint, but as I stated before in my quick, non-scientific review, I wasn't expecting any. What I did get was a simpler, cleaner and regardless of what has been said about my setup, it didn't hurt my temps on either my GPU side or CPU side. I know this comes a s surprise for some as it was for me too, but I can't argue with my own experience. As I said before, others may or may not get the same results. I know a lot find it hard to believe that a PA120.3 would be able to keep my 3 x GTX280's at 50c under load and that by combining it with my CPU loop (also has a PA120.3) that it wouldn't hurt my CPU temps. However, if you take in the fact that both loops were already at that level with separate loops, then why would anyone think that combining them would cause the temps to go up?

As far as the strain on the pump, I would worry, maybe, if I was running the EK Supreme or the Koolance and then some other very restrictive blocks, but for my setup, I did not see the flow go down, but up, which maybe I am wrong on this, but wouldn't that indicate that the pump is not working as hard?


OT: Now I am going to say this one last time for all the folks reading this, I have not been "Shilling" this product, nor will I. I gave an honest opinion of what I liked and didn't like and where I would have done things differently. This is no different when people have asked about EK, :banana::banana:, Koolance or any other product I have owned and used. I call it like I see it, but I do it in a not so hateful way as some have. I have never attacked someone's review, even when I disagree with it, take a look at the Swiftech reviews done by Swiftech, I may not agree with them and I may state that, but I never attacked the review. I know some look at this as "Shilling", sorry, but it isn't, it's called being tactful and respecting the reviewer. If I want to dispute a reviews claim, I do my own testing and post it and then let people decide for themselves. I have no reason to do otherwise since I do not do mods for shows and need support or get free items. I do not do pro reviews, just my own experience with products I buy and if I get excited about something then so be it, doesn't mean I think the entire world should own one. I have had my fill with some of the comments directed at me, especially when I cannot respond to them due to where they are posted. If you want to debate something I state in a comment, feel free to post right after me, I love a good debate and I do not mind being wrong as long as I get a chance to debate it.

Someone call you out for Shilling Norris? It never came across to me as you were.

Thanks for the thoughts there Martian. I am now getting a little curious if i should try my GTZ MCW60 on one loop and MCR320-QP on the other. I was wanting to give it a try and see if it worked, but seeing I can't measure flow or anything in my system, I don't want to blow the pump up or anything.

Quote:

---

Originally Posted by BoxGods

What I meant by less worry was more that if you wanted to add a new part to parts you already had then you had options, or that you could try different options and were not limited to just one serial loop, or just parallel loops, and that you can try different combinations.

For new users who want to start with just a CPU block a single loop is fine, and if they want to add a second rad and some GPU blocks down the road then they can move to parallel loops if that offers better performance without having to buy a new pump--the more options you have the better your chances are for a good result.

Yes there is a configuration utility coming that will allow users to plug in different blocks, radiators etc in different combinations to more easily determine whats ideal for their specific set up. The Loopilator lol. All the data and the database portion are ready, just working on the front end now.

---

Nice!

It would be great to have a little design guidance with parallel. Series is pretty hard to screw up, but parallel is different. Not hard, but at least some rough guidance should be provided so people get the best setup that can get.

Anyhow, thanks for the neat findings on parallel and product to take advantage of it.:up: It's been ignored for far too long and certainly seems to have some application and possible benefits from the results. Maybe some of this math and calculation that's necessary for parallel setups will steer more conversations towards performance and science...THAT certainly can't be a bad thing...:yepp::D

Cheers!
Martin

wow Martin awesome insight. hmm i wanna see some dead pumps :D

Quote:

---

Originally Posted by Hondacity

wow Martin awesome insight. hmm i wanna see some dead pumps :D

---

Keep in mind I'm trying to caution based on experience I've had with large commercial pump design. AC and DC motors are so different, I don't know anything about electrical motors though. I just give them power and expect them to turn for me...lol!

One thing is for sure that the DDC and D5 have been through several design iterations. I have found that the D5 and the DDC are both now "Overvoltage" protected, so it's very possible they are RPM limited somehow to protect things like running at max flow as well.

Maybe someone can check with Laing and ask if there is any concern with running the D5 near max flow (zero pressure) continuously, or if there was some optimal point to target?

Martin were those pumps you were using in your above described scenario positive displacement pumps? If so those should operate much differently than a mag drive pump with a fixed rpm, if you cut off the flow in a lcs with a mag drive the impeller keeps spinning and the pump will eventually overheat due to lack of circulation but a positive displacement pump (mechanically linked) will either burst a tube or blow a breaker (if it's appropriately sized) when you stop the flow. Both pumps are designed to operate against a load but unless the positive displacement type has some kind of rpm limiter it will just keep spinning up until something fails if it has too little of a load.

Quote:

---

Originally Posted by Utnorris

As far as the strain on the pump, I would worry, maybe, if I was running the EK Supreme or the Koolance and then some other very restrictive blocks, but for my setup, I did not see the flow go down, but up, which maybe I am wrong on this, but wouldn't that indicate that the pump is not working as hard?

---

Pumps are backwards from common sense. I don't understand it either, just measured it.

Pumps draw more current and work hard the less restriction they have. Go figure!??:shrug:

This is an old test, and I screwed up by only testing at the outlet, but you can see the pump draw goes up with more flow rate:

http://www.overclock.net/gallery/dat...Pump-D5-10.jpg

It only draws around 16 watts at 0 flow rate and around 29 watts where I could no longer measure. This same thing has occurred with every pump I have tested.

Why I don't know, something to do with how electrical motors work. Maybe one of the electrical folks can chime in why...:shrug:

Quote:

---

Originally Posted by hellcamino

Martin were those pumps you were using in your above described scenario positive displacement pumps? If so those should operate much differently than a mag drive pump with a fixed rpm, if you cut off the flow in a lcs with a mag drive the impeller keeps spinning and the pump will eventually overheat due to lack of circulation but a positive displacement pump (mechanically linked) will either burst a tube or blow a breaker (if it's appropriately sized) when you stop the flow. Both pumps are designed to operate against a load but unless the positive displacement type has some kind of rpm limiter it will just keep spinning up until something fails if it has too little of a load.

---

No they were standard 2" centrifugal pumps. They were a 2" pump with a 5HP 220V electrical motor. Full Load Amps on the motors was 22.4 and I was drawing over 34 amps because my load was so small.

My original setup (very low restriction) was causing the pumps to draw around 34amps which was kicking the circuit breaker. After added a tuning valve of sorts to the outlet and introducing some artificial restriction, I was able to get that down to 28 amps and fix my draw problem.

Anyhow, I don't design pumping systems very often, so it was pretty enlightening for me. I paid over $2K for each of these pumps under contract and had a serious issue that was due to my system being too low in restriction. Something new to me, although after talking with a guy at work that previously managed our city's water system, he made it sound like it's common practice to install valves at both ends of pumps that operate continuously and pressure meters to tune these sorts of things.

Anyhow, little things we ignore in our water cooling loops are a much bigger deal with commercial pumping systems.:shrug:

Quote:

---

Originally Posted by millertime359

Thanks for the thoughts there Martian. I am now getting a little curious if i should try my GTZ MCW60 on one loop and MCR320-QP on the other. I was wanting to give it a try and see if it worked, but seeing I can't measure flow or anything in my system, I don't want to blow the pump up or anything.

---

Think balanced pressure drop on each loop.

The GTZ is much more restrictive than the MCW60 or MC320, so on that, you'd probably want to run the GTZ on one side and the MCW60 and MCR320 on the other side.

You can keep it simple by looking at pressure drop at one point. Pick 1.5GPM as your reference point.
Per swiftech's charts:

GTZ is around a 2.4PSI drop at 1.5 GPM
http://www.swiftnets.com/assets/imag...Z-PD-vs-FR.png

MCW60 is about .8PSI
http://www.swiftnets.com/assets/imag...0/PD-vs-FR.gif

MCR320 is about .6PSI
http://www.swiftnets.com/assets/imag...0_PD_vs_FR.gif

So you'd have around 2.4PSI on the CPU and 1.3PSI on the GPU/Rad.

This probably isn't balanced enough to use parallel. You'd get a fair amount more flow on the GPU/RAD side and your CPU side may do worse than series. Maybe after you add some chipset blocks or a more restrictive GPU block?...

Anyhow, that's as far as I'd go with it...doesn't have to be an exact science. Skinnee is showing that he got gains when his loops weren't perfectly balanced, so I know there has got to be some leeway. How much, I'm not sure..

You are still talking about a mechanically driven electric pump and the same rules for restriction apply, without rpm limiters they will rev until something breaks if not enough restriction is built into the system. A fixed rpm mag drive pump shouldn't suffer from this type of problem.

Thanks for that Martian. I'm switching to distilled water now, so maybe I will just try a couple of different combos and see how the temps vary. If they even do. What is the best way (cheapest) to tell if I am putting to much strain on the pump, or should I not even worry?

Quote:

---

Originally Posted by hellcamino

You are still talking about a mechanically driven electric pump and the same rules for restriction apply, without rpm limiters they will rev until something breaks if not enough restriction is built into the system. A fixed rpm mag drive pump shouldn't suffer from this type of problem.

---

Cool!

I told you I don't know anything about electrical motors. Sounds like there's nothing to worry about with the low restriction then.:up:

Thanks for clarifying..

Quote:

---

Originally Posted by millertime359

Thanks for that Martian. I'm switching to distilled water now, so maybe I will just try a couple of different combos and see how the temps vary. If they even do. What is the best way (cheapest) to tell if I am putting to much strain on the pump, or should I not even worry?

---

It's sounding like the pump speed may not be an issue with these small DC mag drive pumps.

Give it a shot..let us know..:up:

Quote:

---

Originally Posted by Martinm210

It's sounding like the pump speed may not be an issue with these small DC mag drive pumps.

Give it a shot..let us know..:up:

---

Great, thanks Martin. I will start playing with it then :up:

The RPM limiting would explain why the power draw on Skinnee's graphs does not really alter that much from the low flow to the high flow, both are well within the sweet spot on the curve.

I am digging around to see what I can find out on RPM limiting on the D5. Maybe I can figure out how to overclock it lol

I'M KIDDING!

Quote:

---

Originally Posted by BoxGods

The RPM limiting would explain why the power draw on Skinnee's graphs does not really alter that much from the low flow to the high flow, both are well within the sweet spot on the curve.

I am digging around to see what I can find out on RPM limiting on the D5. Maybe I can figure out how to overclock it lol

I'M KIDDING!

---

Oh, that's easy. . .24v Meanwell. :D

Laing pumps have a lot of control circuitry...there should be little worry about pushing it too hard these days. If you want to overclock a D5, get the D5 Strong and hook it up to a 24V PSU. Power consumption gets ridiculous, but performance actually gets a healthy boost (at least compared to the paltry boost a D5 Vario gets from going to 24V) :up:

Quote:

---

Originally Posted by Waterlogged

Oh, that's easy. . .24v Meanwell. :D

---

Why did Tim Allen and Tool Time come to mind when I read that. :rolleyes:

Quote:

---

Originally Posted by millertime359

Why did Tim Allen and Tool Time come to mind when I read that. :rolleyes:

---

Har har har:rofl: