Smaller Tubes Better

[zer0signal667]

It's been said many times, but now it's time to actually get back on topic before this thread gets locked. This is a watercooling forum, not a debate-the-meaning-of-a-culturally-based-but-non-derogatory-Americanized-word forum.

 

[plywood99]

On the other end of the spectrum there are those who'd be happy with a dog turd sitting on their CPU if it kept the CPU cold, and overclocked better. They don't care - they close the case side and from the outside it looks the same anyway.

Hehe, the dog turd may best describe my homemade block:

A nexXxos XP clone that cools rather well, better than my swifty 6000, but I'm sure to some they simply would not put such a thing in their pc.....

 

[BellaCroix]

Anyway... um... erm... Thanks again for the advice Cathar. You (finally) made me understand the tubing size debate. While I'd still really love to see comparissons made using different sized tubing on various blocks I now know what to look for and understand better the impedance created by certain diameter tubing.

I'm going to scurry off now and hide. (Always get scarred when moderators start coming around).

 

[Megadeth_Guy01]

Hehe, the dog turd may best describe my homemade block:

A nexXxos XP clone that cools rather well, better than my swifty 6000, but I'm sure to some they simply would not put such a thing in their pc.....

Why do you have that (disgusting) Metallica CD on your precious workspace that your homemade block is on! lol.

 

[plywood99]

Why do you have that (disgusting) Metallica CD on your precious workspace that your homemade block is on! lol.

Uh Oh, you must be a Dave Mustane fan eh??

 

[Megadeth_Guy01]

ROTF hell ya!

Edit: what did u mill out your block with, I've been wanting to tinker with some copper and make my own block.

 

[Top Nurse]

Jeez guys you sure had a lot of fun while I was out and about! I'm bummed I didn't get to have some as well. Guess I will just go sulk in the corner and dream about low-flow silver blocks. BTW, I will still use 6mm ID tubing in my setup and I don't expect any problems from using it. YMMV

 

[Cathar]

...

In case you hadn't realised, you're the only one talking in absolutes.

 

[madmat]

If a particular size tubing is 1/2 the size of another tubing, say for instance 1/4" versus 1/2", then the former will offer a quarter of the actual area the latter offers. Moving up the scale to a incremental increase in size we find that (roughly) 3/8" = 2 x 1/4" and 1/2" = 2 x 3/8".

The smaller the tubing you use will generate more head on your pump even in a loop with identical tubing lengths due to back pressure which will lower your flow rates and increase your working flow pressure to a point. Once you reached a point of the back pressure overcoming the head on your pump the flow rate will decrease but the pressure or velocity of the flow won't increase, you'll just end up with the pump in effect having the positive side of the flow bypass back to the negative side of the flow inside the housing seeings as the flow will take the path of least resistance.

A vane type pump is (I believe) a dynamic pump and not a positive displacement pump and doesn't generate high pressure well, A good analogy would be to compare a fan to an air compressor with both of them rated at the same CFM, the fan when placed under a load will just lower it's CFM output to the point of zero flow and the load required to do this will be quite insignificant whereas the air compressor under the same load will continue to afford nearly the same CFM flow as it did in an unloaded state.

What all this boils down to is that with the same pump in an identical system where the only difference is the tubing diameter the potential for flow will be afforded the best performance with the largest tubing....the smaller the tubing the lower the cooling capacity of the system overall. This is purely in an apples to apples system though, when you start mixing in pump X versus pump Y with tubing A, B or C and block I versus block M then you're looking at wildly ranging results.

 

[Cathar]

What all this boils down to is that with the same pump in an identical system where the only difference is the tubing diameter the potential for flow will be afforded the best performance with the largest tubing....the smaller the tubing the lower the cooling capacity of the system overall.

It should be noted that this is all still linked to the resistance of the rest of the system. If the rest of the system is so restrictive that flow rates would never get high enough for some tubing size to act as a significant source of restriction, then there really is no need to use some tubing size any larger than that which is required for the application.

Use a whole stack of super-restrictive items, and yeah, even 1/4", or even 4mm, can be justified.

If the total level of system restriction, coupled with the characteristics of the pump being used, suggests a certain (larger) tubing size, then using a smaller tubing won't necessarily cause performance to fall through the floor, and the differences may not even be that large, but the issue is that for something as simple as choosing the "right sized" tubing for the application why would one implement the system in a fashion that is needlessly restrictive and therefore lower performing than it needs to be?

This is where the issues with smaller tubing lies. Yes, it can be used, and to good effect, but only in the correct circumstances, otherwise the system may not be operating at its peak potential, and by that, that includes the particular pump and radiator, as well as the blocks.

Small tubing alone does not make for a poorly performing system, and the performance of a system should not be judged on tubing size alone. Smaller than is appropriate tubing just makes for a system that won't reach it's full potential, even if that lost potential is only 0.5C, or 3C or 5C, or whatever.

I just want it to be clear that I'm not bagging out all small tubing setups, I am just explaining what are the important issues to consider in whether a small tubing choice is appropriate or not.

 

[madmat]

I have no issues with small tube systems either but my point is that if you have a pump with 6' of head in a system that totals 4' with the blocks and fittings but not the tubing factored in then the larger the tubing you run the better your performance will be, if your tubing plus your system equate to a head greater than 6' as seen by the pump you'll have zero flow.

Simply going up from, say, 6mm tubing to 10mm tubing will cut your back pressure nearly in half and ditto for moving from 10mm tubing to 13mm tubing and in a system with a medium head pump and high head blocks (impingement) plus multiple blocks like a TDX and a maze type chipset and GPU block then running small tubing will really hurt your overall performance. On the other hand if you were running all maze type blocks you could likely get away with smaller tubing and see very little impact to overall performance.

One of the guys from DD made an impingement chipset block and the added head from that block, even with dual D4's in series, totally killed the performance of the TDX equipped loop. Once that block was removed from the loop and replaced with a standard maze4 chipset block they ran it with a single D4 and got far better results due to the lower overall head of the system.

 

[Cathar]

Exactly madmat.

The lesson from all of it is that a full system must be a carefully balanced selection of parts, and that goes for the blocks, the tubing, the fittings, the pump, the radiator, etc.

That is the primary thing. An improperly designed system made up of poorly matching components will be outperformed by a well put together system. The tubing size is just one part of the whole gamut of pieces that should be "right sized" to the rest of the system.

 

[TheLonelySwedish]

So how can i tell whether my parts mesh well? I have all PolarFlo TT stuff; i'm hoping this isn't a bad thing.

 

[madmat]

Are they labyrinth style blocks (Maze 4 for instance) or impingement (RBX, TDX) style? What model pump? What's the head on it? approximately how many feet of tubing are you going to use? How many fittings?

Answer those questions, those are what will dictate whether you need to use larger tubing or not.

 

[Lazn_Work]

Really what it means is that sometimes smaller tubes don't hurt performance, but larger tubes never hurt performance.

So if you always err on the side of larger tubes you will never have to worry about them causing any problems.

==>Lazn

 

[plywood99]

One of the guys from DD made an impingement chipset block and the added head from that block, even with dual D4's in series, totally killed the performance of the TDX equipped loop. Once that block was removed from the loop and replaced with a standard maze4 chipset block they ran it with a single D4 and got far better results due to the lower overall head of the system.

That's why I run my system in parallel instead of series. My gpu block is very restrictive, which if I ran it in series, would kill the flow to my cpu block.

You can run restrictive and less restrictive in the same w/c system, just have to know how to work with what you have....

 

[plywood99]

ROTF hell ya!

Edit: what did u mill out your block with, I've been wanting to tinker with some copper and make my own block.

Mill? LOL, I wish I had a mill! I used a dremel and about 20 cutoff discs. Was a long and tedious process but well worth the time for the performance.


BTW, nothing like doing your own block, I done several and have learned quite a bit from it.

 

[Megadeth_Guy01]

Aww, that sucks man, looks good for being dremeled, I have access to a nice mill (non-CNC) so we'll see if I can churn out anything good. Did you thread the holes for your barbs or just braze them or superglue them in? lol.

 

[zer0signal667]

I'm going to scurry off now and hide. (Always get scarred when moderators start coming around).

I'm always around... Just because I don't post doesn't mean I'm not lurking in the shadows

 

[Top Nurse]

Have been watching this thread for a while and it seems to be correct to me I usually fly off the handle when others start bagging on 6mm ID systems as not working well and go on a tangent till I cool off a bit. So I guess the bottom line is that they do work, but you have to be careful with mixing and matching of components correct? So if you want an A-C, Innovatek, or Koolance system it is best to stay in that range of products that are based on the same water cooling philosophy.

 

[plywood99]

Aww, that sucks man, looks good for being dremeled, I have access to a nice mill (non-CNC) so we'll see if I can churn out anything good. Did you thread the holes for your barbs or just braze them or superglue them in? lol.

I use a caulking made specially for aluminum gutters, will stick to ANYTHING and, once it cures make a very tight seal.

 

[Cathar]

So if you want an A-C, Innovatek, or Koolance system it is best to stay in that range of products that are based on the same water cooling philosophy.

If only it were that simple. Some of the waterblocks made by some of the "low flow" companies (not talking about any company in particular here - just the broader picture) actually perform worse at low flow rates than various "high flow" blocks perform at the same low flow rates.

It's all about reality vs perception. A company who markets for low flow wants its customers to perceive that their products are specifically tuned for low flow, when the reality is that they may actually not be.

Of course, finding out the truth on these matters requires good independent testing that tests across a broad range of flow rates and reports the pressure drops, and quite often that sort of information simply is not available.

So again, it all falls back to perception. In the absence of independent, verifiable and thorough test results potential customers get left with choosing whether or not to believe that the company who is selling the stuff actually is correct in what they say.

This is not to say that the low-flow stuff is going to be bad - it's all about degrees of scale. It's not "will it be bad?", it's more a situation of "how far off being as good as it can be is it?". For some customers, they simple don't care to know the answer to that question, which is fine, because they appreciate their chosen system on other aspects that are important to them. Where it gets ugly is when those same customers start to attack other solutions in defense of their choice.

 

[Top Nurse]

Well we will soon be finding out about the Cuplex XT as I just e-mailed the person you suggested. Soon as the reply comes back Sharka will send them a block along with anything else they need to make it happen.

 

[Erasmus354]

Well we will soon be finding out about the Cuplex XT as I just e-mailed the person you suggested. Soon as the reply comes back Sharka will send them a block along with anything else they need to make it happen.

Get sharka to send other AC blocks as well...I am very curious to see a good review of AC stuff, and perhaps even see 6mm vs 8mm comparisons on AC gear.


Also, I would like to thank Cathar for clearing up a lot of the confusion on the boards here with large vs. small bore systems. As always your knowledge of watercooling seems to be unsurpassed.

 

[Top Nurse]

Get sharka to send other AC blocks as well...I am very curious to see a good review of AC stuff, and perhaps even see 6mm vs 8mm comparisons on AC gear.

Also, I would like to thank Cathar for clearing up a lot of the confusion on the boards here with large vs. small bore systems. As always your knowledge of watercooling seems to be unsurpassed.

What Sharka wanted was a rundown on the blocks as they are used by the typical consumer. That being 6mm and 10mm ID tubing. I suppose their are some who might also use 8mm ID as well, but the sales so far have been centering around the aforementioned sizes. They can actually supply the blocks with 6mm, 8mm, and 10mm fittings if desired. They are also willing to supply tubing for the push-fits and pumps if needed.

An extra XT came on the last shipment from Aqua Computer specifically for a review site here in the USA. Any more blocks will probably be at their expense so I wouldn't plan on seeing tests of every A-C block made, but who knows I think I might be able to encourage them to send a Cuplex Pro depending on the outcome of this water block testing. Perhaps one of the new SLI TwinPlex's (has a Cuplex Pro style block) when they show up stateside in the next month or so.

But personally I think it is going to take someone like [H]ardOCP to do a review of an Aqua Computer system as it encompasses some pretty fancy hardware and electronics.

 

[BellaCroix]

But personally I think it is going to take someone like [H]ardOCP to do a review of an Aqua Computer system as it encompasses some pretty fancy hardware and electronics.

Not to slight the reviews done by [H]. I've found them very well conducted and remarkably useful, but... the reviews Cathar posted from Procooling (or whatever it was called) were downright amazing!

There's so many questions to answer about AC gear, hose sizes, flow restirctions, etc. and the cost tends to be so much if AC isn't going to supply the equipment gratis, I'd love to see a couple review sites share costs, then info, pass parts around a little, kick the tires so to speak, then combine all the data into a single review shared across a couple sites.

That way we'll get the technical reviews from the sites set-up to do hardcore benchtesting and scientific measurements as well as "real-world" results from smaller sites just installing it into a computer and overclocking until the magic smoke is released.

 

[Pooky]

Innovatek would really be considered low flow, compared to Alphacool and even, Aqua Computer. They have relatively simple blocks (no pin grids, no jets), 8mm ID tubing and, their flagship pump is a 12v Eheim 1046 with a bit more head pressure than a 1048. The Pro Cooling review of their Rev3 block (http://www.procooling.com/reviews/html/innovatek_rev3_waterblock_revi.php) defines their approach. Bella Croix commented on another thread that Innovatek blocks can go as low as 0.3 lpm and still function.

On the other side though, Alphacool doesn't use the same ideals. They make high flow blocks and their pumps are up there too. The pump they sell with their highest performing NexXxos XP kit has 1500 litres an hour (http://www.alphacool.de/perl/shop.p...m=get_art&menu_id=2&prod_id=100&&art_id=10306), not to mention a radiator that is more likely sold as a 'high flowing' part here comes with the kit also. The one thing they do share with Innovatek, and most other German outfits though, is the 8mm ID tubing, with options to go with 6mm ID if wanted.

Most people tend to call German/European systems low flow, but that's not the case at all. It seems as if it is the same situation outside of Europe as well, except that the tubing sizes are usually larger.

 

[Calis]

do you think that a straw will have more flow than the Nile River?

come to think of it...

 

[Top Nurse]

Innovatek would really be considered low flow, compared to Alphacool and even, Aqua Computer. Top Nurse commented on another thread that Innovatek blocks can go as low as 0.3 lpm and still function efficiently.

I don't think that was me that made that comment because I don't know much about Innovatek coolers.

 

[Cathar]

Pooky, you do raise an excellent point, and it is also something that I raised and argued over at Procooling.

Some of the so-called "low flow" systems are in fact really just "high flow" systems using a "moderate flow" tubing size. IMO, it more or less goes like this in terms of a tubing size's suitability for a particular flow-rate (including the effect of fittings)

4mm or 5/32" = very low flow (< 1lpm / < 0.25gpm)
6mm or 1/4" = low flow (< 2lpm / < 0.5gpm)
8mm or 5/16" = moderate flow (< 4 lpm / < 1.1 gpm)
10mm or 3/8" = moderate/high flow (< 6 lpm / < 1.6gpm)
7/16" = high flow (< 10 lpm / < 2.65gpm)
1/2" = high to very high flow (< 12.5lpm / < 3.3gpm)

Now various sites have taken then time to measure the actual flow rates in what many people call "high flow" systems, and they typically fall in the 3-7LPM range, which is rather interesting because 8mm tubing's "usefulness" range partially straddles the set of flow rates that realitsically occur in "high flow" systems.

It is a very astute point you've raised Pooky. In terms of pumps, the radiator, and even somewhat commonly with the waterblocks, more than a few of the so-called "low flow" systems share an awful lot in common with their "high flow" brethren. About the only thing that differs is the tubing size, and even then in terms of effective usefulness range with respect to the typically observable flow rates, the differences are not that large.

Therein lies much of the answer to just why many of these so-called "low flow" systems do perform very well in comparison to "high flow" systems - and that is purely because there really isn't that much different at all at the heart of it. Many are in every essense a "high flow" kit, but with 8mm tubing, instead of 3/8" tubing, attached.

 

[Master Ninja]

Cathar, remind me never to argue with you about this stuff.

This does raise an interesting question, though; is there a way for consumers to determine the restrictiveness of components? A quick glance at the AquaComputer and Innovatek websites reveals no information indicating the flow rates of the products, and radiators have a similar lack of detail. Pumps are the only things to show how much water they output, but that only prevents us from using a pump that pushes more water than the tubing can carry.

Also, just for my own clarification; the rates you have listed are the maximum effective rates, correct? That is, if I'm using 8mm tubing, then it will create resistance against the pump if it's pushing more than 4 lpm?

 

[Cathar]

Also, just for my own clarification; the rates you have listed are the maximum effective rates, correct? That is, if I'm using 8mm tubing, then it will create resistance against the pump if it's pushing more than 4 lpm?

The effective maximums are just an arbitrary guideline, by which you can assess up to what point some tubing size is going to offer a level of restriction that is so high that beyond that point you really would be better off going to a larger tubing size. This is a guideline against reasonably powered centrifugal water pumps that are in use in water-cooling loops.

Any tubing size creates resistance against the pump from the moment that any water flow moves at all. The level of restriction grows roughly proportional to the flow rate squared. i.e. if you double the flow rate, you quadruple the resistance against the pump.

The suggested maximum figures given are just the point where the tubing is offering so much resistance for most pumps that really it is just better to be using a larger tubing size if targetting flow rates higher than that.

 

[mohammedtaha]

Any tubing size creates resistance against the pump from the moment that any water flow moves at all. The level of restriction grows roughly proportional to the flow rate squared. i.e. if you double the flow rate, you quadruple the resistance against the pump.

The suggested maximum figures given are just the point where the tubing is offering so much resistance for most pumps that really it is just better to be using a larger tubing size if targetting flow rates higher than that.

But don't we use wider tubes anyways. For example, if we have a 1/2" ID system the width of the water coming out of the pump is less than 1/2" OD since the barb is 1/2" OD. Therefore, the resistance wouldn't be that high. But then again this all depends on the type of tubing used. If you use Tygon, you would probably experience better flow, less resistance, than using a cheap 1/2" tube from Home Depot.

I believe the flow depends on the pump first, and then the length of tubing and then the design of the restrictive elements in your cooling and it goes the other way around. Tubing is almost always the last thing to worry about. If you have a diesel powered pump and a 2mm tube you can have more flow than a 1/2" system but it doesn't mean it will cool better. Heat from the pump, heat from friction, the water block might not be able to handle such high flow and the radiator might be badly designed, poor cooling, and no fan attached.

Phew .. done .. now i'm awaiting Cathar's corrections and constructive criticism..

 

[Master Ninja]

The suggested maximum figures given are just the point where the tubing is offering so much resistance for most pumps that really it is just better to be using a larger tubing size if targetting flow rates higher than that.

Then is there a real disadvantage to using a pump that is too powerful? I have a Hydor that does 320 GPH ~= 5.3 gpm and I hope to use 8mm tubing in a new rig. Obviously I would get better performance if I used larger tubing to take advantage of the high flow. So what would the advantage be in spending the extra money on a less powerful pump? It seems like, given a certain diameter tubing, the effectiveness would eventually plateau as pump power increased, so at least the performance wouldn't be getting worse.

 

[Top Nurse]

This thread is starting to get interesting

A-C has a philosophy of quiet cooling and thus perhaps markets their products in that direction. I recently had an Airplex Evo 240 radiator, but ditched it because it seemed quite restrictive (compared to a BI Pro) to me based on my subjective blowing through it. However, there have been plenty of posts here in this forum from people using these A-C radiators that say they work excellent with A-C systems. You don't suppose that the reason they don't go to a BI style radiator is that they don't want to create a less restrictive loop? From what I have been reading here it appears that designing in restrictiveness to the system can be a good thing so long as you have total control over the other factors correct?

BTW, I can't wait till we get some results on the Cuplex XT with all three sizes of tubing available for it.

 

[Cathar]

mohammedtaha, masterninja, you both are expressing valid point and with a level of thought that is excellent.

In answer to your questions, this was all exactly explored in detail here:

How much pump is enough? How much is too much?

Yes - there is such a thing as a "too-powerful" pump, along with the heat that it puts into the water that overwhelms any gains at the water-block and exceeds the capacity of the radiator setup to keep the water cool enough such that actual gains get seen.

In summary from that thread the following (rough) recommendations were made:

- Ideal to use a pump rated at a maximum of 12W of power draw per 120x120mm radiator area.
- Recommended to avoid using a pump rated at more than 16W of power draw per 120x120mm radiator area.
- Do NOT use a pump rated with a maximum power draw of more than 20W per 120x120mm radiator area.

Those recommendations are necessarily very broad to address most scenarios. Everything is really a case-by-case basis.

Really though the way to design a system is by selecting the pump second-last, and the tubing last.

Pick the blocks that you're going to use. Work out the restriction that they are going to offer to the pump. Then pick the radiator based upon your space limitations and noise desires. Then while looking at the flow vs performance curves of the blocks and radiator, and looking at the level of resistance of the system vs the pump's pressure-flow curve, choose a pump that will give a flow rate that ensures that the blocks/radiator are operating at a desirable level of performance (water-flow wise). When choosing the pump, keep in mind the recommendations above. Once the predicted flow-rate is determined, then choose the tubing size to suit. Can choose smaller tubing sizes if flow rates are low.

 

[Cathar]

However, there have been plenty of posts here in this forum from people using these A-C radiators that say they work excellent with A-C systems. You don't suppose that the reason they don't go to a BI style radiator is that they don't want to create a less restrictive loop?

Not really to do with restriction. Less restriction is always desirable.

There would be a different reason why they're using the A-C rad.

Radiator performance is also affected by the flow-rate. Radiators perform better at higher flow rates, and worse at lower-flow rates. Each radiator is different though. The Black Ice radiators start to rapidly lose performance below 3 LPM and really are "happiest" at 4LPM and higher, and so would not be a good choice for a low-flow A-C system.

The A-C rad internally may be better designed to suit lower-flow rates and not lose efficiency as quickly as something like a Black Ice will. I don't really know though - I've never seen good testing for an A-C rad.

Much like waterblocks, radiators are similar. Some can work well at higher flows and perform bad at lower flows. Some can work well at lower flows but be so restrictive that it prevents their use in higher-flow setups. Some are just plain good across the board.

This is also something which is being heavily explored over at Procooling and other places, being the characteristics of an ideal computer water-cooling radiator.

 

[BellaCroix]

Top Nurse commented on another thread that Innovatek blocks can go as low as 0.3 lpm and still function efficiently.

That was my post, don't remember where I saw it... think it may have actually been from an Innovatek/AC forum or something. For some reason I thought the .3 l/hr figure was in reference to AC blocks.

And I'd probably put the period after the word function. If memory serves the claim was that they would function at .3 l/hr rates... not that they'd function well.

 

[Erasmus354]

That was my post, don't remember where I saw it... think it may have actually been from an Innovatek/AC forum or something. For some reason I thought the .3 l/hr figure was in reference to AC blocks.

And I'd probably put the period after the word function. If memory serves the claim was that they would function at .3 l/hr rates... not that they'd function well.

.3 l/hr is much different than .3 lpm which is what flow is normally measured in.

 

[Cathar]

I call shenanigans at 0.3 liters per hour.

At that sort of flow rate, the water itself flowing through the block will heat up by around 2.9C per 1W of heat energy being emitted from the CPU.

This effectively means that even a 30W heat load would be enough to cause the water to boil within the waterblock, let alone that the CPU would then be at well over 100C.