The impact of tubing sizes

[samuknow]

LOL. Still holding onto the dream I see.

That statement of yours would only be true if we're talking about the differences between 1/4" to 1/2". There's a whole range of tubing sizes between the two, as I showed, with far, far, less differences. I don't want to see people walking away and thinking that the differences between, say, 3/8" and 1/2" are anything like 1C. They simply are not. There's actually been plenty of people who have seen the same sort of thing over the years, where they changed to a 1/2" system, and saw no difference at all.

This is exactly what happened. I started with 3/8" because it fit nicely. In the quest for lower temps, I went to 1/2". The only thing I gained was a harder install and more clutter. I would have been better off with a larger radiator. To get 3/8" ID tube on 1/2" barbs, I heated the ends and stretched them over a tool. Worked great.
You have prompted me to go back to 3/8" tube. The 1/2" looks impressive, not worth the added problems though.

 

[aaronjb]

I'll be moving to 3/8" Tygon after seeing these figures. The 1/2" is a bit too bulky, and the 3/8" is much easier to route around bends.

But the push-lock fittings are nice as well. Once upon a time I used 8mm tubing with push-locks and QR fittings, and it was super-easy to work with.

 

[QuietIce]

This is exactly what happened. I started with 3/8" because it fit nicely. In the quest for lower temps, I went to 1/2". The only thing I gained was a harder install and more clutter. I would have been better off with a larger radiator. To get 3/8" ID tube on 1/2" barbs, I heated the ends and stretched them over a tool. Worked great.
You have prompted me to go back to 3/8" tube. The 1/2" looks impressive, not worth the added problems though.

If you have external runs you may want to keep 1/2" for them. Not only is the tubing more rigid and rugged but (usually) we're talking longer runs and no fittings in the way to slow things down. Keep in mind the study Cathar has done assumes 8 connectors (4 choke points) in the loop, which accounts for a large part of the restriction in larger tubing as shown by comparing the 7/16" to 1/2" curves.

But I still bow to the advantages of 7/16" tubing (or even 3/8" quick-connects) for internal runs ...

 

[voigts]

Very interesting info. I have had several threads where I debated that the tubing sizes don't make a big difference as long as the flow rate is good. I however didn't realize that there was so little difference between 3/8" and 1/2" tubing.

 

[Mycobacteria]

Well I would presume he used the same CPU for all the tests...

However a give or take 10% in the results.... is a large amout that means it like + or - up to 4 deg either way!

 

[Moto7451]

Yes but its also a consistent +/-. The temp sensor isn't going to decide that it will be +3 on tuesdays and -1 on wednesdays. Remember that the temperatures themselves are not important, its the delta between them that is.

 

[DSpot]

The biggest restriction to the liberty of different tubing sizes is still my choice of pump, in this case the D5. Using reducers will be too messy and untidy for a loop within the case.

 

[Mycobacteria]

Yes the sensors have a give or take BUT at least you can average out the results, and if you repeat the experiment you can even do some stats to give "confidence" to your results. Whereas a theroy based calculation type approach you cant do this....

 

[aaronjb]

Yes the sensors have a give or take BUT at least you can average out the results, and if you repeat the experiment you can even do some stats to give "confidence" to your results. Whereas a theroy based calculation type approach you cant do this....

Sure you can. His equations work every time. I think he's working on a pretty solid foundation here. I'll take Cathar's calculations over a number of other folks' "real-world" tests with physical setups.

I think he's modeled the setup quite well. Find a flaw in his math or reasoning, and I won't trust the data.

Claiming that his numbers are unreliable just because they weren't done with a physical test setup is out of touch with modern engineering principles. While the actual numbers would likely vary a bit from Cathar's numbers, the temperature deltas would probably be close to spot on.

 

[eightballrj]

I'll be moving to 3/8" Tygon after seeing these figures. The 1/2" is a bit too bulky, and the 3/8" is much easier to route around bends.

But the push-lock fittings are nice as well. Once upon a time I used 8mm tubing with push-locks and QR fittings, and it was super-easy to work with.

Whew... but I have seen many a pushlock leak. I don't trust them in something leak critical.

 

[nikhsub1]

Sticky'd.

 

[Cathar]

The 10% margins of error referring to the deltas being correct to within 10% of each other, not 10% errors in the absolute temperatures, which is obviously ridiculous. What the absolute temperatures are is dependent on the calibration of the thermal diodes, the heat load, radiator fan-power, and so on. Some of those aspects would also affect the deltas too.

 

[Otter]

Whew... but I have seen many a pushlock leak. I don't trust them in something leak critical.

Back when I heard bad things about these and decided to stick with good old barbs and hose clamps, but now that it's come up again, I don't recall any recent complaints about pushlock/quickfit connectors leaking. Have they improved to the point we can trust them around our expensive electronics?

 

[Cathar]

Back when I heard bad things about these and decided to stick with good old barbs and hose clamps, but now that it's come up again, I don't recall any recent complaints about pushlock/quickfit connectors leaking. Have they improved to the point we can trust them around our expensive electronics?

Gabe (of Swiftech) has provided some long term manufacturer feedback about push-fits over at XS.

In summary:

MUST use tube inserts, or very hard/stiff-walled tubing
Pushfits are rated for pressures much high than in a w/c system
Over a period of years, the O-ring can permanently indent the tubing wall, reducing the seal pressure and causing a seal failure

So while they're okay for the short-term 1-2 years, they do seem to have long-term reliability issues.

After much discussion about how to solve the tub-barb interface, it came down to two main suggestions:

1) Compression fittings of an OD matching the tubing, but an ID on the fitting of the same ID as the tubing (i.e. would need to stretch the tubing over the fitting)

2) Much like 1), but with regular barbs. For 3/8" tubing, use 7/16" barbs, and stretch thet tubing over the barb (much like we do with 7/16" tubing over 1/2" barbs).

 

[samuknow]

Back when I heard bad things about these and decided to stick with good old barbs and hose clamps, but now that it's come up again, I don't recall any recent complaints about pushlock/quickfit connectors leaking. Have they improved to the point we can trust them around our expensive electronics?

I just ran 250 feet of 3/8 OD tube for a water cooler and used instant connects. We have almost 100 PSI of water pressure. I don't worry about it one bit. All of our water coolers use them. Never a leak.

 

[Cathar]

I just ran 250 feet of 3/8 OD tube for a water cooler and used instant connects. We have almost 100 PSI of water pressure. I don't worry about it one bit. All of our water coolers use them. Never a leak.

From what Gabe was saying, that's what they're good at - high-pressure systems where the internal pressure helps to push the tubing out to seal against the O-ring. In low-pressure systems, they have issues.

The main issue does seem to be due to the suction port from the pump (negative pressure trying to collapse the tubing). It may very well be possible to get around this by plumbing the tubing length from the preceding item into the pump with regular barbs.

 

[Otter]

Thanks Cathar. Matching compression fittings do sound interesting. Reliability, bling, and a 0.1C temp drop!

 

[Mycobacteria]

does the delta increase as flow rate increases or if you use a better pump?

ie if I get a really good pump will the difference between the small tube and the large tube become more apparent say 2-3 deg difference instead of a 1 deg difference?

 

[Cathar]

does the delta increase as flow rate increases or if you use a better pump?

ie if I get a really good pump will the difference between the small tube and the large tube become more apparent say 2-3 deg difference instead of a 1 deg difference?

I crunched the math a while back 'cos the question was asked on another forum.

The answer as that the better the pump, the less the differences are.

The worse/weaker/more inappropriate the pump, then we see greater differences. This explains why years ago 1/2" tubing did provide some noticable benefits, 'cos available pumps (as well as blocks) were all tailored around high-flow & low-restriction. Since that time it's been learned that restriction is good (in blocks) and as well, pumps have gotten better suited to PC water-cooling (good pressure, medium-flow).

Times have changed. Water-cooling as a whole has moved forwards. Pumps and blocks have evolved as everyone begins to better understand what is required to improved performance.

As a consequence, we no longer really need 1/2" tubing to sustain the super-high flow rates and super-low restrictions required for near-peak performance. Near-peak performance is now achieved at flow rates in which smaller ID tubing does not form a major source of restriction in the system, and any added restriction is now more easily overcome anyway by the more suitable pumps we have access to.

 

[eightballrj]

If I a reading the C/W charts correctly I should be able to deduce that it is much better to have a higher flowrate through the water block than the radiator?


I am asking this because of all of the uproar about Swiftech posting the pics of two Stealths running in parallell. Will that REALLY hurt the temps of the cores on the cards much? There is also a thread about a guy trying to run a dual CPU loop and it got me thinking about how parallel versus series for a dually would work.

Thanks Cathar!