Input on W/C Guide

[hokiealumnus]

Hi there. alacheesu has said I should start a thread because there are apparently several factual errors in my guide and it definitely needs to be factually correct.

So, if anyone has time and is willing, please have a read through my Water Cooling Guide for Beginners.

Bear in mind this is for beginners and not the experienced XS water guru. So...help me correct any factual errors and make a thorough, complete, yet beginner friendly guide.

Hopefully, when it's completed, it will be a good resource to point people to when they're just starting out and asking the basic questions we always answer.

Thanks very much for your time!

[alacheesu]

I'll just copy my response from the sticky:
---------------------
Reading it right now. I have a few comments for the first pages:
1. You divide case options into "cases with holes for tubing" (rad external) and "Mountain mods". I would argue the most common option is neither of these. Most people use a standard, full tower case and put a radiator or two internally.

2. For nearly all products you mention, you link to specialtech.co.uk. This seems like advertising to me. A more honest approach, imo, would be to link to manufacturer specs.

3. The MCP355 w/aftermarket top does better in a high restriction loop than the MCP655, not the other way around as you have written. (not a big issue, but still)
----------------

I also think this section is incorrect. I'm far from an expert on water blocks, but I'm sure someone has some input on that:

High Flow Blocks
(These do not restrict flow as much as the alternatives. High Flow is best when planning on a multiple-block loop.)

* Heat Killer 3.0 (sorry UK folks, Specialtech doesn't carry them; though if you're interested, they are known for stocking things on request), accepted in many places as the best block currently on the market.
* D-Tek Fuzion v.2 (v.1 was discontinued)
* XSPC X20 Delta V3

The UK and US generally prefer high-flow, where the flow rate of the fluid itself pushes the heat as quickly as possibly around the system, allowing the water to do the work. This is better for multiple block loops.

Low Flow Blocks
(These are a somewhat more restrictive, meaning it takes more pressure to push the water through them.)

* Swiftech Apogee GTZ
* Koolance CPU-350
* EK Supreme

Low flow is popular in Europe and the far east. The object of this is to use the blocks to slow down the flow rate of the fluid in order to pick up as much heat as possible, thus using the radiator as the main source of cooling. Problem is, Chinese radiators aren't the best, so that system doesn't work very well.

Good luck with your guide, I'm sure it will be useful for lots of people. I think you've done a good job so far.

[faster3200]

I also think this section is incorrect. I'm far from an expert on water blocks, but I'm sure someone has some input on that:

I was going to say the same thing.

First off, with the exception of the D-Tek block and the EK, the high flow blocks you listed are European and the Low flow blocks are American, so that doesn't even make any sense. Also, you will notice the low flow blocks generally tend to scale much better the more you increase flow, so they should be used with powerful pumps or in single loops like you said. While I am on the note of flow scaling, the comment about slowing down flow in order to pick up heat doesn't make any sense.

I always had the impression that the use of weaker pumps in Europe was for noise reasons. The same reason that sub-1000RPM fans seem to be popular on hardwareluxx.

Also, what is with the Chineese rad comment? It is really random and does nothing to help. With the exception of possibly Feser (according to some), no rads are actually made in China afaik.

Honestly, a guide like this seems silly. Parts change constantly. I think it would be much better to do a guide on the theory of WCing and how to pick the best parts based on theory and not specific items and then just do a yearly list of the top parts.

EDIT:
Also, the explanation of the rads is not very good. You just list a few and say they are good. I would go more in depth about the differences if you are going to do it this way. For example, you just say that high fpi rads need more powerful fans to be good, to me the way you said it makes them sound like they suck. In fact, if you have strong enough fans, they are the best, by a pretty fair margin.

I didn't even see you mention air in/water out delta once. That is a pretty important aspect of picking the right rad and fans.

[alacheesu]

Not sure what you wanted to write here:

To maintain the same cooling capability, you need more fins, which means you need more radiator. What that does NOT mean is that you need more fins.

Fins in the last one should be FPI, maybe? I also think you should emphasise the importance of fans more, IMO. The difference in cooling capacity between a rad with low speed and one with high speed fans is huge. "Is my rad big enough for X" is one of the most common questions here, and you cannot answer that without discussing fans as well.

[hokiealumnus]

In response alacheesu:

1. Perhaps I wasn't clear enough on that point. Basically it was trying to give the reader the easiest options. Most internal mounts require modding of some sort, or a pretty large full tower case, which was given admittedly short play with "Full-tower ATX form factor cases can often fit a single radiator and also contain holes to reach; installation is your option there." What you mention was a subset of the cases with holes. What do you think? Are there enough cases with room for internal rads that do not require any modding to give it its own section?

2. FWIW, I wanted to link to the manufacturer sites. The links there were on request by the owner of the site, Gilgamesh. They are a sponsor and supply us with some of our hardware for reviews. They're also listed in our sponsor list, linked from the main menu. Not trying to hide that in any way whatsoever. The site has to get hardware somehow. I try and make myself feel better by saying it was 'selective linkage' because they're a sponsor, not necessarily advertising.

3. Really? Stock specs put the 655 vario @ 50PSI and 355 @ 22PSI (per sidewinder). I wouldn't think a top could more than double the available pressure. Maybe I'm looking at it the wrong way?

4. FWIW, that was verbatim from Gilgamesh. He knows a lot more about water cooling than I do and I just put it in there without questioning. Faster: This applies to the Chinese rad comment as well. Boss man get whats boss man wants.

5: Perfect way to state that alacheesu, edited.

To Faster: As stated within the article, the parts are not exhaustive by any stretch. They are there as examples only. The point was to make a general guide. I'll also try and keep the parts updated as new ones come out.

Will go into more detail about high FPI rads, good idea. I didn't intend to make them sound like they suck. Just stating they need better fans to operate properly.

EDIT - Rather than going into too much detail on high FPI rads, I just added a simple comment: "This is not to say high fin density radiators are bad by any means; the right ones can even be better than their low fin density counterparts." Does that make you feel better about them?

[faster3200]

3. Really? Stock specs put the 655 vario @ 50PSI and 355 @ 22PSI (per sidewinder). I wouldn't think a top could more than double the available pressure. Maybe I'm looking at it the wrong way?

To Faster: As stated within the article, the parts are not exhaustive by any stretch. They are there as examples only. The point was to make a general guide. I'll also try and keep the parts updated as new ones come out.

3. Was proven a long time ago by Martin.

Also, if that is what you are trying to do that is fine. What about the other parts I mentioned though?

[hokiealumnus]

3. Was proven a long time ago by Martin.

Also, if that is what you are trying to do that is fine. What about the other parts I mentioned though?

I've seen and read that article...there is only one mention of the 655: "I happened to be a long time Swiftech MPC655 (Laing D5) vario user, so it was fun trying out this new little pump." The entire article is a VERY thorough and excellent comparison of tops vs each other and vs the stock 355. How does it prove that the 355 is better for loops that need strong pressure?

Sorry, didn't answer the block comment. Calling those high flow & low flow (less restrictive & more) were based on the comparison by Martin himself. Didn't think that would be a question? EDIT - scratch that; I see you were asking re: western vs eastern. I'm thinking of ditching that stuff altogether. I wasn't comfortable with it to begin with and it's beginning to become an issue as you can see.

EDIT2 - You know what, you guys are absolutely right. After re-reading it (again), that didn't have any place. It's gone.

[faster3200]

How does it prove that the 355 is better for loops that need strong pressure?

Compare this graph to this graph.

Sorry, didn't answer the block comment. Calling those high flow & low flow (less restrictive & more) were based on the comparison by Martin himself. Didn't think that would be a question?

I didn't mean to say that calling the low flow blocks low flow and calling the high flow blocks high flow was wrong. That whole part about the blocks was in reference to:

Low flow is popular in Europe and the far east. The object of this is to use the blocks to slow down the flow rate of the fluid in order to pick up as much heat as possible, thus using the radiator as the main source of cooling. Problem is, Chinese radiators aren't the best, so that system doesn't work very well.

My point being that the majority of European made blocks tend to be high flow, thus this would seem to me to be their acknowledgment that making a high restriction block purely for the theory of decreasing flow to decrease temps doesn't make any sense. The reason Euro blocks tend to be high flow is that they use weak pumps, thus they need less restrictive block to not get destroyed by flow scaling. Just as an example Eheim pumps are far more popular in Germany than here, which iirc one of the more popular ones has about half the pressure of a stock DDC 3.1 (or DDC1 for you).

Also, a thing you might want to add is that a DDC1 can be modded to a DDC1+ rather easily, so it might be a good option for the budget conscious. I know here a 3.1 can be had for $50, while a 3.2 is about $65.

[hokiealumnus]

Sorry, I'm typing over top of you. In my edit I basically say: "I agree"

And thanks to both of you and anyone else that chimes in for your help. Please don't think of my answers as being argumentative. If I'm asking a question, it's just that, a question. I want this thing to be as accurate as possible.

[alacheesu]

In response alacheesu:

1. [..]Are there enough cases with room for internal rads that do not require any modding to give it its own section?

I see what you're saying, but most people either mod their case or use a case like TJ07, Cosmos S, one of the Lian Li's etc. that fit rads with little trouble. Taking a look at the worklog section, it seems most people prefer a normal, full tower case and do a tiny bit of modding to fit the rad(s). I guess it's impossible to have everything in a basic guide, though, so I'm not sure what I'd do.

2. FWIW, I wanted to link to the manufacturer sites. The links there were on request by the owner of the site, Gilgamesh. They are a sponsor and supply us with some of our hardware for reviews. They're also listed in our sponsor list, linked from the main menu. Not trying to hide that in any way whatsoever. The site has to get hardware somehow. I try and make myself feel better by saying it was 'selective linkage' because they're a sponsor, not necessarily advertising.

Heh, OK. My issue wasn't really that you linked there from your own page, but since you came here to have it included in the stickies on XS that's a little different, IMO. Since I'm not an XS mod it doesn't matter what I think, though.

3. Really? Stock specs put the 655 vario @ 50PSI and 355 @ 22PSI (per sidewinder). I wouldn't think a top could more than double the available pressure. Maybe I'm looking at it the wrong way?

The P-Q curve is what's important. Take a look at Martin's old site. A couple of numbers from there:
MCP355 w/XSPC res top: 1.0 gpm, ~6.5 PSI
MCP655: 1.0 gpm, ~4.5 PSI.

4. FWIW, that was verbatim from Gilgamesh. He knows a lot more about water cooling than I do and I just put it in there without questioning. Faster: This applies to the Chinese rad comment as well. Boss man get whats boss man wants.

I'm still hoping one of the experts come in on that one. :P I have my ideas, but again, I'm not too confident in my knowledge on block design.

[W1zzard]

very very good spelling, better than what most of my guys deliver as "ready to be posted"

- pictures in the case section would help
- move the case section further to the end, when people already have a very basic idea of how watercooling works and what components are involved = which have to be fit into a case
- your pics need to be resized to the right size
- res and others: first explain what it does, why you need one or why you dont

[Waterlogged]

It should be noted that those PSI numbers for the pumps are what the housing is rated for, not what the pump will produce. These pumps are also have other uses, some of which include being installed into lines that connect to municipality water lines.

I would also removed all geographical references such as US/UK/Europe. Let the reader decide what type of cooling (high flow / low flow) they want, seeing the regional stuff may make them think that it's all they can, (or worse) should get.

[hokiealumnus]

W1zzard: I'll work on some case photos. Also edited all images except for the fittings; they are now thumbnails to the bigger versions. Does that address your concern? Good idea on telling what the components actually are for. Figured it was evident, but shouldn't take that for granted in a beginner's guide. Point taken on case placement. I'll give that one some consideration.

Waterlogged: duly noted. The references to localities are already gone from the block section.

Thanks to you both as well!

[MpG]

Just a few little points:

As a side note, there is one exception to the rule that you must use clamps with barbs. This is using 7/16"ID tubing on perfect seal 1/2" barbs.

On a list of barbs that I would be even remotely comfortable going clampless, the "Perfect Seal" ones are definitely NOT on the list. What little barb there is has very little edge to it, and once the water (and tubing) warms up a little, you're just begging for a wet tube to shoot off the barb, especially with higher pressure pumps.

I've also seen diluted methanol used (windshield washer fluid), though that one is slightly flammable and I'd not use it personally.

Alcohols don't mix with acrylics either. Their only real practical use these days is for sub-zero applications.

[hokiealumnus]

Compare this graph to this graph.

Ok, I'm very confused. After reviewing those graphs, why is the 655 reputed to be such a better pump? Isn't there a point in a low restriction system that the law of diminishing returns comes into effect for GPH? Is it necessary to really move that much more water? I will say with that data, I'm even more pleased with my 355 + XSPC res top than I already was!

[MpG]

Compare the stock 3.2/355 curve to the D5, and you can see why the D5 is still going strong. Amazingly enough, not everybody is topping their DDC pumps. That's what I hear, anyway. There was also some business a little bit ago with the DDC-2's randomly kacking, which wasn't good press.

As far as flowrates go, most blocks tend to show diminished returns past the 1.0gpm mark, and even the most restrictive blocks are definitely tapering off by the 1.5gpm mark. But there are still slight performance benefits to be found at those flowrates, and in a series loop, a faster flowrate can help minimize the effect of earlier components heating the water for later components.

[hokiealumnus]

Ok, changed the wording to the following, tell me what you think:

Truth be told, any of these pumps will serve most water loops with a block or three in it. If running a high restriction loop, counterintuitively, the MCP355 would be the way to go as long as you went with a quality aftermarket top. See this graph on the D5 vs this one for the DDC. You'll notice that per-gallon/hour, the DDC with a good aftermarket top gives more PSI than the D5. Now, if your loop is not restrictive and you want to move mass amounts of water, the D5 is the choice for you.

Links are in the proper place of course.

[faster3200]

Ok, I'm very confused. After reviewing those graphs, why is the 655 reputed to be such a better pump? Isn't there a point in a low restriction system that the law of diminishing returns comes into effect for GPH? Is it necessary to really move that much more water? I will say with that data, I'm even more pleased with my 355 + XSPC res top than I already was!

I'm not sure where it is reputed as a better pump. It certainly isn't on XS. I would say it is better for chiller usage though as it is better built, although neither pump have high failure rate for what we are doing. If it does have a worse reputation it must be from the stock top, which as you can see by the graph isn't up to snuff. I am happy with my pump as well. That combo is basically as good as you can do for a single pump <$200.

It depends on the block, but I would say you start running into very diminished returns somewhere around 1.5GPM or more. Still, why have less when you can have more for basically free?

EDIT: MPG beat me.

[Kram]

The curve of diminishing returns is actually much lower down, around 1.5-2 Liters/minute. Between 2 and 4l/m you have a difference of 1-2c. Above 4 l/m it will be 1c or less and the pump heat will outweigh it.

If you are getting 1.5gpm! (6 liters/minute) then the extra flow is not free, it costs heat into your loop, you should turn the pump down.


You have heatkiller as a high flow block, when its one of the more restrictive.

Also for tubing, you have this which I think is wrong:

1/4" & 5/16" are all but extinct at this point. There is still equipment available that uses them, but it's mostly to fix old loops.

Its true they are not used much, but this will change as blocks are getting far more restrictive with FAR higher pressure pumps, the tubing size has became *compleatly* irrelevant for most people. It makes very little difference to temperature but its used for compatability.

So with 1/2", you might gain 1-3c at most, but you will have to use large fittings, 90degree bends and it will take up most of your case.

1/4" small fittings, you can do 360 degrees in the space of a 1/2" fitting. you can g-clamp the hose if you want to disconnect something without draining, and it will take less space than aircooling.

a faster flowrate can help minimize the effect of earlier components heating the water for later components.

The water is already moving so fast that the dT is under 0.5c.. so that is compleatly wrong.

[faster3200]

The curve of diminishing returns is actually much lower down, around 1.5-2 Liters/minute. Between 2 and 4l/m you have a difference of 1-2c. Above 4 l/m it will be 1c or less and the pump heat will outweigh it.

If you are getting 1.5gpm! (6 liters/minute) then the extra flow is not free, it costs heat into your loop, you should turn the pump down.

That is not necessarily true. Between 1GPM and 1.5GPM you will gain around 1C depending on the block, after you count the extra heat from the pump. I am pretty sure most people here are running DDC 3.2s which only have a thermal output of 18w, not all of which will end up in the water. That is pretty negligible.

I think it is all about what you think is acceptable price/performance. The gains do diminish, but they are still gains.

Also for tubing, you have this which I think is wrong:


Its true they are not used much, but this will change as blocks are getting far more restrictive with FAR higher pressure pumps, the tubing size has became *compleatly* irrelevant for most people. It makes very little difference to temperature but its used for compatability.

So with 1/2", you might gain 1-3c at most, but you will have to use large fittings, 90degree bends and it will take up most of your case.

1/4" small fittings, you can do 360 degrees in the space of a 1/2" fitting. you can g-clamp the hose if you want to disconnect something without draining, and it will take less space than aircooling.

I would argue that 1-3C increase in temps because of a tubing choice would be unacceptable to most people.

[NaeKuh]

Here is my guide i wrote a long time ago.

http://forums.anandtech.com/messagev...&enterthread=y

Now what i love about my guide is that its very simple and basic. There is no block or part comparison so it takes a lot for it to get out dated.

I feel there are 3 kinds of guides you can write and maxxracer nailed all 3 of them.

the first is the general FAQ or a newbies guide that basically answers questions or teachs you basics. (that is what my guide was modeled off of).

Then the second is a parts FAQ, this would have all the parts that is current and a review to the parts. (this requires too much work to maintain.)

The last would be whatever you didnt include in the first two.

its important that you dont jumble all 3 faq's in 1 big faq, because you'll lose your audience half way down the FAQ.

[MpG]

The water is already moving so fast that the dT is under 0.5c.. so that is completely wrong.

At 1gpm, 266W of heat will give you a dT of 1c. So on a heavier loop (i.e. OC'd proc+SLI/XFire), a dT of a couple degrees isn't unreasonable at all. Boost the flow to 1.5gpm, and it now takes about 400W before you have a dT of 1c. Not to mention that the increased flowrate will help the radiators shed the heat a little faster too. They're all incremental gains, but they add up, and they're available for those who want them.

[shazza]

...
its important that you dont jumble all 3 faq's in 1 big faq, because you'll lose your audience half way down the FAQ.

That's a FAQ

Nice job on the guide ... we are all learning as you go through and get comments on it. Let's only hope the new users will take the time to research and read it!!!!

[Brodel]

The Heat Killer is a available from a few places in the UK. Also, the UK is part of Europe, so saying that the UK likes one thing and Europe likes another is a bit inaccurate.

http://www.overclock.co.uk/product/W...lock_8744.html
http://www.tekheads.co.uk/s/product?product=610430
http://www.chilledpc.co.uk/shop/inde...sort=2a&page=2
http://www.thecoolingshop.com/produc...oducts_id=3583

Hope that helps.

[hokiealumnus]

Thanks shazza!

Brodel: Yea, I dropped all locality language. Thanks for the links, but I've been told to keep specialtech the only linkage in the UK by the boss-man. As you can tell, it's readily available. I'm not discouraging people from going elsewhere, just can't link in the article. Sponsorship and what-not.

I've integrated a good bit of changes at this point, thanks to the good folks here and @OCF; just need to rearrange it to put cases further back and go into some more install detail and it should be pretty much done. Thanks again for all your help!