wb performance graphing

[BillA]

as part of the ever slightly closer to completion wb Test Methodology article,
am intending to use this graph:



it works well, but my question is:
what are the 'best' (-> most useful) units for the x axis ?

being a mossback, kPa are quite useless for me
psi I can understand (and use for measurement, the indicators are scaled that way)
BUT pumps are typically described in 'feet of head'
or 'meters of head'

as many use Eheims, should I use "m H2O" ?
if so, should it then also be retitled "C/W" vs. Head Loss ??

- understanding of course that a portion only of the pump's capability can be allocated to pushing the coolant through the wb

[MadDogMe]

Hmmm?, do both . I'd prefer head because it relates to what we most see. is there a conversion tool or equation? you can tag on the end?...

PS. wish I could see the graph!! . where will the article be hosted?. OverClockers.com?, Here?, or both?...

PPS. Are the graphs for the Rad tests at OC'ers not the revised/correct ones?...

[myv65]

Hey Bill,

With results as shown on that graph, who cares? Just say, "They're all so close together it doesn't make a lick 'o difference which one you use." As a footnote you might add that Cathar's block kicks all their tails.

Or how about, "Given the skill required for a good paste installation, the differences between blocks is below the threshold of significance."

To your original question, I think the proper answer is "flow" on the X-axis. Whether gph or lph doesn't much matter. The problem with pressure is that no one could or would measure delta-P on a block. What they can measure is overall system flow rate.

Whether you go with flow or pressure, you really need an additional graph that ties the two together. How is anyone to know how much delta-P (or flow) they will get with a given setup? At least with flow they can measure it. If they have your data directly in flow, great. If not, then they would need the second graph to tie flow to delta-P.

[bigben2k]

Personally, I really prefer the cw vs flowrate (GPH or GPH). How this flowrate is achieved, is up to the user to put together.

The problem of course is that the headloss/pressure drop across the rad is mostly unknown (for common heatercores at least), except for what you've posted, which unfortunately doesn't include the latest BI products. If those were available, then, and only then would the above graph be usefull, IMO.

My .02

[gmat]

m of H2O sound nice.
That way we can add restrictions (all in m of H2O) and compare to pump head, and even point directly on pump diagrams to find resultant flow.

[BillA]

Originally posted by myv65
Hey Bill,

With results as shown on that graph, who cares? Just say, "They're all so close together it doesn't make a lick 'o difference which one you use." As a footnote you might add that Cathar's block kicks all their tails.

Or how about, "Given the skill required for a good paste installation, the differences between blocks is below the threshold of significance."

To your original question, I think the proper answer is "flow" on the X-axis. Whether gph or lph doesn't much matter. The problem with pressure is that no one could or would measure delta-P on a block. What they can measure is overall system flow rate.

Whether you go with flow or pressure, you really need an additional graph that ties the two together. How is anyone to know how much delta-P (or flow) they will get with a given setup? At least with flow they can measure it. If they have your data directly in flow, great. If not, then they would need the second graph to tie flow to delta-P.

that the 'good' wbs have similar performance I would think is fairly well understood, given how often its been stated
(less than 0.5°C temp difference, below the threshold of computer user's measuring equipment)

likewise wrt the TIM joint influence, seemingly less understood (or accepted ?) despite even more repetitions

the conventional graphing is of pressure drop vs. flow, and "C/W" vs. flow; the above graph being simply a systhesis of the two into "C/W" vs. pressure drop,
- with the intent of looking at the specific efficacy of different wbs 'utilization' of the available head
(bearing also in mind that this article is on wb testing, as contrasted with WCing system 'design')

I'm kind of gathering that the potential interest group (wb designers) is so small (and also so unsophisticated) that the added complexity is not warranted
ah, well

Ben
anyone can get a rough handle on the rad pressure drop (and all other parts of their system) with a simple manometer
but other than Owenator no one has enough interest to do so

MDM
its your ISP, deal with it
the revised rad graphs are only at that link (same site)
if I can find a new hosting provider perhaps that will solve your difficulty

gmat
kind of what I too was thinking, and why I asked

[myv65]

Originally posted by unregistered
- with the intent of looking at the specific efficacy of different wbs 'utilization' of the available head
(bearing also in mind that this article is on wb testing, as contrasted with WCing system 'design')

Ah so. Then I agree with gmat.

[Les]

Units - "Metres Water" is fine, as long as I am allowed roughly convert to KPa by x10.
Title - probably "Pressure Drop"
Have been wrestling with this one for sometime and usually duck the issue by using the same as my data source. Luckily have no "Squish" data of my own.

Thanks for posting graph . Although you have posted "Flow v Pressure Drop" for various wb configurations,I was unsure which Swiftech462 data to use. Not located any "Flow v Pressure Drop" curve for "the Hoot creation", have I overlooked?

[g.l.amour]

myv65 has a point to note where the relevancy of it all is. but these graphs (combi of them) can give u a pretty good idea why the individually good components do lousy, in a particular setup. so i don't like to read between the lines that BillA 's work should not be appreciated.

[bigben2k]

Actually, I think that some people appreciate it very much. I suspect that some people are just not able to use those numbers, simply because there's no step-by-step instruction on how to use it, when they're face-to-face with the choices out there. In other words, the manufactures/distributors are at fault here.

I got a pump, and I did not consider what block/rad combo would go with it. If I had specs, I would have selected my pump more carefully. Right now, it stands with a max head of 14.4 feet, and will achieve more than 500 gph at zero head.

The good thing is that I can estimate my expected flow rate.

Back to BillA's Q: if I had to choose, I'd like to see psi as a pressure unit. It's not metric, as gmat would propose of course, since he's in France!

[dream caster]

I prefer head in meters of water (feet is also useful);
If you could provide us a head flow graph, would be useful because that is beyond most of us can do.

[BillA]

g.l.amour
people 'approach' WCing from different perspectives,
myv65's is engineering which will quickly discern a 'point of diminishing returns'

mine is as an 'experimentalist' (so I have been told), so my interest is in deriving and examining data
from which I deduce some things, and predict others;
but 'need' theory only to rationally understand relationships that may be too complex to directly test

dream caster
this more useful ?


should make a comment:
just added a 462-UH that Gabe sent me, and am stupefied by it's incredibly low flow resistance !
granted, I used 1/2" NPTx1/2" Cu pipe connectors (with 5/8" hose) - but even so . . .
while I suspect it may not cool as well as a 462-U at the same flow rate,
it will seriously impact the efficiency graph that started this thread

I'll re-do it for mH2O now, and work on the 462-UH's C/W #s tomorrow



better ?

[Cathar]

Hmmm, I'm not fusses whatever pressure units are used. If all it means is that somewhere in the article is listed the conversion actors to go to one or two other commonly used units easily, then it's all good.

Pressure is pressure, no matter what units it's stated in. Stick a quick converter footnote, and within 10 minutes anyone who cares will be fluent in swapping between the units.

Using flow rates only is kinda sucky because its gives no information as to whther or not that flow rates is achievable with some random pump.

[Cathar]

That pressure-drop vs flow rate graph is very nice - exactly what I'd be looking for as a useful guide of what pressure translate into what flow rate...

[BillA]

"some random pump" is not a good starting point
if it's got no P-Q curve -> don't buy it

[Cathar]

Originally posted by unregistered
"some random pump" is not a good starting point
if it's got no P-Q curve -> don't buy it

Of course. I really meant "some random pump off the shelf with a PQ curve" so one could look at your data and determine where the pump will roughly sit.

[LiquidRulez]

Originally posted by myv65
As a footnote you might add that Cathar's block kicks all their tails.

Which one is Cathars block C/W?

[gmat]

Originally posted by bigben2k
It's not metric, as gmat would propose of course, since he's in France!

Not me alone. Everyone around the world...

[bigben2k]

Originally posted by gmat
Not me alone. Everyone around the world...

Including Canada, but not the USA (now why is that again?)

[Skulemate]

I'm not sure about that gmat. As a Canadian I am familliar with both systems to an extent... that said, I'd rather have imperial units than the usual "soft" conversion that we get here... (38x89 studs and the like... ).

[Blackeagle]

Because the U.S.A. is stubborn and resistent to change. Never mind that it would improve our competitive standing in marketing with/vs other nations.

It wouldn't be easy for those in their 40's or older, but could be handled. And I hate to see my two daughters now being taught the same outdated system of messures. By the time the US converts my grandkids may have to relearn a new system.

[g.l.amour]

stupefied by the UH indeed. as most of us can't easily buy high head, low volume pumps; that block seems the one to go. i'll await some other tests from billA, but if they turn out the way i expect them to turn out... i could use one

[utabintarbo]

Originally posted by gmat
Not me alone. Everyone around the world...

We were talking about the civilized world!

[Cathar]

Originally posted by LiquidRulez
Which one is Cathars block C/W?

Now that the bits and pieces are coming together for my blocks, I'll send one to BillA. Currently only two blocks of the final design parameters exist, one's with me, and one's with a very eager first customer who picked one up from me last night.

He just emailed me and said he was seeing around a 5C improvement over his old Silverprop Cyclone 5 when running Folding @ Home on a AthlonXP running at 1880MHz/2.0v, which is the sort of results I was seeing in my testing with the production block (and about 1C better than my prototype).

In my testing the Maze 3 (properly lapped and mounted) is almost inseparable from the Cyclone 5 (also properly lapped and mounted), and the Cyclone 3 which is a near Swiftech clone also performs about the same as the M3/C5.

So yeah, I'll send a complete final block to Bill within the next 7 days. I'm keen to see what he sees.

I know that for me that the thermal juction is now starting to become the predominant factor (barrier?) in improving performance further.

[bigben2k]

Originally posted by Cathar
Currently only two blocks of the final design parameters exist, one's with me, and one's with a very eager first customer who picked one up from me last night.


I know that for me that the thermal juction is now starting to become the predominant factor (barrier?) in improving performance further.

You mean Bass-Injected? What a nut!


That sounds about right. Speaking of which, I'll be trying to re-create Les' thermal calculations on your block, so that I can try to apply it to mine. Any help is welcome!


Ok, back on topic. Come to think of it, I think I prefer head, in feet. That way, I can compare it directly with a pump, and do the conversion on the rad, from numbers available, wherever that is!