Cathar(Forster)

It'd be nice to get it perfect first time, every time.

Apologize for the wrong name.
Thought it was Foster, my family name is Round.

I wasn't sure on the initial comment. Was unaware of your surname, and what I could have been mine was mis-spelled. I was wondering if you were referring to some British commentary panel show or something where 3 guys sit behind a desk and talk shit, but sometimes show a modicum of wit.

Then again, that's probably exactly the meaning you were conveying, just that it wasn't some British TV show. :dome:

lol
the 3 techno stooges
gotta love it

In Wigan we play RL.

Perhaps refering to "Doctor Foster went to Gloucester"

Why did you choose this baseplate thickness? What's stopping you from using a thicker one?

An increased cost of conduction is why. As Les's graphs illustrate, when the rate of convection get significantly above 50000W/m²K, base-plate thicknesses below 2mm are pretty much required to get the most out of the design - eventually the copper starts to get in the way. The higher the rate of thermal convection immediately above the CPU die, the less the lateral thermal spread, until eventually all the copper is doing is just getting in the way, rather than actually aiding in any significant fashion.

Update
 

Okay, picked up the replacement XS middle plate two days ago. The replacement plate was machined out of Delrin, and it turned out excellently. Delrin is wonderful stuff for machining, the only drawbacks being that it is not a clear plastic. Still, it does come in black, so a waterblock made from black Delrin would have the appearance of those cool looking black Sony Playstation CD's.

That's all just modding fluff though.

Comparing the copper prototype XS to the Cascade SS (Silver-based modified Cascade).

Now I've picked up a motherboard that reads the on-die diode from the AthlonXP CPU's. The drawback to this is that the temperatures being reported are pretty low, probably about half of what I would've expected to see, and keep roughly in proportion to the sorts of values that Phaestus is seeing in his test-bed.

Using the radiators I've been able to determine that the CPU under the test load conditions is dumping approximately 120W into the cooling loop, but the CPU temperatures are being reported as around 7-8.5C above the water temperatures, which is plainly ludicrous, and should at least be twice that. So much for on-die diode accuracy...

Okay, so using the on-die diode and overclocking stability to "measure" waterblock performance, it goes roughly like this:

AthlonXP @ 2.66GHz/2.2v running Prime95 Torture Test with fixed highest-heat-load test pattern. Prime95TT is not as heat-heavy as BurnK7, but for this particular test CPU, BurnK7 will cause the system to reboot at anything over 2.55GHz, which is pointless for overclock stability testing.

Okay, the results after a number of test runs. Basically reporting the best achieved result. Most runs were either at or just below the following figures.

Cascade SS - 52 jet-tube, silver base-plate
~8.5C CPU rise above water
peak stable 1hr overclock 2685MHz @ 2.2v & 26.0C water temperature

Cascade XS - 149 jet-tube, copper base-plate
~8.0C CPU rise above water
peak stable 1hr overclock 2690MHz @ 2.2v & 26.0C water temperature

The 0.5C difference here probably equates to a 1C difference in reality, but I can't be sure of that given my equipment. That's purely a guess.

Made in silver, the XS would probably offer another 0.5C performance increase, and probably another 5MHz or so. The temperature results are not wholly unexpected, however I was somewhat hoping for a better overclock.

Ordinarily I probably would've proceeded to get a silver XS base-plate made-up, but then a dark horse rode in from over the stormy horizon...

hmmm... don't think you're going to break on this one, but how about droping some hints about that p2, i don't think it has been mentioned on this forum yet.

with the delrin is the xs looking like a viable option? or are you going to scrap it and focus on the p2?

Oops. Got distracted by watching "Scary Movie 2" before completing the second post. :rolleyes:

With the Delrin, the XS is a viably machinable option. Actually I can't get over how clean it looks in comparison to the polycarb. Where the polycarb was visibly "stressed" from the machining process, with more than a few split tubes (not that the incorrect cutter depth helped much there), the Delrin looks near perfect. Some slight "fluffiness" at the edges of the tubes which can be expected on something this fine, but 90% of the tubes look perfect, with only 10% or so with fluff that needed to be cleaned away, and once done, also look perfect. Truly the piece is somewhat like looking at some sort of optical illusion, with the detail so small as to trick the human eye without some magnification assistance.

However, having said all that, it looks like I'll be scrapping the XS, in favor of the dark horse.

A while back at OCAU I hinted on a cheap to make design that would match a White Water, but would be far cheaper to make. The design started out life originally as one of my previously unshown GPU block prototypes. I scrapped further work on it due to limitations that I'm hesistant to explain here, as even saying that much may be giving too much away. Well I decided to re-prototype it, with a few added tricks and tweaks, but basically focused around a budget-block frame of mind. In simulations the figures hinted at something very good and I took the design a little further (I can't resist), not really trusting the figures so much in terms of absolute performance, but at least as a way of improving performance without really adding much cost.

Anyway, that block is what I've referred to as the Prototype #2, or P2 for short. The figures hinted at substantially better than Cascade performance within certain constraints, but overall I was basically expecting more like White Water performance.

I got the P2 and stuck it on, and on the first mount I was seeing (on the same test-bed as above) 1C lower CPU temperatures than with the Cascade SS, and ~5MHz shy of the Cascade SS's overclocking performance. I pulled the block off and noticed an uneven thermal paste imprint, which meant that there's basically more to offer. I remounted a couple more times and on the third try, the temperatures remained the same (1C better than the Cascade SS), but was now seeing 10-15MHz higher overclocks.

So basically the copper-based P2 is significantly out-cooling the Cascade SS, and allowing for stronger overclocks, however small the gains may be, it's still forwards progress.

Given the choice between the near-equalling performance but very expensive to make XS which is at the absolute end of the road for further improvements with it, and comparing that to the ever-so-slightly better P2 which is very cheap to make and has a lot of room for improvement (at a dollar cost), then basically this makes the XS a redundant block.

I can't believe it. Your actually getting close. Keep thinking. There is away and it has to do with the cups. Think about making this "secondary impingement effect" last longer within the cup. I would venture to say, the cups in your cascade are less than optimal but your limited in design choices being that small. This is something I'v been working on for the past year and a half trying to get it right with my limited machining , money(kids gotta eat). Your WWW is what got me thinking about a pure impingement block. But as you stated in your post, being restrictive has been the problem for me. The performance is shocking though paired with a strong pump. There is one more step beyond that can aid convection even more that was discussed between us last year. Making it practical is the challenge but it does work well.

thanks alot, that was alot more then i had hoped to get out of you.
not sure if you'v covered this or already refused to devulge, but have to ask anyways, can you give us a remote time frame to a first glimpse at the p2, i know it depends on alot of variables, many of which are not directly under your control, but can you let us know if you're looking at 1 month or 1 year until we can start to get a glimpse at this mystery block?

great work! sounds like it's going to be a winner, and also good to hear about the success witht the xs, even if it won't happen, always good to push the limits a bit.

Thanks Syscrusher. I have explored lot of options with the Cascade approach including what I described above, but have settled on this P2 thing as being the best way forwards from here.

As much as BillA will hate me saying it, am fairly sure I'm seeing h of over 100K here (on average), and am looking to boost that to the ~120K mark with the next set of enchancements.

Once I reach that level and am confident, then I will reveal all.

Honestly from the way you describe the P2 on OCAU as being so much simpler and also such a good performer I really hope to be able to see it someday. There is an elegence in something that achieves so much with very little. Market wise I'm sure it'll be the winner, a Lamborgini @ Volkswagens price hopefully. While the alure of the all out high dollar Uber block with the flashiest design and highest price tag will still be sought after by those more concerned with bling than performance, I think the mainstream enthusiast is more concerned with the almighty c/w rating and how much they have left over for some Tygon. Pesonally I spent pretty freely and just want the most bang, not the most bang for my buck, and from scoping posts for the past few years I am a lil' outside the mainstream. The P2 sounds as if it'll cover all three bases though. The most bang, the most bang for your buck, and the bling of owning a Cathar designed block.

Okay, it's really late here and I probably shouldn't give even this much of a hint away, but here it is with my mind being tired and all. Nature is such a wonderful teacher.

"What is one of the most violently turbulent and destructive forces in nature?"

WoooHooo TORNADO???!!! Run for your Aunty M DD

I would lean more towards a hurricane or typhoon. ;) Anyways, it sounds very interesting to say the least from the price/performance standpoint. Also the design....trying to think of what you are doing with the block design and not quite sure :)
Edit: maybe a waterspout? A tornado of water :eek: http://www.southdown-amateur-radio-s...aterspout.html

Ahhh but what natural occurence produces the highest velocities and LOCALIZED TURBULENCE ? Thats what impingement is all about.

Tsunami?
Taxmanian Devil?
Thermonuclear weapons?
Drunken frat boys in office?

A woman with PMS?

In that case the most effective form of cooling would be the "chill" that occurs before the storm. I did say that the design is nature based though, not chaos theory based.

On a side note, it appears that the Tasmanian Devils are all about to die out. There is some mysterious disease which is affecting the population. Their mouths start off with sores which progressively gets worse until they can't eat and die. This has just occurred in the last six months or so, and it appears that none of the Devils have any immunity to it. If you ever wanted to see one in real life, then in the next few years may be your last chance.

Their situation is also not helped by some idiot in the last year bringing over foxes into Tasmania (which was previously fox free) to hunt for sport, and the foxes have now started to multiply and kill off the local Tasmanian wildlife, which was one of the last most natural and unspoilt fauna reserves in the world.

Of the items on your list, the last two are basically human-based, and it seems sad that perhaps that may be the true answer to the riddle....

Well that was my depressing thought for the day.

Hurricane!!!

Hmmm, swirly bits...

Anyone ever gotten water to spin rapidly, and then watch for how long that characteristic continues even after turning corners?

Hmmm rifled jets?

Hey...does that mean you'll have to make 2 versions- one for south of the equator and one for above LOL

Nope. Olympics anyone? :dome:

And it has to swirl faster if you taper your chamber towards the outlet (conservation of angular momentum).

Ooh! I know! Mini spirals! (Slaps self, so others don't have to) jlrii: If you're forcing the spiral, it's irrelevant.