Pro/Forums

Pro/Forums (http://forums.procooling.com/vbb/index.php)
-   General Liquid/Water Cooling Discussion (http://forums.procooling.com/vbb/forumdisplay.php?f=9)
-   -   Jet impingement nozzles - defiling my white water block. (http://forums.procooling.com/vbb/showthread.php?t=5141)

Since87 12-05-2002 11:02 PM

Jet impingement nozzles - defiling my white water block.
 
1 Attachment(s)
I can't help tinkering with my white water block before I've even installed it in a system. I took a reamer to the outlet barbs. I've opened them up fairly substantially and nearly gotten rid of an abrupt decrease in diameter that these barbs have.

The question I have now involves a potential mod to the inlet portion of the block.

I've done a little research on fan shrouding and pump inlet design issues, and it seems pretty clear that abrupt changes in the shape and size of the waterflow path are undesireable because of eddys that result in wasted pressure drop. (Anyone who can phrase this in a more technically correct way, please do so.)

I'm thinking that if I could make some sort of insert to fit in the inlet barb, and smooth the transition from the circular cross-section of the barb, to the long slot in the nozzle plate, it might be of benefit.

So:

Would it be of benefit?

Would it only be of benefit if the shape was exactly right to 1 mil tolerances?

Would the shape be dependent on the flowrate?

Anybody able to put a ball park quantity on the amount of benefit that might be achieved? I realize I haven't provided any dimensions here, and I won't be providing any dimensions without Cathar's consent. I'm just wondering if someone (hmm Dave maybe) could easily give an educated guess at this.

Anyone have any suggestions on how to make an insert like this? (Looking for a DIY method, not "Draw it up on a CAD system and have it milled." Modelling clay, plaster mold, cast JBWeld, sounds feasible to me.)

TIA

Cathar 12-06-2002 12:04 AM

Before you do anything further - please install and test it first - that way you at least know if you're making things worse or better...

As a side thing, I ran a series of tests using the nozzle plate under a barb external to the block to what the flow from the nozzle. Basically it comes out pretty straight. The flow resistance of the block is fairly high due to the restriction, but not stupidly so. Flow rates were always about 15-20% lower than other very free flowing blocks (Maze 3, Cyclone 3, Cyclone 5) depending on what pump is used.

Given that you're fighting the restriction you may gain maybe 5% greater flow rates than in the configuration that it's currently in. The performance difference for such a small flow rate gain will be so low as to be undetectable on anything other than BillA's testbed (<0.1C is my educated guess).

I've already found that the nozzle ejects a straight "clean" flow straight down even despite being off-centre (for AMD CPU's) so you won't be gaining anything in that respect.

What may be of more benefit is having a 1-2mm "lip" hanging down from the bottom of the nozzle plate that partially extends into the channel areas to create something that looks more like a nozzle (1 per channel)

This was something that I considered but could not find a way to cheaply machine it without adding an extra $20 on top of the cost.

bigben2k 12-06-2002 09:18 AM

1 Attachment(s)
I hear you, Since87, and I agree 100%.

When I designed the inlet to Radius, I looked high and low for ASME nozzle specifications, and I found them (check the thread, last page).

In short, a straight (abrupt) change in diameter will cause the pressure to decrease by 80%, from inlet to outlet. With an ASME nozzle, the pressure drop falls within a range of 60 to 80%, so it can be up to 25% less restrictive.

In White Water, I would extend the visible fin tops into a blade shape. I will also be doing this with Radius.

Blackeagle 12-06-2002 05:15 PM

A good point about the fin tops Bigben. But how much gain vs costs would this add to a block that Cathar may start to try to market beyond recouping of his costs of his research expeditures?

It seems that each new design sparks more refinments and improvements.

This is always a interesting place to spot into.

bigben2k 12-06-2002 05:29 PM

I think Cathar covered that: it's cost prohibitive.

It doesn't stop anyone from manually fixing it though.

One more thing: the inner surface should be smooth as glass! A rough surface is a great restriction (even though (although?) it's over a very short distance).

To answer your question Since87, I'd go with either a 2 part epoxy block, shaped to fit (drop-in), or resin.

Since87 12-06-2002 10:42 PM

Quote:

Originally posted by Cathar
Before you do anything further - please install and test it first - that way you at least know if you're making things worse or better...
As you point out, short of sending it to BillA for testing, I won't likely be able to determine the result of any mods I do. I've been assuming I could use the results Bill gets, for the block you sent him, as a baseline if I ever wanted to have my mods tested. Is there any reason my block (pre mods) wouldn't match the one you sent to Bill fairly closely?

I don't have anything to test this on myself. I'm on a Celeron 300A @ 450 right now. (Waiting on Granite Bay and Barton right now.) If I come up with something that I think will make a definite improvement, I may send my modded block to Bill.

I think I'm pretty safe in reaming the outlet barbs. I doubt it makes a very significant difference. If there is a way it could be detrimental, I'd be interested in learning how.

Quote:

Originally posted by Cathar
Given that you're fighting the restriction you may gain maybe 5% greater flow rates than in the configuration that it's currently in.
Because I'm hoping to put a microchannel GPU block in series, I think I'll be wanting as little restiction within the CPU block as I can get. Any idea how restrictive your GPU block is going to be relative to the CPU block?

Quote:

Originally posted by Cathar
What may be of more benefit is having a 1-2mm "lip" hanging down from the bottom of the nozzle plate that partially extends into the channel areas to create something that looks more like a nozzle (1 per channel)

This was something that I considered but could not find a way to cheaply machine it without adding an extra $20 on top of the cost.

Is that $20 AUD or USD? If it got an extra 0.01 C/W, (At an acceptable pressure drop.) I wouldn't balk at $20AUD. At 0.005C/W I'd give it a lot of consideration. If you think a fairly substantial improvement in performance could be achieved, it might be worth considering an optional nozzle plate. I really really hate to say this, because I love the fact that you discuss this stuff in public, but maybe you should keep these ideas to yourself. :(

Since87 12-06-2002 11:09 PM

Quote:

Originally posted by bigben2k
In short, a straight (abrupt) change in diameter will cause the pressure to decrease by 80%, from inlet to outlet. With an ASME nozzle, the pressure drop falls within a range of 60 to 80%, so it can be up to 25% less restrictive.
I'm not sure how relevant the numbers for an ASME nozzle are since, in this case, the nozzle would be going from a circular cross-section to a roughly rectangular cross-section.


Quote:

Originally posted by bigben2k
In White Water, I would extend the visible fin tops into a blade shape. I will also be doing this with Radius.
I'd thought of the nose cone idea as well, but I haven't figured out a means to attach them without serious risk of messing up the microchannels, or rather extreme difficulty. In addition I'd guess that, unless the nose cone is soldered on copper, you lose some cooling from the water hitting the top of the fin. Does the improvement in flow counterbalance that loss?

Something like that could be integrated into the barb insert I was proposing, but that's well beyond my fabrication capabilities.

bigben2k 12-07-2002 11:02 AM

Quote:

Originally posted by Since87
I'm not sure how relevant the numbers for an ASME nozzle are since, in this case, the nozzle would be going from a circular cross-section to a roughly rectangular cross-section.

I'd thought of the nose cone idea as well, but I haven't figured out a means to attach them without serious risk of messing up the microchannels, or rather extreme difficulty. In addition I'd guess that, unless the nose cone is soldered on copper, you lose some cooling from the water hitting the top of the fin. Does the improvement in flow counterbalance that loss?

Something like that could be integrated into the barb insert I was proposing, but that's well beyond my fabrication capabilities.

I think any bit helps, but I agree the resulting difference may be negligeable, and may very well not lower CPU temps by a measurable amount.

I don't see where you guess that the nose cone must be soldered to the top, nor how it would have to be soldered to the fins. One certainly wouldn't want any stagnant water though.

I would simply solder copper pieces to the middle plate, and file/sand them to shape. As for the inside of the barb, I'd simply shape a block of plexy/resin/epoxy and attach it inside the barb (just about any adhesive would do).

All of this can be done in a way that is completely reversible, but as Cathar pointed out, it would be a user customization, because he's not going to be doing all that, since it definitely doesn't justify the additional expense/cost.

Since87 12-07-2002 11:57 AM

Quote:

Originally posted by bigben2k
I don't see where you guess that the nose cone must be soldered to the top, nor how it would have to be soldered to the fins. One certainly wouldn't want any stagnant water though.

If the water hitting the top of the fins is providing any cooling, then that cooling will be lost unless a nose cone has a good thermal contact with it. Short of soldering it, it would be difficult to get a decent thermal contact.

Cathar 12-07-2002 06:01 PM

Quote:

Originally posted by Since87
If the water hitting the top of the fins is providing any cooling,
It does.

In fact, there is a very small benefit seen (~0.5C - but I could be wrong - hard to quantify something that small on the equipment I have) by sticking some arctic silver on the bottom of the nozzle plate where it comes into contact with the tops of the fins as the middle plate now comes into play as part of the cooling effect.

That's something you can try...

Since87 12-07-2002 08:33 PM

Quote:

Originally posted by Cathar
In fact, there is a very small benefit seen (~0.5C - but I could be wrong - hard to quantify something that small on the equipment I have) by sticking some arctic silver on the bottom of the nozzle plate where it comes into contact with the tops of the fins as the middle plate now comes into play as part of the cooling effect.

That's something you can try...

Interesting idea.

I just noticed that the top of the baseplate and the bottom of the nozzle plate appear to be lapped too. Impressive.

Have you looked at how well the AS stays in place with waterflow through the block?

I've heard that AS forms a foil when used at the pressures used for clamping pelts. I wonder if the silver gets pressure welded to the copper. Hmm... stir some diamond grinding compound into the AS. Thinking about the six ton press at work... :D

Glad you went ahead with another lot Cathar. Just in case I get too carried away.

Cathar 12-07-2002 08:47 PM

Quote:

Originally posted by Since87
Interesting idea.

I just noticed that the top of the baseplate and the bottom of the nozzle plate appear to be lapped too. Impressive.

Have you looked at how well the AS stays in place with waterflow through the block?

Yes - all the plates are lapped on both sides, although the inner surfaces are only done to 120-grit. The visible top plate is done to 300-grit, and the base to 600-grit. For the internal surfaces I wanted to ensure even pressure for the O-ring, plus it assists in internal heat-transfer, which I can tell because when I lap the base, which is done after the insides have been lapped and the block assembled, the top of the block heats up quite noticable as I'm lapping the base.

AS will assist in the fin-middle plate transfer slightly. One may ask why not just make the fins higher, and the answer to that is that this drops the water-velocity and results in lower performance, so it's all a bit of a delicate balancing act.

In the one block I did the AS on the AS did not appear to come away from the surface at all after a few weeks of 10lpm (2.5gpm) flow, so in that respect it seems to be fairly resistant, and any AS trapped between the middle plate and fin tops should be pretty much right for the long term.

bigben2k 12-09-2002 05:24 PM

Quote:

Originally posted by Cathar
It does.

In fact, there is a very small benefit seen (~0.5C - but I could be wrong - hard to quantify something that small on the equipment I have) by sticking some arctic silver on the bottom of the nozzle plate where it comes into contact with the tops of the fins as the middle plate now comes into play as part of the cooling effect.

That's something you can try...

Thanks. Note to self: try a copper top too!


All times are GMT -5. The time now is 02:18 AM.

Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com
If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk...
Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...