"Jet" block ver. 1.

[bigben2k]

Originally posted by Tuff
Sorry JayDee from the way you present yourself on this forum...I would have thought you would be beyond mistakes and guessing when making a block.

Well, his spelling could use some work

[jaydee]

Originally posted by bigben2k
From a designer's perspective, yeah, the block is going to bring any pump to its knees: you've only got 9 jets. IMO, you would do well to calculate the pressure drop, and try to tune those jets for a pump's optimal range.

You got the blocks mixed up. The Jet Block #2 has no flow and is way to restrictive and has 10 jets. Jet block #1 has 9 large jets and has pretty good flow. I need to seal it back up and throw it back on the new test system. Will do tonight.

Cathar's design compensates for the pressure drop by having a whole bunch of tubes, so it works well with any pump . As the flow rate increases, the pressure drop increases exponentially, so big pump owners need to have a nice pump, with a lot more head than flow, to run nicely.

I understand that. Unfortunatly I am trying to make use of the equipment I have at work and tiny jets are not an option. I was trying to make up for it by making large jets and bigger holes which I am not sure of it worked or not as I have yet to test it.

I really wouldn't give up on the jet block just yet. LeeJSmith's work is top notch, and uses a brass tube, for the jets, which I think is great because at higher flow speeds, a smooth inner surface in the tube is going to be less restrictive, and should result in a very focused jet.

Yeah, his work is looking damn good. Will see how the 9 jet block works out. If it isn't to good I am going to drop it for now and get back to my Lemon Block as it is doing real well. It is one of the few blocks I made that had any real thought and real time put into it. All the Rev. blocks and these 2 jet blocks were thought up and developed in less than 30minutes at work as that is about all the time I can squeeze in every now and then.

"Cascade" isn't hard to make, but it's hard to make it right. I think it would be a perfect exercise for you, JD.

I could make it. I am just trying to do it a little differently. I always new these jet blocks wouldn't be Cascade performance though. Not what I was shooting for.

[jaydee]

Originally posted by Tuff
[b]JayDee's
Quote:
Did you not read in Cathars posts where he got the ideas for the Cascade?

If I remember correctly it had something to do with him washing dishes

JayDee...I might be missing something here..Washing dishes or JayDee's idea...hmmm...Unless JayDee means washing dishes in some other language?????

Yep...I guess the credit is yours.

I am not taking FULL credit. Just pointing out I started working on this concept quite some time ago before the Cascade was even known. Read the first few lines in the first post of this thread and go to the REV 1 link and check the date on the post. Just a point to refute you ignornace in saying I am just starting to do this, which it total BS.

originaly posted by Cathar
Basically drawing upon the White Water concepts, Volenti's directed jet pipes, BigBen2K's cups (rather than dimples), and jaydee116's mass-jet array block, and bringing them all together into a cohesive and (somewhat) affordably machinable design.

Anything you do now with Cups is a clone block...not that its a bad thing...but why try to improve on a design that Cathar has already nailed? OR why repeat the mistakes that "I" already made with that block design?

No, dumbass, I am not trying to improve on his design. I am trying to make a version I can easily make with the tools at work and still get "decent" performance.

Sorry JayDee from the way you present yourself on this forum...I would have thought you would be beyond mistakes and guessing when making a block.

Tuff

Whatever dude. I always made it clear I am a trial and error person and not a scientific one. I know the faults of my blocks before they even get on the test bed. But it isn't going to stop me to see how they work and posting the results. And the results are taken a bit more carefully than from an onboard probe that reads 5+C swings every 9 seconds. And where in this thread did you see my block rules all like you made yours out to be? Do you have a real reason to come in my thread and trash my results or are you just pissed you got called on your ignorance in your thread?

[Tuff]

Quote:

No, dumbass, I am not trying to improve on his design. I am trying to make a version I can easily make with the tools at work and still get "decent" performance.


Hmm....Funny how you are trying to improve on his design....yet you are still fumbling around on MY first attemp.

Hats off to you JayDee...nothing like redoing all the ground work.

As for my ignorance of the temps n such and wild swings. Ask BB...and he will tell you that everyone gets those swings. It all depends on the sensitivity of the thermosensor and how often the readings are taken.

Again... you speak of average temps.


I could have posted the average reading done every 1 min or so...but WHY???? That does not tell me what temp the cpu is actually running at all times.

Just for you JayDee...my temps were 36-37 deg These were the average temps reported.

READ THIS.

Heat has to transfer into the block before it can be removed..therfore the heat registers on the Thermosensor BEFORE it is taken away. IE spikes in temp. THIS explains the swings I had reported.

I have come across many people in my life...who ALWAYS think they are RIGHT.

Tuff

[jaydee]

Originally posted by Tuff
Quote:

No, dumbass, I am not trying to improve on his design. I am trying to make a version I can easily make with the tools at work and still get "decent" performance.


Hmm....Funny how you are trying to improve on his design....yet you are still fumbling around on MY first attemp.

Hats off to you JayDee...nothing like redoing all the ground work.

Ok, so you have a laser and are trying to work with it to make one of these blocks with? And quite frankly i give a rats ass what you have done because of the following:

As for my ignorance of the temps n such and wild swings. Ask BB...and he will tell you that everyone gets those swings. It all depends on the sensitivity of the thermosensor and how often the readings are taken.

Again... you speak of average temps.


I could have posted the average reading done every 1 min or so...but WHY???? That does not tell me what temp the cpu is actually running at all times.

Just for you JayDee...my temps were 36-37 deg These were the average temps reported.

READ THIS.

Heat has to transfer into the block before it can be removed..therfore the heat registers on the Thermosensor BEFORE it is taken away. IE spikes in temp. THIS explains the swings I had reported.

I have come across many people in my life...who ALWAYS think they are RIGHT.

Tuff

Sorry Tuff but this is absolute and utter BULL SHIT! Temps DO NOT swing over 5C in less than 9 seconds PERIOD! Especially under a constant load. Keep on talking, this is a good laugh! If BB really thinks that he needs to buy something other than that P166 system he is on and do some REAL testing instead of theory!

[leejsmith]

Originally posted by bigben2k
"Cascade" isn't hard to make, but it's hard to make it right. I think it would be a perfect exercise for you, JD.

Thats a big understatement IMO it's very very hard to get it right.

My best was only 2C better than the maze 3 i use to compare againts and that was after many changes to cup size , tube size and tube length. I had cathar and ben pointing me in the right direction with the ratio between them all and even when i thought i had it right it was still no better.

[bigben2k]

Thanks Lee.


Guys, we've been over this before.

The wild temp swings actually come from Sandra. CPUBurn works better at maintaining a steady temp, for load testing. Incidentally, that's what JD uses...

You two (Tuff, JayDee116) really need to try to get along!

Tuff, you just need to come forward, and say how you tested your block, and what your reference point was. As JayDee stated, you can't use someone else's temps as a reference point: you have to get a known block, and compare yours to it, under the same conditions, and you have to do it a few times, to make sure that you account for the TIM joint variations.

That's exactly what JD is doing, and he's got it right. I'm sure he hasn't tested it more than a couple of times, but he's still prototyping: when he's got a final product, I'm sure he'll put it through the ringer.

[MMZ_TimeLord]

Tuff,

Tuff, you just need to come forward, and say how you tested your block, and what your reference point was. As JayDee stated, you can't use someone else's temps as a reference point: you have to get a known block, and compare yours to it, under the same conditions, and you have to do it a few times, to make sure that you account for the TIM joint variations.

"under the same conditions" refers to the SAME EXACT HARDWARE. Just change the blocks and NOTHING ELSE.

Heat has to transfer into the block before it can be removed..therfore the heat registers on the Thermosensor BEFORE it is taken away. IE spikes in temp. THIS explains the swings I had reported.

That is incorrect, a 5 degree spike in 9 seconds is not likely even with Sandra... you are talking a 5 degree jump AND drop in 9 seconds. 1 or 2 degrees I could understand. What are you doing moving a cutting torch across it?

IF it is indeed doing that, I suggest you get something else to load your CPU like BB was mentioning (CPUBurn). This will level out your temps.

BTW... I hate using average temps that have swings of more than 2 degrees in the measurements. It just stinks of marketing.

[jaydee]

Originally posted by bigben2k
Thanks Lee.


Guys, we've been over this before.

The wild temp swings actually come from Sandra. CPUBurn works better at maintaining a steady temp, for load testing. Incidentally, that's what JD uses...

This is not so. I HAVE Sandra on my test bed and use it aswell. After equilibrium has set in under load I NEVER see a swing of over 1C (for hours and hours!) and thats on the Thermocoupler aswell as the onboard! You simply do not get 5C swings. Thats like going from idle to load. Only time I EVER see a 5C+ swing is from hitting the start button and loading into windows and that still takes over 20 seconds! And the TC picks it up WWWWWAAAAAAAAYYYYYYYY before the onboard does. It takes a couple minutes for the onboard to read a 5C swing. Unless it is busted of course!

You two (Tuff, JayDee116) really need to try to get along!

Why? :shrug:

Tuff, you just need to come forward, and say how you tested your block, and what your reference point was. As JayDee stated, you can't use someone else's temps as a reference point: you have to get a known block, and compare yours to it, under the same conditions, and you have to do it a few times, to make sure that you account for the TIM joint variations.

That's exactly what JD is doing, and he's got it right. I'm sure he hasn't tested it more than a couple of times, but he's still prototyping: when he's got a final product, I'm sure he'll put it through the ringer.

How many times does this need to be feed to him? I thought we went into detail over and over across multiple threads on this issue and would think it would have gotten through by now, but everytime I think that here he comes again with the same BS as before! And this time trying to flame ME! BAHAHAHAHAHAHA! Like I said, quick to forget and slow to learn.. A bad combo! This has been a hoot though!

[Tuff]

My results were with MBM using the Diode...My results have been compared to the MAZE 3 that I have, which my block beat by quite a margin.


Sandra was not used in my testing. I used CPU burn.

Blocks were seated and unseated...same day same hardware same Temp in room.

Tuff

[jaydee]

Originally posted by Tuff
My results were with MBM using the Diode...My results have been compared to the MAZE 3 that I have, which my block beat by quite a margin.


Sandra was not used in my testing.

Blocks were seated and unseated...same day same hardware same Temp in room.

Tuff

Oh yeah, you where using IDLE temps for the performance guage. Forgot................... Not that it matters because your testing methods are severly flawed. If you want to be taken seriously listen to the people telling you how to do it better or go to another site where they will not call you on your flawed methods.

[bigben2k]

Originally posted by jaydee116
Why? :shrug:

LOL!

My results were with MBM using the Diode...My results have been compared to the MAZE 3 that I have, which my block beat by quite a margin.


Sandra was not used in my testing. I used CPU burn.

Blocks were seated and unseated...same day same hardware same Temp in room.

Hum... I know that there might be a setting in MBM that reads the on die/CPU diode incorrectly, but I just haven't played with it long enough. Obviously something is wrong, and we really need to figure out what it is, in order to validate your claim. Not that I doubt you, the Maze3 isn't hard to beat, but it would be nice to have a more accurate idea of what's going on.


Back to the topic...

JD: did you consider making the dimples deeper?

I think that you might be taking a performance hit, because of the mismatched diameters: the jet needs to be smaller than the dimple. Otherwise, the water in the dimple stays (essentially) trapped, and you end up with a performance that's on par with a jet against a flat baseplate, a la Swiftech.

Also, I don't think that you accounted for the return flow properly: your jets probably interfere with each other, because there's no tube to make space for the return coolant.

Your jet array of 9 or 10 should work. You might look into the brass plugs that we proposed for "Cascade": that might work out to be very easy to do. Heck, even the casing of a rivet might work!

[jaydee]

Originally posted by bigben2k



Back to the topic...

JD: did you consider making the dimples deeper?

I think that you might be taking a performance hit, because of the mismatched diameters: the jet needs to be smaller than the dimple. Otherwise, the water in the dimple stays (essentially) trapped, and you end up with a performance that's on par with a jet against a flat baseplate, a la Swiftech.

Also, I don't think that you accounted for the return flow properly: your jets probably interfere with each other, because there's no tube to make space for the return coolant.

Your jet array of 9 or 10 should work. You might look into the brass plugs that we proposed for "Cascade": that might work out to be very easy to do. Heck, even the casing of a rivet might work!

Yes, on deeper. I mentioned that already. The OD of the jets is smaller then the ID of the demples by almost half. They just are not deep enough and the pipes are to small to only have 10. I need about 30 to keep flow up, but I havn't figured out how to pull that off with what I have.

Jet block 1 however wit hthe 9 pipes seems to have enough flow. I DO need to make the dimples deeper though. I am not even half way through the 1/8" base. I am considering makeing a 1/4" base with 1/8" deep holes and adding a 3rd row of pipes to make 12 pipes. this goes agains my goals, but my goals can't be achieved so I will have to adapt. Anyway I will try and run the 12 pipe version out on the laser today as I just tuned it up and itis cutting and engraving MUCH better than it has since I been here.

[jaydee]

I just made a new one. It has 19 jets. The OD of the jets is .09" and the ID is .06". Should be plenty of flow through this if i can make the base work out right. I don't like the pattern to much but I think it will work. Will post pics when I get home tonight. Have to do some actual work now....

[Tuff]

Here is what I am thinking...if JayDee makes the holes deeper he will be giving up too much copper.

As for right now the copper is most likely transfering to the sides as well as through his holes, then being removed by the water.

For example...you will be removing heat at both 1/8th in and 1/16th in. You will want to maximize the amount of heat removed at the 1/16th in or deeper portion of your block.

If you can..make the holes smaller and deeper...this should maximize the amount of water/copper ratio and in return give you lower temps.

Tuff

[jaydee]

Jet Block Ver. 3.


--

--

--

--

--


Pip OD's .090", ID's .060", lasered area is 1/8" deep. As far as I got so far....

[MMZ_TimeLord]

All I can say is ... SUUUHHHH-WHEEEET!!!!

[jaydee]

I hooked up Jet Block version 1 (9 pipes) an hour ago and so far it is doing very well.


XP1700+@1800mhz
1.75Vcore

Load temps:
Air: 26C
Water: 30C
TC: 41C
Onboard: 39C


Maze 4 results were:
XP1700+@1800mhz
1.75Vcore

Load Temps:
Air: 26C
Water: 30C
TC: 43C
Onboard: 40C

ALOT of tweaking could be done the Version one, but this exites me a little more about this new 19 pipe version. Will try to get it finished up in the next couple days....

[Tuff]

Originally posted by jaydee116
I hooked up Jet Block version 1 (9 pipes) an hour ago and so far it is doing very well.


XP1700+@1800mhz
1.75Vcore

Load temps:
Air: 26C
Water: 30C
TC: 41C
Onboard: 39C


Maze 4 results were:
XP1700+@1800mhz
1.75Vcore

Load Temps:
Air: 26C
Water: 30C
TC: 43C
Onboard: 40C

ALOT of tweaking could be done the Version one, but this exites me a little more about this new 19 pipe version. Will try to get it finished up in the next couple days....

Any idea why the variance of TC being 2 deg and Onboard only being 1 deg?

Tuff

[Tuff]

And just out of curiosity...Make one without Jets to see if there is an advantage of the Jets over just an inlet.

Tuff

[jaydee]

Originally posted by Tuff
Any idea why the variance of TC being 2 deg and Onboard only being 1 deg?

Tuff

Onboard sucks. It doesn't rise and fall linarily. Thats why we don't like to use the onboard.

Do you mean just the cups in the base and no pipes or holes in the middle peice? If so no problem. Will do tomorrow. Sounds like a good experiment.

[Tuff]

Originally posted by jaydee116
Onboard sucks. It doesn't rise and fall linarily. Thats why we don't like to use the onboard.

Do you mean just the cups in the base and no pipes or holes in the middle peice? If so no problem. Will do tomorrow. Sounds like a good experiment.

Yep thats what I mean

Tuff

[bigben2k]

That's more like it


Have you thought about using a tube that extends inside the block? I think you ought to give it some thought, because it's going to allow the flow to come out nicer, and not interfere with the incoming flow so much.

Like I said, even the case of a rivet would work!

[Tuff]

Originally posted by bigben2k
That's more like it


Have you thought about using a tube that extends inside the block? I think you ought to give it some thought, because it's going to allow the flow to come out nicer, and not interfere with the incoming flow so much.

Like I said, even the case of a rivet would work!

This seems like a good idea.....so now you have 3 variations to try. Though I feel that putting the tube inside will act like a jet in itself.

Tuff

[bigben2k]

The tube will unfortunately add some flow restriction though, but performance should increase dramatically, back to where it should be.

The pressure drop comes from two points: the transition from the 1/2" tubing to the individual tubes, and the tubes themselves.

With some tuning, you can get the right length of the tubes. LeeJSmith's experiments, with Cathar's posted graph pretty much show how to do it.

The design really is the best that we can do, but it's infinitely difficult to tune it right. I can't even imagine how many hours Cathar spent trying out different variations, and testing each one multiple times.