Prototype Free Flow CPU Pump Block

[flar]

DDC at 25% is more quiet imo and has a higher flow rate than DC-LT and Asetek variants.

Where it really matters is multiple radiators.

That is the common assessment out there and I can believe it. I'm pointing out that you are comparing solutions that are already dialed in for silence out of the box to the full bore performance of solutions that would have to be manually dialed back for a similar application. Even with 2 radiators in their Ghost build, JayzTwoCents dialed back their Apogee to keep it quiet.

Please don't take this as me having any doubt that your solution is good - I actually think it would meet my needs in the current prototype form if I extrapolate from the data you do present. This is just my Scientific Method hat pointing out what kinds of comparisons I think would focus the objective data to better drive the points home.

 

[Luke NK]

Maybe a noise normalized test would be better.

 

[Wahaha360]

That is the common assessment out there and I can believe it. I'm pointing out that you are comparing solutions that are already dialed in for silence out of the box to the full bore performance of solutions that would have to be manually dialed back for a similar application. Even with 2 radiators in their Ghost build, JayzTwoCents dialed back their Apogee to keep it quiet.

Please don't take this as me having any doubt that your solution is good - I actually think it would meet my needs in the current prototype form if I extrapolate from the data you do present. This is just my Scientific Method hat pointing out what kinds of comparisons I think would focus the objective data to better drive the points home.

Maybe a noise normalized test would be better.

I understand your view, to that I say, I will add to the chart, it's a WIP.

 

[Wahaha360]

...

 

[prava]

I understand your view, to that I say, I will add to the chart, it's a WIP.

But... The product you tested was an all-in-one pumb+block combo? And you are comparing to pump+top combo?

Isnt it apples to oranges?

 

[lehman]

For flow tests, and possibly noise tests. It would make sense to add a cpu block of comparable flow resistance to see how your block compares since yours includes the cpu block.

Is your cold plate a custom design or is it from one from a current product? I designed mine around the Optimus Cold Plate.

Lehman

 

[gwertheim]

For flow tests, and possibly noise tests. It would make sense to add a cpu block of comparable flow resistance to see how your block compares since yours includes the cpu block.

Is your cold plate a custom design or is it from one from a current product? I designed mine around the Optimus Cold Plate.

Lehman

What's your technical background?

Personally I have worked as a pipe fitter on larger construction projects such as hospitals and office buildings so I do understand the concepts better than most. Hydro dynamics, thermo dynamics are concepts I understand and can apply to a computer loop. Hell if I did a hardline build my cuts would be quite accurate because I have done it on a large scale even if I haven't built a loop in a computer before.

Based on my experience of flow analysis on a large scale and stuff like that requires expensive equipment if you want to do it right. As @Wahaha360 could explain, machining time is very expensive and might not be worth the 1 or 2 degrees at the end of the day.

You need to keep in mind that this is a one man operation with limited resources. Yes he wants to give you the best product possible but there is a point where the cost out weighs the benefits.

If you are a mechanical engineer, have a background in this area or went to MIT or Cal Tech then I will shut my mouth.

I am not trying to be an ass, I am just trying to be realistic and remind people that there are limits to the resources these creators have. In many cases these are passion projects developed by people with a full time job and other financial commitments.


Also if I'm not mistaken even though the law isn't my area of expertise. Optimus can take you to court for stealing their design if you try to profit off of it.

 

[srekal34]

LOL the post above made me laugh ? ?
you do not need to have a PhD in rocket science to understand the fact that pump running just with top and pump running with the block attached will have different restrictions, therefore results are just incomparable.

Moreover, lehman is not stealing anything from Optimus, he is using an end user product from Optimus as part of his new product. Is Bosch suing Koeniggsegg for using their fuel pumps in their cars?

 

[Allhopeforhumanity]

If you are a mechanical engineer, have a background in this area or went to MIT or Cal Tech then I will shut my mouth.

LOL the post above made me laugh ? ?
you do not need to have a PhD in rocket science to understand the fact that pump running just with top and pump running with the block attached will have different restrictions, therefore results are just incomparable.

Moreover, lehman is not stealing anything from Optimus, he is using an end user product from Optimus as part of his new product. Is Bosch suing Koeniggsegg for using their fuel pumps in their cars?

Not who you were referring to, but I have a PhD in mechanical engineering focused on numeric simulation (not MIT or CalTech, but UCLA isn't the worst R1 institution), and have some application experience in thermal-fluids; primarily heavy equipment/high-power electronics cooling in this specific field. What @lehman and @srekal34 are getting at actually makes a lot of sense.

There is quite a bit of subtlety in optimization when it comes convective cooling that extends to CPU block design. Everything from the jet plate fin height/pitch, to the number of bends, to the types/diameters of fittings will all impact the optimal pressure/flow inputs to find the peak in noise normalized performance. The real complexity is that the majority of these factors don't scale linearly, and many have parasitic/convoluted effects on the other terms. Thus, testing in the most 1-to-1 scenario with as many extraneous parameters controlled or normalized as possible is really the only way to provide sound results.

However, to @gwertheim 's point, cost and time-investment are also considerations that shouldn't be discounted. It might very well not be worth it for @Wahaha360 to completely optimize the block in the first pass, as long as it's competitive advantage is maintained in its small size and proves to be "good-enough" performance wise to warrant considering over AIOs, then he can always make further improvements in Rev2 once the initial investment is recuperated.

 

[gwertheim]

Not who you were referring to, but I have a PhD in mechanical engineering focused on numeric simulation (not MIT or CalTech, but UCLA isn't the worst R1 institution), and have some application experience in thermal-fluids; primarily heavy equipment/high-power electronics cooling in this specific field. What @lehman and @srekal34 are getting at actually makes a lot of sense.

There is quite a bit of subtlety in optimization when it comes convective cooling that extends to CPU block design. Everything from the jet plate fin height/pitch, to the number of bends, to the types/diameters of fittings will all impact the optimal pressure/flow inputs to find the peak in noise normalized performance. The real complexity is that the majority of these factors don't scale linearly, and many have parasitic/convoluted effects on the other terms. Thus, testing in the most 1-to-1 scenario with as many extraneous parameters controlled or normalized as possible is really the only way to provide sound results.

However, to @gwertheim 's point, cost and time-investment are also considerations that shouldn't be discounted. It might very well not be worth it for @Wahaha360 to completely optimize the block in the first pass, as long as it's competitive advantage is maintained in its small size and proves to be "good-enough" performance wise to warrant considering over AIOs, then he can always make further improvements in Rev2 once the initial investment is recuperated.

I was trying to refer to a level of education which is related to the field of engineering and/or experience in related fields

 

[Allhopeforhumanity]

I was trying to refer to a level of education which is related to the field of engineering and/or experience in related fields

I understood what you were getting at. I just saw a "this is my time to shine" opportunity and wanted to chime in on the subject .

 

[lehman]

Just an average guy who read a book one time.

What I was trying to get at in the previous post is quite simple. His flow numbers through his block are not directly comparable to flow numbers just through the pump.

It would be expected that the flow would decrease going through the cold plate and channels in the block. It would be easier to compare numbers between his combo pump/block and a pump and separate block.

Yes, this would add some work to get this information. I was just giving my opinion.



As for my use of the cold plate by Optimus, they are aware and have seen the technical drawings.

Lehman

 

[gwertheim]

Just an average guy who read a book one time.

What I was trying to get at in the previous post is quite simple. His flow numbers through his block are not directly comparable to flow numbers just through the pump.

It would be expected that the flow would decrease going through the cold plate and channels in the block. It would be easier to compare numbers between his combo pump/block and a pump and separate block.

Yes, this would add some work to get this information. I was just giving my opinion.



As for my use of the cold plate by Optimus, they are aware and have seen the technical drawings.

Lehman

Just wanted to make sure you aren't getting sued or anything like that

 

[flar]

I absolutely agree with all of the statements that @Wahaha360 is providing a great service to the community here and that we are all basically just beggars at their design table. And the design looks great.

And I love the perfectionist approach to wanting to exceed the reference designs.

What I was getting at with my feedback about the testing is that it is an unfair comparison that is actually biased against the FreeFlow. The "first glance" measure indicates that it is only 90% the flow rate of the top reference designs tested, but that ignores the fact that the other tests were missing an important flow inhibitor. The prototype may actually already be better than the reference designs.

And, for my own preferences, even if the block was only 90% as good as the reference setups, the block already beats most existing "AIO" boxed solutions and it looks better to boot. I'd take it right now even at 90%. The only comparison data I don't see is to an Apogee pump/block unit, but I don't care too much because the FreeFlow looks much better. Knowing the relative performance to the Apogee solution would be more for bragging rights than a purchasing decision point.

(The Apogee specs seem to just list the pump specs and ignore the impact of its block so I haven't found any comparable published flow rates for the pump/block combined output...)

 

[lehman]

Apogee with a ddc 3.1. This is the original version, not the current one. I have not found actual testing on that one yet.

This is a 12 watt pump so flow could be increased with an 18 watt pump. Though noise would increase.

http://martinsliquidlab.petrastech.com/img/SwiftechApogeeDrive-PumpCur.gif

 

[LXCXH]

Will it be sold? I'm thinking about to replae my A4 case with a customer water cooling. Please let me know if you can sell it.

 

[panzermuffin]

Would the pump need an AiO header on the motherboard? Sorry, never had an AiO/custom loop before ?

I'm considering the Gigabyte Aorus X570 I Aorus Pro Wifi motherboard which doesnt have an AiO header.

 

[PBJ]

Would the pump need an AiO header on the motherboard? Sorry, never had an AiO/custom loop before ?

I'm considering the Gigabyte Aorus X570 I Aorus Pro Wifi motherboard which doesnt have an AiO header.

You can use a regular fan header.

EDIT: Pump headers usually run at max RPM by default and you can configure a regular fan header to max RPM if necessary in the BIOS.

 

[flar]

Would the pump need an AiO header on the motherboard? Sorry, never had an AiO/custom loop before ?

I'm considering the Gigabyte Aorus X570 I Aorus Pro Wifi motherboard which doesnt have an AiO header.

Note that the Fr33Flow is just a block - the pump is added separately so you can choose whatever pump you want. Many pumps come with a 4-pin molex for power anyway and just use a 3-pin fan-style header for monitoring. So, it depends on which pump you choose and that pump may be too powerful for an AIO header anyway. AIO headers, like fan headers, are often capped at 12 watts and DDC pumps can draw as much as 18 watts.

But, there are pumps out there that are either powered by a fan-header style connector or they are low enough in power draw that they can be rewired to do so. In that case, as @PBJ mentioned, a dedicated AIO header isn't really all that special beyond the default speed profile assigned by the BIOS.

 

[panzermuffin]

Wonderful, thank you guys. So both fans of the radiator via Y-Splitter into the FAN header and the DDC pump MOLEX directly to the PSU and the PWM cable into the CPU header?

I'm still deciding which CPU-cooler to get initially. Can't decide between LT240 with wonky/dangerous mounting system, or with SecuFirm2 or L12S and wait for this block.

What are your plans?