Current review project (MCWCHILL)

[BillA]

yes, the chiller heat balance sux

pHaestus
can we get the hot side inlet and outlet temps and flow rate ?
you could use your regular flow meter and record the various offsets for your thermisters

??

EDIT: the only variables we have are the flow and deltaT,
there is something we are not 'seeing'

[pHaestus]

hmm that would depend on if I can scrounge up the stuff needed to plumb in the digitecs. Lemme dig around.

[pHaestus]

Will have to be another day for that. I am home with son today (wife has car) and I need parts from plumbing section. I can't understand why the numbers look reasonable as f(VCore) and f(Frequency) though while the flow rate numbers blow up. Still thinking on this one...

[Les]

Originally posted by unregistered
pHaestus
can we get the hot side inlet and outlet temps and flow rate ?
you could use your regular flow meter and record the various offsets for your thermisters

??

EDIT: the only variables we have are the flow and deltaT,
there is something we are not 'seeing'

To obtain full benefit from these hot-side measurments also have to measure the TECs' V and A to seperate their Wattage variation with pumping load and temperature.

[BillA]

I saw little change in V and A for each TEC over the temp range I tested, but it was linear for both TECs

from 16.517A to 16.667A at -1.9 to 5.0°C (chilled coolant outlet) with a load of 39.4 to 94.4 Watts

[Les]

Originally posted by unregistered
I saw little change in V and A for each TEC over the temp range I tested, but it was linear for both TECs

from 16.517A to 16.667A at -1.9 to 5.0°C (chilled coolant outlet) with a load of 39.4 to 94.4 Watts

Interesting.It is the only data I have seen for 220w TECs. The ~1% variation is less than I expected.
Best I can do:-

Since87's spreadsheet does not include change in V and A with load.

[Since87]

Originally posted by Les
Since87's spreadsheet does not include change in V and A with load.

Yes, I skimped there. The information needed to calculate that correctly is not routinely published with TEC specs, and I never got around to adding an approximation.

[pHaestus]

Looks like previous flow numbers weren't at equilibrium. I have been running 1.5-2 hours for equilibration and numbers now look normal. 2 more data points to collect; will put numbers up before bed.

[pHaestus]

ok this is still not crystal clear. Delta T (hot-cold) still increases as flow rates increase unlike how it behaves as a f(MHz) or f(Vcore). I think this may be due to the increased pump heat getting dumped at lower flow rates due to throttling. Will put graphs up in another few hours when all points are collected. No anomolies this time though.

[pHaestus]

No spike in coolant temperatures this time...



Heat extracted was graphed using Les's equation from his previous graphs.

If delta T (hot-chill) is expanded then it is still slightly increasing as flow rate increases rather than decreasing as before.


Test numbers here:

http://www.procooling.com/users/phae...lownumbers.jpg

[Les]

[Les]

Some ramblings
The above graphs are simulations for the Whole Cold-side of the system(Chiller Cold-side + cpu & mb + pump+Tubing/instrumentation)
Ri is interpreted as the Resistance between the system and Ambient* Air
Pissing about with the paramters can give results showing similar trends to the observrd, e.g


To explain the increase in Watts extracted with Flow rate need a rate dependant source and/or a rate dependant resistances(Rc and Ri).
Flow dependant source: - Pump Heat production increases with Flow rate** (Eheim1048,Limited to mfrs specs of 10w?) (and Rc decreases)
Flow dependant Resistance:- (1) cpu & mb, Resistance Rc decreases with increasing Flow-rate.
(2) cpu & mb, Resistance Ri decreases with increaing Flow-rate? (connected in series with Primary cpu and mb sources)
(2) Tubing/instrumentation ,Resistance Ri decreases with increasing Flow-rate

*ambient in the Primary Kyotherm calculations is interpreted as the Hot-side Water
** Since87's numbers here http://forums.procooling.com/vbb/sho...5&pagenumber=7

[Les]

Originally posted by unregistered
....
pHaestus
can we get the hot side inlet and outlet temps ,
........

Perhaps better if could, first, get the temp changes across the MCW5000-a Waterblock to help sort out the Heat Imputs. Across the MCW5000-a and the Eheim1048 would be better still.

[BillA]

ck, both hot side wbs and pump

[Les]

Originally posted by unregistered
ck, both hot side wbs and pump

Assume you mean outlets of Eheim1049 and MCW5000-a in the cold loop.?
If so, agree.

[BillA]

yes but
actually after both to attempt a total energy balance
but doubt it would work as the errors for the missing bits and pieces will be large relative to the total
- especially on the cold side

[pHaestus]

I have not the needed number of probes to take ALL the suggested measurements. At the moment the setup is:

pump -->Temp1-->chiller-->Temp2-->WB-->flowmeter-->valve-->res-->pump

Assuming losses to imperfect insulation being minor, the most efficient placement would be to insert temp3 after waterblock but before flowmeter. That would allow one to get deltaT across waterblock and across the pump.

Alternatively I can rig up a bunch of DOW sensors inline and measure everything. Not gonna be the specs that you want to hear though (0.5C accuracy and 0.125C res). I guess I could rig up some other solution with thermocouples but that is a road I prefer not to embark upon.

With my available thermistors I could plumb either a third probe into the cold loop as noted above or I can plumb a second probe into the hotside loop to measure delta T across the hotside block. I must remind you that at the moment I have 2 hotside loops that are separate and one has a pelt running at 12V and the other at 13.5. This undoubtedly complicates serious investigation of the hotside loop; if required I could merge that into a single loop with parallel chiller blocks and parallel radiators to simplify temperature collection.

[Les]

Originally posted by pHaestus
... the most efficient placement would be to insert temp3 after waterblock but before flowmeter. ......

Best option in my opinion.

[Les]

Originally posted by pHaestus
....
Heat extracted was graphed using Les's equation from his previous graphs.
..........

Needs an Anti -Freeze and a temperature correction.
Assuming using a glycol based ~30/70 Anti-Freeze/Water then suggest should x ~ 0.9.
Anti-freeze correction using Scott Gamble's data**
Temp correction,Heat ^ ~ 1%..*

* Maya Data http://www.mayahtt.com/tmwiz/default.htm
** From article http://www.overclockers.com/articles609/index03.asp

[Les]

A pretty picture:-

The "Billa Die Insulation" is for the Heat Die alone and possibly only applies to coolant at 25c (Estimated from 25c data).

[pHaestus]

All:

Second baby arrived on Monday so I haven't even been down to the computer testing facility (ie basement) since. I'll try and get some setup done this weekend perhaps.

Seems to me my strategy should be to:

1) plumb in third temp probe for chilled loop
2) insulate the CPU wb as well as I can manage

I'll run a few tests then and hopefully be ready for actually reviewing this thing.

[pHaestus]

Well crap. I went downstairs today and drained the chiller loop, plumbed in the temperature probe, and refilled it. Now for whatever reason I am getting nothing out of the swissflow meter. Nada pulse. I pulled the flowmeter out to inspect it and rinsed it with a ton of hot water. Replaced it and still nothing. Only other thing I can check is the wiring, but it looks to be soldered properly together and heatshrinked so I dont think it'll make much difference.

If I can't get the flowmeter to work then I am probably done with this testing. I'll hopefully get a chance to mess with it more tonight.

[pHaestus]

I will actually finish this review up in the next few days (on my return flight from holidays most likely) and so it should be posted before too long. I shipped it back to Swiftech a week or so ago; Bill did you receive it yet?

BTW this post is somewhat for keeping readers informed and somewhat to see if Joe's vbb3 rewriting will cause this post to appear on the front page in my worklog

[BillA]

"Bill did you receive it yet?"
nope
re-send, nothing here