This is a bit embarrassing... Reason being, I've already asked this question to Sven, and he has already answered it, but I'm still confused

And I've pestered him so much in the last three weeks that I dare not bother him any more.

The conundrum is the flow sensors. The one I like , basically because it has a negligible pressure drop (0.020 bar @ 5 lpm, as opposite to the 0.15 bar of the Digimesa "usual suspects") while maintaining decent accuracy (it's made by Digimesa as well after all).

However, this one has a very low ppl count (130). In my current solution I had to increase the reading window (or is it called resolution? lost in translation...) from the default 2s to 8s because the deviation is calculated like this (to get a value in l/h):

(1 / ppl) * 60 * 60 / window span in seconds

So, resolving it, that's (1 / 130) * 60 * 60 / 2 = 13.85 l/h deviation with the default 2s window.

Now, for the Aquaero, which is what I'm going to use now (not that my previous solution wasn't fit for purpose, is that the software and configuration are IMHO awkward beyond words), I really don't know how to proceed. On one hand Sven told me that 130 ppl is too low count to obtain an accurate reading.

On the other hand, the guys @ effizienzgurus , backed up by Sebastian (from Aquacomputer) on this post on the forum, say that the Aquaero not only reads the ppl value, but also the time between pulses, therefore it would be more accurate with lower ppl values.

So, summarizing, it's 130 ppl too low for an accurate reading, or I can get away with using this flow meter? And how is the deviation calculated on the Aquaero?

Finally, although not completely on-topic, Digimesa recommends to calibrate the flow sensors once they've been deployed. In an open system this wouldn't be an issue, you just need a bucket with level marks and a stopwatch, but in a closed loop, the only way to know how much water you've moved in a given time is... yes, another flow meter but this time one that measure mass rather than flow... Or an analog flow meter... but this is catch 22, because unless you're planning to leave them there forever, the moment you remove them, you've already changed the layout, therefore you would have to calibrate the sensor again.

Any suggestions on how to break the catch 22?

Thanks for reading

Zitat von »fox3«

What pump(s) do you plan to use? That sensor does not start to read until 3 l/m.

A DDC 3.25 w/EK DDC v2 top on one loop (Res > Pump > TFC 360 > MIPS NB Block > MIPS Mosfet Block > MIPS Mosfet Block > Res)

and either

Two Aquastream XT Ultra (Res > Pump1 > TFC 480 > TFC 360 > Pump2 > DD 4870x2 FC > Apogee GTZ > Res)

or

Aquastream XT Ultra (Res > Pump > TFC480 > DD 4870x2 > Res) [leaving headroom for adding a 2nd card]
Aquastream XT Ultra (Res > Pump > TFC360 > Apogee GTZ > Res)

all with DD FatBoy barbs and Tygon 1/2" ID silver tubing. And as little BitsPower 90s and 45s as possible.

If with either setup I'm not close to 5 lpm I'm going to be mighty disappointed

Right now, with the same blocks on the main loop (CPU and GPU), two PA120.3 in series, two DDC 3.25s in series with EK DDC V2 tops and using 12 mm push-in fittings with semi-rigid PU tubing (and more 90s than you can shake a stick at), I'm doing 4.1 lpm

Zitat von »fox3«

I am using the Digimesa FHKUC 70 hose, http://www.uptom.cn/data/digmesa/fhku/938-3870AE032_GB.pdf, and am using 165 ppl with no problems to the aquaero. I have not calibrated it in closed loop but did a bucket test and it is giving similar results to closed loop.

I originally had the AC sensor installed, the one that does a 180 ° U Turn with a 5.6 mm nozzle and am getting exactly the same flow reading with this sensor I am using now

I wouldn't touch those with a barge pole Not even with a RD-30. Flow is linear, pressure is not... (I mean the u-turn ones)

Zitat von »fox3«

It really is not too important to know exactly what the flow is but rather the temps are more important no matter what flow you can achieve. It is prolly better to use the flow sensor as a means of shutting down the PC in the event of blockage in the loop or pump failure more than being able to determine the actual flow. Sorry if I am unable to help you, but the only way to know for sure if 130 ppl is too low is to try it.

Agreed. I'm not dispensing petrol , and the main objective is indeed to shut down the computer in case of catastrophic failure. I guess I'd go with these and see what happens. Thanks a lot for taking the time to reply though

Zitat von »fox3«

I just bought 2 DDC 3.25's with V2 tops, should have them by mid week, hope they are quiet like the eheims. Planning to split my loop up and put the chipset blocks with the GPU and leave the CPU on its own with the 2 Ultra's.

Right now I have an AC cuplex xt di and BitsPower NB/ SB with 3 mosfets on one loop with 2 aquastream XT Ultra pumps and 3/8" tubing plus feser 480 rad, getting less than 3 l/m flow. I am going with the PMP-400 pumps on the GPU loop.

Good luck with your setup

Hmmm... sounds about right. There's no much info available regarding these pumps, but general consensus seems to be that @100 Hz they're like the D5 vario @ setting 5. And the Cuplex XT is pretty restrictive (although it does look like sex on fire, if I may )

If they can't "pull their weight" so to speak, I will split the loops, leave them for the CPU as well, and add two DDC 3.2s I've got in the "dark chest of wonders" (euphemism for "box where you put all the stuff that you bought and didn't perform as expected, or by mistake, or it has been superseded") on the GPU loop.

Possibilities are endless

OK, here's how one can de-rail his own thread

Little update: I'm going to give the G1/2 flow meters a go. But there has been a little change on the pump department. After all I don't think the XTs are powerful enough

So, a friend of mine has bought the Aquastream XTs, and I'm going with 4 DDC 3.25s (2 on the NB/Mosfet loop, 2 on the CPU/GPU loop).

And may the coolant flow!

Zitat von »fox3«

Sounds like a plan

Where are you getting that sensor from? I had a problem finding a supplier for any Digimesa products at a decent price.

I'm getting it from Farnell. But forget the "decent price" part. The company I work for has an account with them, and I get a nice discount (plus not paying VAT). Otherwise the expenditure is preposterous (is bad enough with all the discounts).

The "normal" price is £74.65. There's another G1/2 flow meter for a more reasonable £44.08, but it's only 65 ppl

Zitat von »fox3«

That GPU throws off a ton of heat. It is not impaired by the heat as much as the CPU is. If you are gaming then these two will be dumping heat into the shared loop.

Mebbe consider putting the CPU and GPU on seperate loops and adding the NB and MOSFETS to one or the other or split them between the two.

Hmmm... I might have to look into that. Thanks for the suggestion