Choice between two evils

[MrToad]

Well, I have most of what I need gathered for my next rig. I've changed it's shape about 15 times in the last six months, but finally we got there

Now, I have one last niggle where I'd appreciate some (constructive) input.

As I don't trust the MB to power off my rig in case of catastrophic pump failure, because it attempts a civilized "soft power-off", which can (and will) be interrupted by a BSOD, instead of a more abrupt "hard power-off" (I don't mind losing a hard drive if that ensures I don't lose the rest of the hardware), I'm going to use a flow meter connected to a "safety device" of sorts (www.keiang.de for more info).

Now, I have two of those available, each one with it's own ups and downs, as usual.

First option is the all-too-well known Swissflow SF800. Upside of it is that it has a negligible footprint and it's very precise. Downside, it has a considerable pressure drop (0.25 bar @ my projected flow rate). For the sake of comparisons, is like an extra Stealth Rev2 on the loop (minus the heat dump, of course).

Second option is the Digimesa FHKU G1/2. Best thing is pressure drop. Completely negligible. 0.020 bar @ my projected flow rate. This is like adding a DD 90 degree elbow. Bad things, is huge, although I've found a way to fit it in the loop without being (too) obtrusive, and, this is more due to limitations on the design of the "safety device", where the deviation of the reading is "(1/ppl)*60*60/2=lph", than the meter itself (which only emits 65 ppl), it produces a rather inaccurate output.

Common sense tells me "Use the Digimesa, who gives two dead monkeys and a flying f**k about accuracy, this is not a laboratory test device", but I thought I may ask just in case I'm overseeing something.

As you'll notice, there's no poll option for "use the rpm output of the pumps and trust the MB", for the aforementioned reasons. I don't want to sound abrasive or abrupt, but I've already pondered the alternatives that don't involve a flow meter, and I'm not happy with them.

I've considered the alternative of connecting the rpm output of one of the pumps to the safety device, but if the pump that is not connected to it fails, only one pump is not enough to keep a decent flow on my loop.

Constructive input and discussion is welcome

[MrToad]

For the sake of completeness, this is how the loop goes:

EK MultiOption Res Linked > DDC 3.2 w/EK top > DDC 3.2 w/EK top > PA120.3 > PA120.3 > Flow Meter > Fuzion w/Quad Nozzle and Washer > Stealth Rev2 > BlastFlow Siberian Modular 8800GTX VGA Block > Res

The blocks will be cooling a QX9650 and two 8800 GTX.

After giving up on sourcing the Sanyo Denkis (I'm fed up of making HRM's Customs happy) I will use the Zalmans on the rads.

The BlastFlow may be replaced by a Stealth Rev2 if it doesn't perform as expected. Alternatively, if it gives me a (pleasant) surprise, a second one may find it's way into the loop.

[ladderman]

Even though I have a SF800, if I was you I would go for the least restrictive of the two if you have all your components in one loop. Flow would be the most important factor in my decision. Plus you mention that you are just using it to start a shutdown if the flow rate stops.

I did some basic flow tests with the SF800 and it’s about the same as a EK 8800 GTX FC.

Mine will be in a loop with just one 8800GTX and the CPU.

[MrToad]

I think I'm going to follow the "common sense" (and ladderman's chosen) option.

With a caveat.

I will use the Swissflow to check what kind of impact the BlastFlow has on the overall flow, because that's where I need a bit of accuracy, then remove it and hook the Digimesa.

In the end of the day is not going to act as a measuring tool but as a safety device. And the least intrusive a safety device is, the better

Thanks ladderman for your input and everyone else for your interest and votes

[Xilikon]

Before taking a final judgement about the SF800, wait till Martinm210 do his own pressure drop testing with the right adapters that he is making right now. He have a sample that I sent him.

The SF800 hole seems to match 3/8" tubing so the drop should be negligeable and it's the adapters provided by Swissflow which is the main source of restriction.

[billb]

Why not sense pressure instead of flow?
A simple pressure switch on a tee has no restriction.

[Wolfsburg18]

For the sake of completeness, this is how the loop goes:

EK MultiOption Res Linked > DDC 3.2 w/EK top > DDC 3.2 w/EK top > PA120.3 > PA120.3 > Flow Meter > Fuzion w/Quad Nozzle and Washer > Stealth Rev2 > BlastFlow Siberian Modular 8800GTX VGA Block > Res

While there has been debate you could really test the fact of having two pumps but at different points in the loop one just after the res and and other maybe after the fuzion followed by the other radiator. This might better help maintain loop pressure/flow.

[MrToad]

Before taking a final judgement about the SF800, wait till Martinm210 do his own pressure drop testing with the right adapters that he is making right now. He have a sample that I sent him.

The SF800 hole seems to match 3/8" tubing so the drop should be negligeable and it's the adapters provided by Swissflow which is the main source of restriction.

Well, I'm just going by the manufacturer's published pressure drop curve. However, I can wait until Martin finishes with his tests. I've been waiting six months, so a bit longer is not going to make any difference

Why not sense pressure instead of flow?
A simple pressure switch on a tee has no restriction.

Because I don't have any safety device that can work with a pressure sensor, and my knowledge of electronics is not good enough to produce one

While there has been debate you could really test the fact of having two pumps but at different points in the loop one just after the res and and other maybe after the fuzion followed by the other radiator. This might better help maintain loop pressure/flow.

I can't test pressure, as I lack the appropriate equipment. However, I can measure flow, so I might actually do the experiment

[Conumdrum]

Flow rate and pressure on a WC system are related to the pump being on right? Both of these sensors measure flow. What about a pressure sensor on a T right after the pump? A good T has little pressure drop.

If you made an easy 30 second delay relay setup with a pressure sensor coupled to the power button (or even the 120 VAC to the PSU) it would shut down if pressure didn't come up on 30 seconds. It would shut power off if the pump lost pressure any time after 30 seconds. The 30 seconds is very adjustable by diff relays and/or adjustable relays, depending how fast your pump comes up or how much time is safe.

Building it is a non issue. Where I work I can source a sensor/relay/relay block easy. Problem is cost on a lot of these items. Sensors like this are for industrial equipment and not cheap. Sensor might be $100+, relay $25, block 15$.

I'm sure by now it's been done. A pressure sensor on a T would help us WC'ers but the cost........

Been done, linkies to stuff like this? I wanna do it, but rather link to "It's been done already,NOOB!" linky.

[Xilikon]

While a pressure sensor is a great idea to detect the lack of flow, it is useless to detect any flow issues. A prime example of this (taken from a real story) is that someone is always puzzled to see only 0.8 LPM no matter how the sensor is installed and the output computed. After he decided to drain the loop to fix a different issue, he noticed the Storm block is being clogged so he cleaned it then the flow picked up a lot to 1.6 GPM (IIRC, he got a Iwaki or something very powerful).

If we know what should be the target flowrate by using the Martinm210's flow estimator, we can use the SF800 or any flowmeter to see if it is running correctly or if air bubbles is impeding the flow.

[billb]

Sensor might be $100+,

I'll bet these are less than $10.
http://www.gemssensors.com/content.aspx?id=1198
These guys will even send a free sample:
http://designflexswitches.com/design...specifications

[Conumdrum]

Wow billb, thanks!

So Martin got a sample, looking forward to the test. I'd hook the output up to a time delay relay to the PSU AC input, depending how long it takes for pressure to stablize on power up.

If I don't screw it up totally I'll post a build log for it.

Edit:
Looked up and called on the first link ya gave me.
.75-4psi adjustable
2% variablity at full range (close enough)
1/4 NPTM brass fitting
SPDT switch
18" leads
Factory preset at 1.25 PSI
$42US
5 amps 120/240AC

Amps is too low. The other link is 15 amps, but +/- 20%, and at our low PSI range, I think it would give false readings and cause shut downs.

So putting one of these right before the pump input in a T would tell us if we have a blockage, pump broke, or a major blowout. I don't see why anyone would put it after the pump.

Comments?

[Waterlogged]

I'm sure you can find a suitable pressure gauge that works on 12VDC, all you'd need to do at that point is tap a line into the green sense wire from the MoBo and add some power and your pretty much done.

[bogg]

I've considered the alternative of connecting the rpm output of one of the pumps to the safety device, but if the pump that is not connected to it fails, only one pump is not enough to keep a decent flow on my loop.

I'm sure you've already considered it, but i thought i'd just point out that the BigNG & sensorhub will monitor rpm from 2 pumps, and if either one fails it will perform a Hard shutdown if the soft shutdown fails.

[sanhacker]

I'm sure you've already considered it, but i thought i'd just point out that the BigNG & sensorhub will monitor rpm from 2 pumps, and if either one fails it will perform a Hard shutdown if the soft shutdown fails.

Bingo!
That's exactly what I'm doing at the moment. I'm also using the miniNG to power those two pumps. That allows me to lower the voltage to the pumps while the sensor hub has the RPM leads to ensure liter/minute rates do not go to zero.

[MrToad]

I'm sure you've already considered it, but i thought i'd just point out that the BigNG & sensorhub will monitor rpm from 2 pumps, and if either one fails it will perform a Hard shutdown if the soft shutdown fails.

Thanks for pointing it out

I did consider it, but I didn't like the fact that's not standalone. It needs the computer to perform it's duties, that's why I went for keiang's solution

[Xilikon]

While you need a computer to configure the BigNG/Sensor hub/miniNG, they can run themselves with the help of the computer so even if the OS crashed, it can still monitor beside full power loss (if there is no power, your computer wouldn't be running anyway so you are still safe).

[MrToad]

While you need a computer to configure the BigNG/Sensor hub/miniNG, they can run themselves with the help of the computer so even if the OS crashed, it can still monitor beside full power loss (if there is no power, your computer wouldn't be running anyway so you are still safe).

I see what you mean, and it makes sense. However, I already have the full "kit" so to speak, plus a very nice display to go with it.

Switching solutions at this point is rather expensive

[Xilikon]

That's true and that's understandable I thought you are just looking at a solution and that's why I made this remark. The pressure switch is a nice idea worth considering.