How to do disaster management?

[Gooserider]

I have a question that occured to me last night. What kind of DISASTER RECOVERY methods do people use, and what is the best way to implement them.

These are my thoughts, I would like to see other folks comment on them, point out anything I might have missed, or tell me how they deal with a given failure type. I didn't find anything useful when I searched on the subject here, If there are threads on it already that I missed, please let me know.

I intend to have the system I'm building running 24/7 (I think it's better for the hardware, and I don't like waiting for bootups) This naturally causes me concern about how to handle different system failures, especially the ones that might occur when nobody is at the console. I'm doing a plain WC setup, no pelts, extreme tech. etc. All of my cooling hardware will be in or on my case. I will be running Linux, but the functions would be similiar on that other OS, even if the software is different. (i.e. Linux uses lm_sensors, MS stuff uses MBM or equivalent)

As many folks have pointed out, there are risks in WC'ing a system, I think there are also risks in an AC'd system, just different ones. However I would think that advance planning and care in system design can manage the risk so as to minimize the damage risks.

DEFINITION: A properly managed risk is one where a failure causes no hardware damage to the system beyond the failed part itself. Data corruption or system downtime is undesireable but totally acceptable if it prevents hardware damage.

Failures in the PC hardware itself I don't see as an issue because they will handle (or not) the same way no matter how the system is cooled. (Note that WC'ing is supposed to reduce the odds of hardware failures by giving better temps, but that's a different issue)

Five failure modes exist for a WC system that I can see. Each would have a different set of symptoms, detection methods, damage potential, and optimum handling to avoid damage to data or hardware.[list=1][*]Partial pump failure / flow restriction[*]TOTAL pump failure / flow blockage[*]Radiator cooling (fan) failure[*]LEAKS[*]Excess temperature rise[/list=1]
(Combinations of the above might happen, but I suspect that one would actually happen first, and trigger others if not handled)

To take them one at a time:

1. Partial pump failure or flow restriction -
Symptom - Temperature rises to new equilibrium point, how high depends on level of failure.
Detection - Flowmeter and / or Normal temperature monitoring, mobo or digidoc based.
Handling - Varies and depends on severity. Certainly generate an alarm (can't fix it if you don't know it's broke, and can't count on user checking) Severe case should be treated as total failure. Less severe case might turn on / increase speeds on fans to increase cooling levels and bring temps back down.

2. Total pump failure or flow blockage.
Symptom - Temperature rises to MELTDOWN levels - I'm not sure how fast... (opinions anyone?)
Detection - optimal would be some kind of flow detector, (flow meter, pressure switch, etc.) If that doesn't exist, look for temperature increases.
Handling - Trigger alarm? Shut system down. Opinion requested how fast would depend on speed of temperature rise and detection method. (Can a digidoc talk to a mobo BTW? also, can a Digidoc and a mobo share temp or fan speed sensors?) If there is enough time, shut down gracefully ('shutdown -h now') otherwise trigger a relay to interrupt power and slam system off. (deal with any resulting data loss / corruption later)

3. Radiator cooling (fan) failure.
Symptom - Similiar to 1 or 2 above, but slower (more water to heat)
Detection - Fan speed monitoring optimal, or look for temp increase.
Handling - Trigger alarm, turn on / increase speed of any redundant fans, other case fans. If temp rises beyond 'comfortable' levels, shut system down, preferably gracefully (there should be time)

4. LEAKS!!!
Symptom - FLOOD, Possible major component damage!
Detection - Nothing I know of as standard PC equipment, I would consider modding a basement water detector with the sensor(s) in the bottom of the case and other vulnerable spots. I would also consider a level sensor in the res, if you use one.
Mods to detector -

• Multiple sensors (sensor is just two contacts seperated by a space, I see no reason you couldn't have multiple sets wired in parallel)
• Add output to trigger shutdown relay

Handling - Two parts here, one beforehand, one after it happens:
Beforehand - Consider putting conformal coating or other waterproofing on all boards. Questions -

• Would this potentially void warranties?
• Could it have any thermal consequences for mildly hot parts?
• Could it have any functional effect (changing capacitive/inductive properties of traces for instance)?
• Would this influence any other mods on the board?

After the leak happens SHUT DOWN ALL POWER IMMEDIATELY!!! That should minimize any damage from water shorts. (most electronics can deal with being soaked if they aren't powered, and are dried well before being powered back on, its power and water together that do things in...

5. Excessive temperature rise - This is kind of a last ditch defense intended to catch anything that didn't trip another alarm.
Symptom - Temperatures trending towards potential meltdown. CPU temperatures exceeding normal operation range significantly.
Detection - I see two points needed, 'Concern' and 'Panic'. 'Concern' I would trigger at say 10*C over normal maximum. Panic I would trigger at 10*C over 'Concern'. (or well below CPU damage threshold) Concern level triggering I would use on-board monitoring. Ideally Panic level I would trigger with an off board monitor to ensure a process wouldn't keep it from triggering.
Handling - Concern - start a graceful shutdown...
Handling - Panic - Shut down power immediately!

I know this was long, but tell me what you think...

Gooserider

[sevisehda]

I've been running an OCed system for a year now and Havn't had any problems except the ones I created. Tangent...Like the time I wondered how low I could go and dumped 6 gallons of ice into my bong. Not realizing everything in the loop would get covered in water.

As to prevent damage:
1) Go over your system weekly making sure there aren't any small leaks. I had an almost microscopic leak on a block and it took me 3 weeks of use to spot it even after I hooked a compresser to it and did the bubble test.(Bad solder joint or something).
2) Make sure your tubing isn't wearing away.
3) Check coolant levels
4) Listen, dying machines often make evil sounds

I would also set my MB or OS to shutdown at critical temps. Even with a pump failure the block and waters thermal mass will prevent an instantaneous failure. If you had an external monitor like a PIC or STAMP you could have it monitor temps, flow and rpm then flip a relay if one should go beyond a threshold.

[jaydee]

Hose clamps and over heat shut down set to a low temp that will not lock the computer up before it can shut down the comp. Been working 24/7 for a few years now and multiple systems just peachy.

[SparkedFire]

i thought about getting a probe and setup that would detect water if it was to leak. place a probe under the WB, pump, tubes, etc. It would send an alert (alarm, buzzer) and do a sys shut down.

never did impliment thoe.

If you smell something odd, then panic.

[Gooserider]

Well, maybe I'm over paranoid... But this system is a really big chunk of my disposable income - I can't afford to blow it up!

Thanks for the kind words at any rate, they do sound reassuring, but I know that I have a tendency to not be as good as I should be about checking stuff. I'm trying to design around that as well.

SparkedFire: i thought about getting a probe and setup that would detect water if it was to leak.

That one is actually one of the features I thought would be easiest to do... Home Depot sells a nice little alarm box about the size of a pack of smokes, runs off a 9V battery, and has a cable with a probe on it. If the probe gets wet, the box sets off an alarm about as loud as a smoke detector. Cost was around $10 as I recall. Harbour Freight Tools sometimes sells a similiar box for even less.

I took an older one apart once, and it was pretty simple inside. I'd be willing to bet it would be possible to pick a signal off the alarm speaker and use that to trip a relay that would kill the system power. I might consider replacing the battery with a wall-wart or tapping into a PC supply line, but it really isn't needed, the batteries last for years. The only question might be whether it would have enough power to trip the kill relay.

The sensor is just two bare metal pins about 1/4" apart, mounted in a plastic insulator. (it apparently works by the water shorting out the pins) I see no reason why you couldn't hack the sensor cord to put lots of parallel test points into it. The only caution is that the pins will detect contact with a case part as a leak, so some care in mounting is indicated.

Just guessing, but I think you could probably build the whole setup in a few hours and it wouldn't cost more than $20-30. As a bonus, it should be possible to use the same kill relay for any of the other above alarms I suggested where slamming the power is appropriate.

Gooserider

[bigben2k]

Well, you're right to be cautious.

For leaks, since they can come from any point, Conformal coating is still, IMO, your first defense.

Otherwise, a flow sensor, and a temp gauge (in the water) is my second step, while letting the mobo (with OCP) protect the CPU.

Hose clamps are a life saver

Otherwise, to detect water, you can use the old salt pill in a clothes pin: make contacts there, and put some salt: when the water dissolves, the contacts trigger.

[pHaestus]

leaks don't usually just come out of nowhere. Use hose clamps and then put UV dye in the water when you fill. After you clamp everything in place then turn on a uv lamp and verify you don't have any leaks around barbs or clamps. .

[jaydee]

Well, maybe I'm over paranoid... But this system is a really big chunk of my disposable income - I can't afford to blow it up!

First off of this is such an important and your only comp that you don't want damaged then drop the idea of messing with it at all. Get a good adjustable speed air cooloer and be done with it. Save up a few hundred and buy a mobo/ram/PS/vid card to test on. Don't need a case for this or expensive parts. If I was worried about breaking things I sure the hell wouldn't be testing blocks on my comp. Your going to break things when putting on and taking off blocks consistantly (testing in other words). Water will get dribbled, mobo's will die, ect.... Got a stack a dead mobo's to show for it over the last 3 years and expect at any moment to break things. I am willing to afford though. If you can't then don't mess with a perfectly good system.


But anyway water sensors are not to hard to make and customize.

http://hallbeng.jlab.org/firesafety/watersensor.htm
http://www.electronic-circuits-diagr...msimages/4.gif

www.google.com search Water Sensor Circuit and it will come up with ideas on it. You can also get a flow switch. If the flow stops you can set it up to trip the relay and shut it all down.

[Cova]

Depending on the size of your CPU block, most of those failures can be captured just with regular temp monitoring and a "shutdown -h now" when temps get too high. Here's my experiences with various components not working (either I forgot to plug something in, or simulated a failure). These are all cooling a OD'd TBird @ 1.85V using a Maze-2 block

1. Rad-fan failure. If system is idling, temps go up fairly high, but systems keeps running and remains stable. Just playing MP3's (about 3% load) I can turn off every fan in the system except the PSU fan and listen to music for hours without a crash. Under a load, I overheat and lock up after about 10 minutes, but as it is a very slow temp rise, it locks up and cools back down before any damage is done. Conclusion - even on a heavily loaded server a rad-fan failure will give you at least 10 minutes of warning based on CPU temp reading for a clean shutdown.

2. Pump failure. I once forgot to plug in the pumps AC cord after some system maintenance. First boot (block and coolant at ambient) I made it into windows, logged in, and started Outlook before the system locked. Hit reset-button, got to windows logon prompt. Hit reset button, got almost to logon prompt. Reset two or three more times, all crashed during windows boot process. Powered off machine, found pump problem, plugged it in, turned on comp, booted into windows fine and ran the rest of the day as expected. Conclusion - even on a heavily loaded server a pump failure will give you at least 2 minutes of warning based on CPU temp reading for a clean shutdown.

3. Partial pump failure. I haven't tried to simulate this, or any other type of obstruction of flow while the system is running. Based on my observations running with no pump at all though if you have the system set to shut down at a certain temp, if this restriction is severe enough the existing protection will catch it, if it is not that severe, the system will continue running stably.

I've never had to deal with a leak before. But based on my experience running with no pump, the mass of copper in a good-sized block along with whatever water remains inside the block not moving (I don't think most blocks will drain completly without being rotated some) can absorb enough heat for a system to run for a minute or two. Even if you lose most of the coolant out of the block, you should still have a minute of temp-warning for a clean shutdown.

[Gooserider]

bigben2k Well, you're right to be cautious. For leaks, since they can come from any point, Conformal coating is still, IMO, your first defense.

I agree, but I'm concerned about what it might do to warranty coverage? Even if it does trash the warrantee, I might still go for the conformal coating since they probably wouldn't cover water damage anyways, and there isn't that much else to go wrong on a modern mobo.

Otherwise, a flow sensor, and a temp gauge (in the water) is my second step, while letting the mobo (with OCP) protect the CPU.

Makes sense, but so far I haven't been able to find an affordable flow sensor - any suggestions? I'm even watching e-bay and not having much luck there.

Hose clamps are a life saver

I agree, I wouldn't build a system or any other plumbing without them! The most I might be persuaded to do is to use zip-ties instead of worm drive clamps, and that takes special circumstances.

BTW, My current plan is to use braid reinforced polyethelyne (sp?) tubing - the kind often used in soda fountains, etc. My local hardware store has a good selection, I know the stuff is pretty strong, and it seems very resistant to kinking. Any reason you know of NOT to use it?

Otherwise, to detect water, you can use the old salt pill in a clothes pin: make contacts there, and put some salt: when the water dissolves, the contacts trigger.

Neat idea, but I think I'll stick with the electronic detector idea - MUCH faster response times, when seconds might count!

pHaestus leaks don't usually just come out of nowhere. Use hose clamps and then put UV dye in the water when you fill. After you clamp everything in place then turn on a uv lamp and verify you don't have any leaks around barbs or clamps.

Great idea... I hadn't been planning on putting anything in the water beyond Water Wetter or equivalent. I'd always thought the dyes and stuff were just for the folks doing showoff rigs, didn't realize there was a practical reason for some of it. (I'm not planning to mod my case except as it adds definite functionality)

jaydee116: First off of this is such an important and your only comp that you don't want damaged then drop the idea of messing with it at all. Get a good adjustable speed air cooloer and be done with it.

Some ways I agree with you jaydee116, and I'm not planning to mess with this box on a continuing basis once I get it built... One of the reasons I'm doing so much work on investigating everything and planning what I'll be doing is so that I can put it together once and be mostly done with it.

I would go with the air cooled approach, except that I can't stand the racket I already get from my old PC with TWO fans - one little job on the CPU cooler and one on the PSU, plus a couple of old SCSI drives. I already know that I can't do a "bladerunner grade" silent system, but I want to get as close to it as I can, and water cooling seems like the way to do it.

Cova Depending on the size of your CPU block, most of those failures can be captured just with regular temp monitoring and a "shutdown -h now" when temps get too high. Here's my experiences with various components not working

Thanks for the input Cova, I really appreciate the 'real world' numbers, it gives me a bit better feel for how to do the disaster planning.

I know some of this may seem like excess paranoia, but you get into certain mindsets when you spend years working on the kind of high reliability / redundant systems that I have.

Gooserider

[logosmani]

Originally posted by Gooserider
I agree, but I'm concerned about what it might do to warranty coverage? Even if it does trash the warrantee, I might still go for BTW, My current plan is to use braid reinforced polyethelyne (sp?) tubing - the kind often used in soda fountains, etc. My local hardware store has a good selection, I know the stuff is pretty strong, and it seems very resistant to kinking. Any reason you know of NOT to use it?

not sure exactly what kind of rigidity reinforced polyethelene has but it may put undue stress on your waterblock mounts, so watch out for that. Only thing i can think of to watch out for. Unless the braids conductive of coz.

[Static-D]

Before you go with a water detection sensor you may want to make sure that it will set off when using distilled water. I work in a refinery and we use a setup in our clean rooms that detects changes in moisture, if there was a leak you could use this device to kill the entire system if moisture above a certain percent is detected.

[bigben2k]

I think that a Disaster Recovery effort starts with a good design. Avoid using undersized barbs.

The braided stuff is fine, but stiff. If the block is lifted above the CPU, it will fry. I'll be using 3/4" ID braided vinyl tubing myself, along with some regular 1/2" ID simple vinyl tubing.

It should also be understood that the OCP has limits: it will not protect your CPU, if you suddenly take off the HSF or waterblock, while it's running.

A lot of the risks involved have a potential to appear when the rig is first run, while other risks are more long term. You ought to categorize them.

I'd forget about the water detector: the flow sensor will pickup on a fault. OC has an article on making one, out of an old fan, and copper fitting. Otherwise, there's one that was posted recently.

[Gooserider]

logosmani: not sure exactly what kind of rigidity reinforced polyethelene has but it may put undue stress on your waterblock mounts, so watch out for that. Only thing i can think of to watch out for. Unless the braids conductive of coz.

Far as I know the braid isn't conductive (I think it's nylon or some such) and it will be pretty well isolated anyways. The stuff is somewhat stiff, but not excessively so, and I think I can arrange it so that the curve of the tube will tend to push the block onto the CPU rather than pulling it off. (I'm somewhat more worried about the weight of the blocks, I might try to find a way to add some additional support to them)

Static-D: Before you go with a water detection sensor you may want to make sure that it will set off when using distilled water. I work in a refinery and we use a setup in our clean rooms that detects changes in moisture, if there was a leak you could use this device to kill the entire system if moisture above a certain percent is detected.

Good point, I will need to check that. Of course I am planning to use water-wetter (or equivalent) and the water will pick up stuff from the plumbing so it won't stay 'distilled' for long. (Extra true if I end up using a used heatercore.)

bigben2k I think that a Disaster Recovery effort starts with a good design.

Agreed totally, which is one reason I'm doing so much research. I am getting the knowledge needed to DO a good design.

Avoid using undersized barbs.

I am planning to. I'm not thrilled with oversize either, but I don't think it's a problem to get barbs that match the tubing, and to make sure that the barbs on each end of a tube are the right size. I see some folks mentioning reaming out their barbs, but I consider that a bit risky as it would take away strength as well, for a gain that I don't thing is that significant. (I especially don't care with the pump I'll be using... I just got an Iwaki MD20 off of e-bay)

The braided stuff is fine, but stiff. If the block is lifted above the CPU, it will fry. I'll be using 3/4" ID braided vinyl tubing myself, along with some regular 1/2" ID simple vinyl tubing.

Sounds about right, I am planning to go to great lengths to ensure my plumbing is well supported.

It should also be understood that the OCP has limits: it will not protect your CPU, if you suddenly take off the HSF or waterblock, while it's running.

Understood. I am planning a 'last ditch' relay to crash off the system if anything goes wrong and it looks like it's approaching meltdown. Wonder how I could add 'block mount failure' detection. Perhaps an N.C. microswitch attached to the block and pushing on the board or CPU??? That might do it OK. How about putting a switch under the hold down clamp point?

I notice that many folks use springs to mount their blocks, doesn't this increase the odds of a block lifting? What are the pros and cons of springs vs. solid mounts?

A lot of the risks involved have a potential to appear when the rig is first run, while other risks are more long term. You ought to categorize them.

Agreed, although most of the risks I see as both short and long term following a 'bathtub curve' trend. I would say that leaks and comonent failures are somewhat likely initially due to 'infant mortality' Once things have been running for a while, I would expect long term reliable operation, but then as they get older I would expect to see some 'age failures' from things like motor wear, etc.

I'd forget about the water detector: the flow sensor will pickup on a fault. OC has an article on making one, out of an old fan, and copper fitting. Otherwise, there's one that was posted recently.

Hmmm.... I've been getting a different message electronic Flow meters seem a bit hard to come by, and folks say they add alot of resistance (unless one uses the big bucks magnetic jobs). OTOH any flow restriction or stopping is going to produce a temperature rise. While I agree that a flow sensor would EVENTUALLY detect a leak, it might not do so until most of the water was pumped out of the system. Water detectors are easy and cheap, and presumably would give a very early warning, potentially while the flow sensor is still happily measuring the rate at which the pump is dumping all the water out of the system!

Gooserider

[BladeRunner]

I've got a little involved in this and although it's not finished the way I want it yet, as I've a liquid level detector to fit to the res, and water detectors to the base of the PC case, I have a very simple low restriction flow sensor / shutdown system that also wont allow the PC to be started unless the pump is plugged in and on.

more on that HERE

The key to lack of leaks is really in the hands of the, individual, the design and application. I like the festo style push fittings for all the other system blocks than CPU in my system, (the CPU benefits from a bore size that makes push fit types unsuitable). The fittings are rated 10 bar air and so far I've had no leak issues with them. They are also simple to disconnect reconnect when messing about with the system. With the CPU block using standard barbs I always use one larger than the tube, (barb ID same as Tube ID tygon r-3603 on 5/8 barbs for instance), so the tube requires boiling water immersion and a lot of effort to get it on. Once on no ugly, and possible leak causing hose clips are not required, as the tube will not pull off even using all my weight and strength with the barb in a vice. As far as blocks go I solder them together, and make sure to air pressure test them with 4 bar air, the block immersed in water. The block is place under an upturned glass bowl underwater with all air removed and left overnight if there is even the slightest air leak it will show as trapped air in the upturned glass bowl.

I'm not arrogant enough to believe I will never have a leak or disaster but the risks can be significantly reduced with some care and forethought, the most important thing is logical working procedures, when putting it all together. I have an acceptable situation in my mind, and that is one major water-cooling disaster every two years, to me is acceptable to have silent Water Cooling over noisy air-cooling........... so far 1 & a half years on...... (crossed fingers), no leaks or death by water-cooling yet

[bigben2k]

Originally posted by Gooserider
Hmmm.... I've been getting a different message electronic Flow meters seem a bit hard to come by, and folks say they add alot of resistance (unless one uses the big bucks magnetic jobs). OTOH any flow restriction or stopping is going to produce a temperature rise. While I agree that a flow sensor would EVENTUALLY detect a leak, it might not do so until most of the water was pumped out of the system. Water detectors are easy and cheap, and presumably would give a very early warning, potentially while the flow sensor is still happily measuring the rate at which the pump is dumping all the water out of the system!

Gooserider

You really can't do anything about the HSF/WB popping off the CPU: just keep it bolted. That's a design requirement.

Some flow meters add a lot of resistance, but flow sensors usually don't, if properly selected. I can't believe that you didn't find the one that was presented recently.

BTW, you're off your rocker about the restriction adding heat: while the heat will appear at the restriction, removing the restriction will only cause the heat to appear at another restriction(s) within the loop Just forget about the heat thing.

The flow sensor will protect your CPU, even if there is little flow: that's all you need. When there's no flow, then the CPU will eventually, over a certain period of time, overheat and die.

The level sensor is a good idea, as long as none of the coolant is evaporating. Even better is the Conformal coating, and an allowance for the leaked water to exit the case. The real problem, is figuring out the possible leak points.

[Gooserider]

BladeRunner: I've got a little involved in this and although it's not finished the way I want it yet, as I've a liquid level detector to fit to the res, and water detectors to the base of the PC case, I have a very simple low restriction flow sensor / shutdown system that also wont allow the PC to be started unless the pump is plugged in and on. more on that HERE

Nice article, similiar to what I'm planning on the PSU setup though I'll probably do the wiring a bit differently. I plan to have a relay that will kill the power to the PSU and pump if a problem is detected (If I have a leak I want to get the pressure off the system and keep the leak minimized if possible)

I didn't see anything about the level detector, though I think that should be pretty easy, at least for a 'low level alarm. I plan to put a sight tube on the res. to make visually checking the actual level pretty easy. If I put a float in the sight tube with a magnet in it, I can do proximity switches to get a rough quantity reading. Alternatively, I might be able to do something with adapting an automotive gas gauge... BTW, do you happen to know of a US source for those flow sensor switches?

The key to lack of leaks is really in the hands of the, individual, the design and application. I like the festo style push fittings for all the other system blocks than CPU in my system, (the CPU benefits from a bore size that makes push fit types unsuitable). The fittings are rated 10 bar air and so far I've had no leak issues with them. They are also simple to disconnect reconnect when messing about with the system. With the CPU block using standard barbs I always use one larger than the tube, (barb ID same as Tube ID tygon r-3603 on 5/8 barbs for instance), so the tube requires boiling water immersion and a lot of effort to get it on. Once on no ugly, and possible leak causing hose clips are not required, as the tube will not pull off even using all my weight and strength with the barb in a vice.

I'm not crazy about those type of fitting, I don't like something that makes such a critical issue of what sort of hose is used with them. I prefer barbs. Maybe I'm over conservative, but I tend to use a barb size that matches the hose, and either worm clamps or tie-wraps. Clamps only cause leaks when the're improperly tightened.

I won't argue with your results, but I really would use at least a tie-wrap on your barbs, no matter how tight they seem. I have seen pressure cause a hose to stretch just a little bit as it ages, and either start seeping or blow off the fitting. A clamp (or tie wrap) puts a limit on how far up the fitting the hose can stretch.

As far as blocks go I solder them together, and make sure to air pressure test them with 4 bar air, the block immersed in water. The block is place under an upturned glass bowl underwater with all air removed and left overnight if there is even the slightest air leak it will show as trapped air in the upturned glass bowl.

I've been debating with myself about whether to solder or screw my blocks together. Soldering seems more reliable, and would allow a thinner wall section, but also means not being able to get inside the block to service it. OTOH, screwing onto a layer of permatex should give a tight seal, after all it's how they put bike engines together and they (mostly) don't leak

I'm not arrogant enough to believe I will never have a leak or disaster but the risks can be significantly reduced with some care and forethought, the most important thing is logical working procedures, when putting it all together. I have an acceptable situation in my mind, and that is one major water-cooling disaster every two years, to me is acceptable to have silent Water Cooling over noisy air-cooling........... so far 1 & a half years on...... (crossed fingers), no leaks or death by water-cooling et

Seems reasonable to me, though I hope to make the disaster as minimal as I can.

bigben2k: You really can't do anything about the HSF/WB popping off the CPU: just keep it bolted. That's a design requirement.

I can't bolt it, the mobo I'm using doesn't have the bolt holes, so I have to do a clipon. What I'm thinking is that my case has a mobo tray with a cross peice already to hold the boards in place. If I extend that arm to the CPU area, I can put in some supports to take the plumbing weight and possibly put an arm with a spring on it onto the block to give it some extra support.

Some flow meters add a lot of resistance, but flow sensors usually don't, if properly selected. I can't believe that you didn't find the one that was presented recently.

If I saw that one, I must have missed it, or assumed it was a non-US available product. As I asked BR above, can you point me at a US source?

BTW, you're off your rocker about the restriction adding heat: while the heat will appear at the restriction, removing the restriction will only cause the heat to appear at another restriction(s) within the loop Just forget about the heat thing.

I think you mis-read what I said. I wasn't suggesting that a flow restriction would ADD heat! What I was saying was that if a major flow restriction or stoppage FAILURE occurred in a running system, then heat would go up in the CPU's and blocks. A temperature sensor located on or under one of the blocks would see this increase, and be able to trigger a relay to kill power or take other corrective action.

The flow sensor will protect your CPU, even if there is little flow: that's all you need. When there's no flow, then the CPU will eventually, over a certain period of time, overheat and die. The level sensor is a good idea, as long as none of the coolant is evaporating. Even better is the Conformal coating, and an allowance for the leaked water to exit the case. The real problem, is figuring out the possible leak points.

I figure where water is, there is potential for leaks so assume that any hose could split, any barb could blow off, or any block could fail. Try to design around it. The case drain is easy, got a drill? The rest takes a bit more thought.

Gooserider

[sevisehda]

I'm not sure why theres a reason to use a flow sensor. A decrease in flow results in increase in heat so just add a small thermister to the cpu block. The point is to kill the system when it gets too hot. A decrease in flow may not necessarily hurt the system too much so calibrating the cut-off point for a flow value is difficult.

For a water level sensor use the same sensor as you would to detect a leak but in reverse. Place the leak sensor in your res and when it no longer detects the leak then your water level has dropped. This could double as a leak sensor because obviously if your coolant level is dropping than that water is going somewhere.

[Solarian League]

i'm going to get an water cooling system that probably will have pelts on it so i'm obiviously new to this whole watercooling systems.

anyway i have a couple ideas.

for the leaking detector, depending on how big it is, if its small one you could place one below each barb joint. example.

look at the cpu block, say its an maze4 with both of the barb sticking out top of it, now you could place a little funnel below both barb then have the leak sensor in the neck of the funnel. so when one or the other start to leak it would trigger off the leak detector and bam, system goes down, and you could optionaly have an additional setup.

allright, you got say one pannel from a 3.5 exterial drive, just drill a few holes and put in a few lights, say one for pump, one for radator, one for cpu, one for video card, and then one for the resvor. now you have a small say 9 volt battery behide the pannel then you hook the lights up into ur auto kill system so what happens is.

when one of the leak sensor sense water, it kills the system and trip maybe a breaker, or somthing i ditto, but it trips and one of the light on the little pannel comes on and it will stay on because you have a little 9 volt battery to keep it on so the user will know why the system triped and where the leak was.


just an idea might be too complex but i don't think it is overly complex heck i might try it myself my dad and i am not pro electricans but were decent with electrical items so i might try to make that little mod myself would be a good first warning.

by that i mean, if you system just turns off you know something bad happened but you don't know WHAT happened, so.... with that little light pannel you would atleast have a idea as of what happened.

now you could probably wire up few more lights for other possiable failure, such as flow, if the flow gets too low, bam one of the light comes on, then another one is you can mount an mirco switch on your block, if it lifts off bam another light comes on and system is shutdown, aka an relay that is just blowen the second the block lifts off the dice. then another one could be for the resvor, you could have a little resvor level sensor when it goes below a certain level bam another light comes on and it could optionaly kill the system but then again you already have leak detector all over the system at the leak point.

now i don't know how much it would cost but i would guess some cheap LED lights, a few of em, and some wiring, probably a rechargable battery, and probably a few relay/switch whatever however u plan on tripping the system, then some sensor the total cost should be i hope maybe $50 or less, depending on what you want to do with it i guess.

its just an idea, and admitly i like the idea myself, i'm newbie to watercooling and this idea might help me push my watercooling system to.

[BladeRunner]

I think that maybe you are all going overboard a little. It's almost as if you are bolting another four engines, an extra set of wings and sticking a few more pilots in a Jumbo Jet just in case some parts fail. Trouble with that is you get to a point where its so overly complex you miss something simple, so the plane still crashes .

I'll have to do an air pressure test with my clipless pipes but I can assure you they will not pop off, slip off, or leak, Just think about it with the barb in a vice using all my strength and body weight all I can do is stretch the tube a bit. I used to have the whole system with barbs done in a similar fashion, and none ever leaked or came off. I much prefer the festo push fittings now because I can make smaller neater blocks and connect and disconnect them quickly & easily without worry of leaks. The tube isn't an issue once you have some of the correct stuff, I think it was about £15 for 20 meters of the 8mm

The liquid level sensor as said I haven't done yet but I'll be using a sealed optical type liquid sensor I have here atm.

Part of the reason for soldering my blocks is with the more complex shapes of the VGA ram blocks and Splitter-X in my system, it would be very difficult to make a seal grove without access to CNC, and like you say the blocks can be thinner walled and smaller overall if soldered.

It doesn't really matter what target you set yourself but I think you have to be realistic when using water cooling and that's why I say "major disaster every two years". I would probably double that to every four years if I just built the system and then didn't dick around with it at all, but as I'm constantly modding it and upgrading parts of it adding new blocks etc, the possibility of a screw up on my part is much higher. It wont change the world but if you beat your target you'll get a nice warm feeling about it too.

[Solarian League]

you might think it is overkill but heh if its what $50 to say $100 to do a little bit of disaster prevation its well worth it like that funnel and the water sensor with little lightboard on front of computer heh

atleast if something goes to hell you will know what it probably was before opening the case

Originally posted by BladeRunner
I think that maybe you are all going overboard a little. It's almost as if you are bolting another four engines, an extra set of wings and sticking a few more pilots in a Jumbo Jet just in case some parts fail. Trouble with that is you get to a point where its so overly complex you miss something simple, so the plane still crashes .

heeeeeey what's wrong with that if i was ever an jumbo jet designer i would damn well make sure i have 4 wings, 10 engines, 22 pilot and extra gas, extra fins, extra windows, extra seats extra EVERYTHING heh

[BladeRunner]

Well there you go, your jumbo jet would never crash as it would never get off the ground Looks as if you've made my point for me..

[bigben2k]

Originally posted by Gooserider
...
I figure where water is, there is potential for leaks so assume that any hose could split, any barb could blow off, or any block could fail. Try to design around it. The case drain is easy, got a drill? The rest takes a bit more thought.
...

No, a hose split is not reasonable, neither is a blow off. To make sure it doesn't happen you use hose clamps and forget about it. That's what I meant by "good design".

What has happened historically, is that a brass barb in an Aluminium top would corrode, and break a seal: just don't do it. Hose split has happened once, when the water was allowed to get so hot, that it melted the tubing, so don't allow that to happen either, that's all. The same heat can melt a plastic top, and cause an o-ring failure: I've seen that twice. They were the result of either a lack of OCP, or OCP failure.

[bigben2k]

Here's the latest flow switch that was found:
http://forums.procooling.com/vbb/sho...=&postid=79336

[Pyrotechnic]

I'm not sure if a leak detector would work 100%

There are lots of points in a system for a leak, and even a few drops on your memory, video, mobo..etc while the system is running is BAD. The only time i see where the leak detector would be sucessful is if your having a flood in the case, like a hose came off (impossible when using hose clamps). In the event of that, i think by the time the detector reacts, your system is would be dead anyways. Unless you put about 50 leak detector senors in there, i dont think it will be 100% effective.

Infact, where there needs to be almost 0 risk to a system, high quality air cooling is your best bet. Set up your computer to shutdown when the tempurature reaches a certain point, and to shutdown when the fan stops. All you have to do is maintain the fan by keeping dust from building up.

If you still want watercooling, consider using a separate case, like a external SCSI drive enclosure to house the radiator and pump, and run to lines to the computer, this minimizes the amount of places leaks could occur within the case, you only have the waterblock, and a properly sealed block should be no problem. Of course, setup the temperature and fan shutdowns, and that should be pretty safe. You could even put a few leak dectors on the block, which would be semi effective since this is the only place for leaks within the computer.