Archive for the ‘How To & Misc’ Category

Ahh the joy of mixing metals in a closed water loop…:)  While many water coolers have had excellent success with running copper/brass/nickel over the years with plain water, we have seen many examples of where certain conditions result in not so favorable results.  While we often call copper/brass loops a “Similar” metals loop, I think we are also forgetting that “Similar” is not the “Same” and we have a LOT more than just copper and brass in our loops.  Your typical water cooling loop has a mixture of copper, brass, nickel, and tin.  Note, that I think I’m the first example of the “TIN” corrosion by an unintentional experiment I had been carrying out over the last year or so..:)

The manufacturers pretty much all say “use our coolant” which includes corrosion inhibitors, yet we persist in thinking nothing is wrong with this mixing of “Similar” metals.  I’m not a corrosion expert by any means and have typically had the same or similar good success without the use of inhibitors.  I am however becoming more of a believer of corrosion potential as these repeated problems persist and as I have now experienced a recent loss myself.

Last year when doing my fan testing series, I filled up two radiators with water as part of my testing rig templates.  One was a Swiftech MCR120, and one was a Hardware Labs SR1 140.  Upon digging those radiators out in preparation for my radiator testing bench rebuild, I noticed that the MCR was still full of water, but the SR1 140 was empty.  I also noticed what appeared to be water stains on the bottom of the SR1.

Could this be corrosion?  I thought..

Oh my, my SR1 has become a victim of corrosion!!

But I thought the idea was if you run copper/brass loops, corrosion wasn’t possible?

Well…it is..

My SR1 is now a leaking sieve, so I decided to do a little digging in on galvanic corrosion.  I think most people including myself have been thinking about corrosion between copper/brass/nickel, but I don’t think we have been thinking about the solder in radiators.

What is Galvanic Corrosion?

Per Wiki:

Galvanic corrosion is an electrochemical process in which one metalcorrodes preferentially to another when both metals are in electrical contact and immersed in an electrolyte. The same galvanic reaction is exploited in primary batteries to generate a voltage.

So it’s not all bad..after all Galvanic corrosion is what starts your car in the morning..:)

What is needed for Galvanic Corrosion?

Per the corrosiondoctors.org:

  1. Electrochemically dissimilar metals must be present

  2. These metals must be in electrical contact, and

  3. The metals must be exposed to an electrolyte

Of particular interest to me is #2, I didn’t realize that the metals had to be in electrical contact, but that does explain a few things I’ve been seeing.

Now to make sense of my SR1 loss:

Ok, so in a closed loop of water, while water is initially non-conductive, it only takes a short time in a water loop to become contaminated and conductive.  Once conductive it now satisfies the “Electrolyte” criteria. #3 is done.  The stagnant condition (and not at all typical) probably made this many times amplified.

Also the metals must be in electrical contact.  In my radiator example the soldered connection of the radiator fins is clearly a good metal contact. #2 is satisfied.

And finally they must be dissimilar metals:

According to the corrosiondoctors.org anodic index, it appears a normal copper solder radiator has about a .30 galvanic potential.

Yep, seems to make sense I guess.  How about a few other examples others have shared:

ALUMINUM/COPPPER

Aluminum and Copper in direct contact.  Pn0yb0i gave an example of what running an aluminum/copper block can do after an extremely long 4 year run here.  He was going to reuse the block after cleaning, so I ask him if I could use some of his pictures if I sent him a replacement block sample for free.  He accepted happily and I feel better that he’s got a new block too..:)  Anyhow, here are a couple of photos that he shared and gave me permission to use.

He said he used distilled water plus pentosin and purged every 2 months.

And to compare the anodic index between copper and aluminum.

In general most manufactures have given up on attempting to make aluminum/copper blocks, which has led to eliminating that problem.

However, as water cooling has become “Art” as much as it is performance, there has been a dramatic increase in Nickel plating of blocks.  It is handy not having to deal with tarnished copper and a lot of people like the shiny surface of Nickel plating which in itself has caused problems as well…

Nickel Plating

The hot topic in the forums has been in regard to nickel plating failures.  Manufacturing processes have been improving regarding the plating quality.  Electroless plating is the latest preferred method which is supposed to plate the parts more evenly.

If you look at the anodic index again and compare nickel to copper, you can see it is actually very similar in index meaning their corrosion potential is very small.  In the case of metal corrosion, opposites attract and nickel/copper are very similar.

But….the difference is still there.

We have seen failures on blocks and we have seen failures on fittings.  The one commonality I have seen in all the various forum examples is “stagnant” water.  Just like my radiator example, where you normally see plating fail, is where water sits still.  I believe this stagnant condition is what promotes the #3 electrolyte condition.  The longer the water sits still near metals the more contaminated and electrolyte like it gets.  We typically see plating failures between surfaces such as the GPU block and acrylic or delrin top.  We also see it between CPU nozzle plates and the CPU block bases plated in nickel. In fittings we see the plating failures at the threads…again where the water is stagnant.

The other reason I think the small index still causes problems is simply due to the plating being very thin it just doesn’t take much to show.  Also since copper is the anode to nickel, it works in an undermining process where the copper goes away, and the nickel flakes.  Also as the copper goes away and undermines the nickel, it creates a pocket where that electrolyte enhancement (stagnant water) grows even faster.

I do think a “Perfect Plating Job” could avoid the issue with plastic tops, if there was a perfect nickel plating over the copper block and that was the only metal in direct contact, you will have essentially removed the “electrolyte” variable.  If there is no way for the electrolyte (water) to get between the copper and nickel, then life is peachy.  I just don’t think plating is ever perfect.  Any little microscopic pin hole, scratch, or thread wearing into the plate will expose the copper allowing the reaction to occur.

What is the problem?

  • We are mixing metals.
  • Some of the mixed metals have direct electrical contact.
  • Our water is becoming an electrolyte with stagnant water conditions in some areas.

Sacrificial Anodes

One thing that hasn’t really been explored much in water cooling is the use of sacrificial anodes.  These are used quite regularly for corrosion applications where the idea is to make the electrolytes go after a more active metal instead of the metals you are trying to protect.  The anode need to be in electrical contact with the other metals and will over time corrode and need replacement.  You see them in household water heaters and on ships in saltwater, and bridges along the coast.  Most industries that have some sort of corrosion problem lean toward either or a corrosion inhibitor or some sort of anode to provide that protection.

I don’t see why you couldn’t have some sort of zinc barb insert or something that could be easily replaced though. I’m not quite sure what sort of deposits the zinc would make, but it should theoretically work in preventing corrosion from occurring. I’m not sure???

I could see that as being a possible solution for folks that would rather not run anything than water. That’s what we do for water heaters (inhibitors not possible), why not for water cooling?

Anyhow, not sure if sacrificial anodes would work or not, but I’m really curious to try. It could be a solution for giving plain water loops corrosion protection without the fuss of a coolant with inhibitors.  You would just need to attach a piece of zinc to each of the mixed metals blocks/rads and see what happens.

Conclusion

I don’t think it is possible to completely stop galvanic corrosion from occurring, but we can reduce it by:

  • Eliminating direct electrical contact of dissimilar metals (Plastic top/unplated copper base blocks)
  • Reduce Electrolytic Conditions – Reduce areas where water is stagnant, flow is your friend.  Regular maintenance and complete cleaning of the block/pieces probably helps too.
  • Improve plating processes and increase plating thicknesses.
  • Slow the process with corrosion inhibitors in the fluid
  • Slow the process using a sacrificial anode in system running plain water.
But I don’t think you will completely stop corrosion.  The idea is to keep it at bay long enough and/or reduce it for the intended service life. Unless you made the entire loop of one metal or kept all the metal parts from touching one another, you will have the potential for galvanic corrosion to occur.

Cheers!

Martin

XSPC H2 Tower Case Preview

Posted: June 10, 2011 in How To & Misc
Tags: , , , ,

Welcome to one of my first water cooling case previews, the XSPC H2 Tower.  There are a plethora of cases out on the market, yet very few include enough space to really build a very high performance water cooling system (large & quiet radiator/fan combinations).  With water cooling your heat exchanger (Radiator) is user defined or case constrained.  For users willing to have external radiators, you can have nearly unlimited capacity, but users that wan’t a clean and tidy internal system are constrained by the case space.

Radiator size needs are defined by three things:

1) Heat Dissipated – CPU only loop vs multiple CPU/GPU blocks makes a big difference.

2) Fan Speeds – High Speed fans can make a radiator dissipate up to 5X more heat than very low speed fans…BUT!  equal noise gains too.  If you want complete silence, you need to mitigate the fan difference with more radiator.

3) Performance Level – An average system may be happy with cooling performance at a 10 degree delta between water temperature vs. ambient, but an extreme performance level is possible down to 2-3C levels.  While ambient is our theoretical wall, there is room for improvement in radiator capacity.

Bottom line, water does offer the option to address all three, but they all mean more radiator and there simply isn’t enough room in your average air cooling case to get all that and retain an internal design.

I’ve been using a tech station for a couple of years, and while it does meet my ease of use needs (Two quad rads), I’ve grown tired of the exposed/dusty conditions.  I’ve been looking for a means to go back to a computer case, but I also didn’t want to compromise having the ability to use large quad radiators with low speed fans.  I had considered some of the “Cube” style cases, but I preferred something that was more traditional “Tower” in shape and simply large enough to house the quads.  Upon requesting a test sample XSPC Rasa CPU block, I was presented with an opportunity to review this XSPC H2 Tower Case which I thought fit my needs perfectly.

I would like to thank Paul from XSPC for sponsoring this review and my future build!

This initial review I am calling a preview because I haven’t and can’t fill it up until after completing my round of CPU block testing.  I’m planning a rebuild and relocation of my hardware to this massive tower case….the H2! First a picture of the main reason behind my desire to move to this case…factory quad radiator support…no mods needed!

4 x 120mm x 15mm spacing fits like a glove!

SPECIFICATIONS

First let’s take a look at the specifications. The size is very large, in person it’s actually even bigger than expected.

I will note that the bay covers are sold separately.  Unless you’ve got 8X bays worth of gear to fill that up, don’t forget to order the covers. This wasn’t apparent to me until the end of my assembly, but it is spelled out in the specs.

It is also worth noting that the case material is “Anodized” aluminum, so it is fairly light in weight and the anodizing is more durable than paint.  The color is more of a matte black with the brushed finish, so it’s not overly glossy and I think makes a great neutral dark theme for most builds.

Dimensions:
618.9 x 246.6 x 696.8mm (DxWxH)

Shipping Dimensions:
67 x 67 x 20 cm, GW 12.296Kg

Material:
Brushed aluminium – Black anodized

Screw size:
6-32 UNC

Features:
8x 5.25“ bays *
3x HDD bays (6x with optional extra HDD cage.)
1x SSD tray (up to 6x with optional extra trays)
6x 120mm fan grills
– 1x Quad 120mm
– 2x Single 120mm
1x Acrylic window
*Front 5.25“ bay covers sold separately

While the specs don’t mention it, factory it comes with the top  4 x 120mm radiator spot and you can buy brackets to fit a factory 3×120 or 3×140 sized radiator for the bottom.  I sized it up myself and very certain with very minor modifications, you could also easily fit a quad radiator in the base as you can see below:

You could probably even run a quad on one side and a triple on the other adjacent to the PSU, so the potential for very large radiator capacity is definitely there which is what I was after. Let’s have a look at the packaging and shipping protection first…

While generally most testing is only good relative to the same test bench, it’s also nice if you can get some level of absolute accuracy for a sense of scale and to speak more of a common unit of measure. I do this testing all for hobby, so none of my meters are sent in for any sort of monthly calibration or certification.  While I thoroughly enjoy my little testing tools, they are not proven accurate and I simply don’t have a means or desire to do so while hobby testing. It’s very expensive and cost prohibitive at the recreational level I’m working within.

I figured the next best thing is to run my own checks using basic tools I have readily available.  Those basic tools include a  graduated container, a stop watch, and section of tubing to be used as a static pressure head water filled tube manometer.  In spirit of sharing information, I wanted to include this in a mini “how to testing” blog of sorts.

Some of these methods can also be used for casual testing.  If you think you might have a flow problem, but don’t have a meter…you can test your flow rate fairly accurately by the method described below.  Also if you think your pump is not up to par, you can (with a fair amount of work) also test pumps using tube manometers and bucket/stopwatch for flow rate.  I’ve done this before…long ago, but it’s very much a plausible method to extract a more scientific level of data or to test the accuracy of meters.

Testing Flow Meter Accuracy

Testing flow rate (Gallons Per Minute) is fairly simple.  Find a nice large known volume to fill and measure the time required to fill this known volume.  In my case I had an old 5 gallon race gas can that had gallon increments.  Mixing 2 stroke gas requires some level of ratio precision, so these cans are a bit more graduated than a normal gas can.  If such a graduated volume isn’t available, you could also create your own by using a smaller known volume and marking the container as you fill it up with the smaller known volume container. (A large bucket and an old empty one gallon milk jug would be one cheap and readily available option).

Moving onto the next variable…Time in Minutes. I took my DROID phone (Not the first time I’ve used it for science!), downloaded a stopwatch app, and proceeded to test in gallon increments.  I measured the time in minutes and seconds to fill the container from zero to the 5 gallon mark.  Here is the test setup and results:

Per the King Instruments Site, the larger King Instruments 7520 should read within 2% of full scale or .02X5 = 0.10 GPM.    I got around 1.7% of the measured scale or .06GPM maximum.  Flow rate calibration appears to be within specification…very good!

Testing Pressure Meter (Manometer) Accuracy

Testing lower levels of pressure is also fairly simple.  A manometer is nothing more than a pressure meter that measures the pressure  difference between two points.  If you disconnect the negative pressure terminal, it is now exposed to atmospheric pressure.  When you zero out to the atmosphere, you now have gauge pressure.  If you fill a column of water on the positive terminal, the net column height (Max level minus level in terminal tube) is the actual gauge pressure differential measured.   The tricky part is finding something plumb that you can measure a column of water.  Note that the tubing doesn’t need to be straight, but your measurement from water level at the top to lower water level needs to be plumb vertical.  Excuse the spider webs…:)  I just connected a funnel and clamped it to a column in our vaulted ceiling area.

Per the Dwyer Site, the Series 477-5 is capable of reading .5% of full-scale.  Full scale is 30.00PSI or 831 inches h2O.  So the possible error is .5% of 831 or 4.1 inches of water.  I measured an error of 1.2% of 100 inches or about 1.2 inches which is well within the 4.1 inch specification.  Looks like my manometer is also reading just fine within specs..:)

Manometer Pressure T fittings

In manometer testing, I think there is also some importance in the T line fitting used.  If it’s not a large smooth pass through fitting or the fittings are not perfectly identical, there could be some induced pressure differential from T fittings.  Like the venturi in a carburetor, if you reduce the sectional area, you will create a vacuum.  There is also the possibility with a large opening for the pressure junction that you could get odd flow momentum effects such similar to a pitot tube. I have seen some odd things in the past that makes me a bit careful about these fittings.  I think it’s good practice to spend some time on these fittings to minimize disturbance of flow and ensure equal pressure measurement conditions.  Water flow does change states and behavior as it increases in velocity, so keeping it down to nice slow speeds may help minimize measurement oddities.

I created my own T fittings from 1/2″ copper tubing and brass fittings soldered.  This leaves a clean and thin walled 1/2″ ID fitting with minimal turbulence and restriction.  They also have very small port openings leading to the manometer in an effort to minimize both restriction and any pitot tube like effects. I have used these custom manometer fittings for some time now and have been pleased with repeatability of pressure drop or pump tests.

Conclusion

While absolute accuracy is not really important at all in comparison related tests, it’s nice to know that my meters are within at least manufacturer specification.  That’s one way to do some checking on your flow and pressure meters if you’re interested and I’m pretty happy now that I finally checked mine.  There is also not much information on this, but I think it’s worth your time building your own T fittings when doing manometer related measurements.  These test methods can also be used by other reviewers that don’t have a manometer or flow meter.  I think it’s always good to do some of the more down to earth type tests like this now and then.  A number displayed on a magic box is worthless without the understanding of what it’s actually doing.  While time consuming, these forms of pressure and flow rate testing are also very educational.  I think they should be mandatory before anyone starts testing with a flow meter or digital manometer….:)

Looks OK from what I can tell..time for more pump testing..

Cheers!
Martin

Catching up from a long overdue update.  I’ve added and cleaned up many pump/component updates as well as incorporating a new parallel GPU block feature. Give it a try and let me know how it works for you.

You can find out more and download it via my optimizer page here.

Cheers!

Martin

Test

This is just a test…