Friday, 22 May 2009

Solving Zalman Reserator 2 problems

See here for the introduction.

Sure enough, it wasn't even a month before my little corner of peace turned first into a little nuisance and after a while into a nightmare:

Problems #1 and #2 - flow indicator
First I began having problems with cricket-like sounds that the flow indicator was making. I contacted my reseller and asked them for help. Aside from a bit of mumbling I overheard ("customer sucks" style), they said they would check with Zalman and get back to me. Well, in the three weeks it took them to get back to me, I started getting the second problem - the flow indicator stopped spinning which caused the unit to shut down the pump and start beeping violently. And that's not something you want with your water cooling. If the water stops flowing, it's a matter of minutes before the water in the blocks is too hot to provide any cooling for the components.
At this point I just took the unit to the retailer and hoped that it would be fixed ASAP. Naturally, it took them more than a month to actually give up and give me a new unit.
Problems #3 and #4 - service level
Since the retailer's service personnel was so helpful, I received the new unit alone. Having used all the coolant for the previous setup, I naturally complained. So I was given some of my original solution from the first unit plus a bit of coolant from the second package. And to top it all off, they poured the stuff into tap water washed plastic flasks (originally a sweet beverage). No amount of complaining made any difference. Together with this I was also given a lecture on how the coolant is too thick and that the instructions were wrong to suggest a 1:4 mixture. This supposedly caused the original indicator failures. A 1:10 mixture was more than enough. Blah blah. Naturally I also couldn't convince them to give me a new degassing tube. So much for customer service... :(
Problem #5 - pump too weak
I tried to remove the plugs from the tubes to make myself a new degassing tube, but that's impossible without cutting the tubes themselves. Of course, having no degassing tube made degassing even harder. The pump in the Reserator unit is simply too weak and after 4 hours of trying I had to give up. The pump simply couldn't push the water down the tubes and no amount of shaking, raising / lowering of both the unit and computer helped. Calling service again had no effect. Also writing mails to Zalman support also bore no fruit. So in the end I simply decided to buy a second pump. I chose the Laing DDC-1T and it arrived in two days. The Laing pump has similar properties as the Eheim in the Reserator. While also being a quiet pump, it has a rather nasty problem of vibrating alot. Only using lots of foam around it made it quiet. Two pumps in the loop finally managed to push the water through.
Having solved the problem, I also assembled the rest of the loop with the NB block and VGA block. After the system was degassed properly, the Zalman pump alone was able to push water through the system, but I rather kept the second pump in operation since the flow indicator turned spookily slow in comparison.
Problem #6 - very high flow resistance
At this point I should mention that the entire system offers significant water flow resistance. If you wish to empty the tubing, both the computer loop subsystem as well as the Reserator unit subsystem offer such resistance that one has to blow really hard into the tubes to make the water go through. I mean *really* hard. My wife saw me a few days ago when I was making the final cleanup / repairs and she yelled at me to stop as she thought I would have a stroke :) My face was just so red from all the blowing. }:-)) This kind of resistance certainly demands a pump with much higher head pressure than the Zalman integrated pump provides.
Let's put it this way:
Initially when I assembled the original CPU only loop I calculated the water flow and it was only 27 l/h (a very generous calculation) instead of the pump rated 300 l/h. Completing the loop with three elements didn't much change this. Maybe the flow fell to 25 l/h. In any case, the flow indicator spins really slow with Zalman pump alone.
Adding the Laing pump into the loop increases the flow to approx 60 l/h (also a generous calculation).
Problem #7 - flow indicator again After a month of bliss, problems started again. The flow indicator was stopping again and I couldn't figure out why. The reseller service refused to help me any further and they directed me to Zalman support, which I already knew to be quiet.
Zalman USA
So I turned to online forums, asking my peers for help. I received some great feedback as well as phone numbers of Zalman USA office. Naturally I called them and spoke to a nice guy named Keith and he agreed for me to send him an email with description of the problem. So I did and I got a reply the next day. This was great: finally some support. We exchanged questions and answers three times all of them within a single week and Keith was kind enough to give me all the requested info. He even promised to ask the Koreans to send me a replacement pump, flow indicator, degassing tube and coolant. This was truly a pleasant experience and I was really happy with Keith. To be more precise - I still am.
Well, to be sure, Zalman support was once again quiet. After waiting for two months I finally gave up on the replacement parts and proceeded to fix the unit following instructions by Keith and the forums folks. In order to do that I had to order some coolant, but unfortunately it was unavailable in Slovenia. Yep, both our distributors sell Zalman's water cooling kit, but none of them provide replacement parts of any kind. So I ordered from Germany, paying 13€ for coolant and 17€ for delivery :P
Problem #8 - Algae buildup
Having such a weak coolant solution predictably led to algae buildup in the system. While I was discussing things with Zalman USA and waiting for replacement parts that were never to come, algae started growing in the system. Fortunately for me, the buildup was slow enough that the system wasn't clogged by the time I started fixing things. But a thin film was clearly visible on surface of the coolant solution.
Problem #9 - Screw quality
Disassembling the unit revealed a few more faults with the unit. The screws used are extremely low quality. Even with a proper screwdriver I simply ruined a few of them. They are so soft that any force will damage them. One would think they are made of alluminum, that's how soft they are. But they react to magnets so I guess they are some iron alloy after all.
Problem #10 - Flow indicator materials
The flow indicator itself is made of clear plastic and the centerpiece (the spinning gauge) some blue plastic. Here's a pic of what I pulled out:

Note the two metal weights and the plastic around them. One weight is some iron alloy with good magnetic properties so that the sensor can detect the indicator spinning. The other doesn't have magnetic properties (and is also not affected by rust).
I cleaned up the metals and re-glued them into the plastic. To prevent further corrosion, I painted the whole thing with some water resistant paint. Excess plastic was cut off.
Since at the time I still didn't know what causes the indicator to stop spinning, I performed multiple tests for spin resistance. The indicator spinned no matter how low a water flow I generated, so I left it at that and proceeded with fixing the rest of the unit.
Problem #11 - Sharp tubing turns in the unit
Here they are:

I should mention that these imaged don't do the actual situation justice. Both tubes were bent so badly that the internal tube dimensions were at most 2mm x 8mm, thus severely restricting water flow. The springs that are supposed to prevent this bending don't quite do their job.
Having bought a couple of L joints beforehand, I was able to fix this problem like this:

The bottom joint was particulary problematic since there wasn't enough space and I definitely wanted to keep at least two of the sealing rings. Not being able to cut the L joint to optimum length left me with a bit of an angle which surely creates some turbulence, but I'm still betting this solution is a lot better than the original one.
Problem #12 - Air buildup
During all the time having these problems, I've had two occasions where air would start building up in the tubing. Initially I attributed this to algae, but it also happened again even after I had already used the new coolant. I'm guessing I just didn't make a strong enough solution which led to the original algae colony not being completely killed off. I have in the mean time cleaned the system again, also by using quite a lot of alcohol and I'm now using a much stronger coolant solution (1:3).
Anyway, I have also determined that this air buildup was responsible for the flow indicator failures. While I could certainly see the air in the computer loop tubing, I could never see it in the unit tubing. However I am now sure that air buildup was in both cases present also in the unit tubing itself.

So, this is it. After all the troubles and work fixing them I now have what I originally purchased. The conclusions from the previous post are still valid and I'm also happy that I managed to fix a problem where lots of things were working against me.

I still / again believe this is a very good piece of equipment and I think it's a shame Zalman removed it from their portfolio instead of fixing it. Its thermal properties are no doubt much better than Reserator 1's which stays in the portfolio, probably due to much simpler design which lacks many of the shortcomings of this unit.

Zalman Reserator 2 (review ?)

Well, seems I can't keep my own schedules, but still it's time to start on the promised reviews.
The Zalman Reserator post will be in two parts, first about the product and then about all the issues I've been having with it. Yes, you've read this correctly - even Zalman sometimes makes mistakes. Having used their products for some years now, I thought that nearly impossible, but this particular product has quite a few shortcomings I will describe in the second article (hopefully also done today). For now I will focus on the good side since after all the modifications I made this product now does what it was intended to do, I just had to work a bit for it:

I've bought Zalman Reserator 2 last July. You see, I have my computer set up in a corner of my living room with the box under the desk and the rest of stuff on it. Since I like to squeeze every ounce of power from my computers, naturally that also means I have a 150W oven under my desk. So, off to the first online store I went and ordered me a nice watercooling setup. The general idea was of course to move the heat source from under my desk to above it.

Just a few days later, the postman rang with a nice little package:

Well, OK, not so little. The box itself was actually quite big :)

Trembling with excitement, I went ahead and opened it:
First thing inside was the radiator / reservoir unit itself. I must admit, I didn't expect to see such quality of workmanship. All corners are nicely rounded, so there's no chance of you cutting yourself. The transitions between various parts are smooth and any screws holding the whole thing together blend superbly into the design of the unit.
The paint finish is superb with silver and matte black adding a high quality touch to the overall unit impression.
Here's a snapshot the other parts in the package although I'm sure you've seen all of this in all the proper reviews around on the net. There's two stands, an expansion slot tube bracket, the CPU and VGA cooling blocks, a degassing tube, 4 meters of tubing and a flask of coolant liquid. Naturally, a manual too.
Following instructions in the manual, I could have the unit assembled in a few minutes, but I decided to enjoy the process and took a good hour to assemble the basic CPU loop, turning and marvelling at every component in the process. This was after all my most expensive needless computer component bought to date.

The degassing process didn't go as well as I expected. I spent a good half hour before I couldn't hear any bubbles any more. Today, almost a year later I know exactly what to do to degass the unit, but back then I was obviously too much of a n00b. In case you're wondering: you have to tilt the unit clockwize (observed from the front) until you hear the bubbles. Then tilt it some more and just wait until the bubbling stops. The entire procedure can be completed in a couple of minutes with a maximum of one reset necessary. My original problem was that I was tilting and shaking the unit to both sides, as was described in the manual...
Having degassed the unit, I plugged it into the simple CPU loop I made and let it run overnight to check for leaks. Naturally, there were no leaks whatsoever so I finished up and closed the computer. Here's how the original loop looked like inside my Antec P182 case:

The VGA block was of course useless for my 8800GTS so I immediately went online and ordered the appropriate block for that as well. What would be the point of buying a component to move the heat when I wouldn't even apply it to the hottest component in the computer?

A few days later I received a small box with my 8800GTS cooler:
Again, a high quality looking component, nicely presented in two brushed alluminum shades. Attached were instructions and all the required parts to assemble the beast onto my gfx card.

Having had to buy a new VGA cooler, I was left with the original block which now suddenly got the appeal of becoming a nice chipset block, even though it originally wasn't designed for that function. Oh well, if the designers only knew what folks do with the stuff they worked so hard to design }:-))

So, here's the snapshot of the final, three-block loop i assembled. The order of components is CPU - Chipset - VGA. I figured the temperature delta is the largest on the CPU so it seemed to be the best candidate for the cold water. VGA on the other hand can take temperatures higher than CPU so it doesn't matter it the water pouring into its water block is already at 70 degrees centigrade.... :D OK, OK, I admit, the order was suggested by gurus on the numerous forums I read before even going for the setup.
I did have to clip one of the mount holes on the "chipset" block in order to be able to fit it next to the CPU socket, but otherwise the installation went easy. You will also notice that I removed the heat sink from the power regulation modules on my Asus P5K-E WiFi. I figured since the top row didn't have any, the back ones also don't need it. For this reason I didn't want to cut the NB / modules heatpipe in case I'd ever need to go back to air cooling. Until today, almost a year later, there have been no adverse effects due to this decision although I must admit that initially I feared it a bit.

After completing the modifications, I went ahead and started testing the whole system. I must say I'm very impressed with the results:
The cpu (E8400 @ 3,44GHz) went from 40 / 65 (idle / load) to 40 / 55 degrees centigrade.
I never had any numbers for the NB, so I cant give any, but it works, so I guess it's cooled sufficiently.
The VGA, however, was the biggest jump: Initially it was 55 / 80 degrees. Now it's 50 / 55. No matter how much I work it out, the temperature difference won't go higher than that. And I'm running it overclocked to the max 650 (core) / 2000 (RAM).

You probably noticed the relatively high idle temperatures. Well, I can also explain that: I'm running this computer 24/7 and at this moment the room temperature is 29 degrees centigrade. Yes, we're having quite a hot week here in Slovenia and no, I have no air conditioning installed.

What I did notice was that the Reserator unit manages to cool all three components and even with the computer constantly running, the water will go no higher than 10 degrees above ambient, the CPU will have water temperature at idle and max +15 degrees when running prime over night. The graphics card always has water temperature +10 - 15 degrees, depending on load.

I even managed to push my E8400 to 4.6GHz, but that required some insane voltage - if I remember correctly, I had to use 1.65V or something like that. And it also got to 65 degrees after a good night of prime so that was a bit too high for my personal taste.
Now I just have it overclocked to 3.44GHz which is the highest it will go with the lowest voltage the MB suports (1.1V). In case you're wondering, my particular CPU overclocks a bit strangely - if I want to go even to 1.85GHz, I already need 1.35V so I just figured a measly 12% increase in performance isn't worth it.

Obviously, now I have the heat all placed above the desk with practically none of it coming from the box below. Which was my initial purpose for this little beast. And it worked :)
Also this setup is practically completely silent. The pump in the Reserator unit is practically inaudible and I only have two very low RPM fans left in the case to cool the other components. I guess the only quiter PC I ever had was my first 286 AT which had no hard drive and no fans. And even it lost the quietest status the moment I installed my first second-hand 10MB IBM double height MFM disk :)

Overall, an excellent product, unfortunately removed from market due to its many shortcomings. See the next post, I'm guessing I had to solve pretty much all of them to make the unit work properly.

Another snapshot of it installed in my corner of zen and peace :) :