2 MCP-350s, "quasi" dual loop?

[Ross]

Well, I received my 2x "Swiftech" MCP-350s (Laing DDC/Delphi) on Friday, unfortunately my 775 WB and new rad won't all be here til Tues. FYI, I wanted to try 2 of these as a replacement for my current MCP-600 pump since they are smaller, "square" (easier fitting/better looking mounted in the case?) and more quiet...even two of them.

The way I had things figured out was to run them in parallel, which no one seems to ever recommend doing with pumps. I was glad to see that Cathar added some data on them to this thread (http://forums.procooling.com/vbb/showthread.php?t=10825&page=8&pp=25) and I was most particularly interested in this post from the man himself:

A second DDC may well be a good idea, and given its characteristics it would generally be desirable to put them in parallel as you suggest. In-series would suit the incredibly high flow-resistance blocks (i.e. even the NexXxoS XP will only be borderling restrictive enough from a cursory examination of the curves to warrant an in-series setup).

In parallel at 12v, we are getting a curve that looks much like the MCP600 @ 13.8v, with heat dump being a little higher, but also slightly higher head. Haven't worked it out, but from a cursory glance it looks like using the MCP600@13.8v figures would be a very close estimate.

In parallel @ 13.2v, which is very easy to achieve even with a small DC-DC converter which can be bought for ~$30US or so, and the putting them in parallel looks something like the 2 x MCP600 @ 12v in series curve, but with more heat...


So parallel seems like a fairly good deal in this case in terms of performance which pretty much cinches it for me. What I will run in this setup:
2x MCP350s
MCW-6000 CPU block
MCW-50 GPU block
MCW-20 Chipset block (still debating if I am going to put this back in the loop)
Swiftech res
BIP2 2x120mm rad. I currently have a Watercool HTF2 Dual, but it so thick (~2.25") and I am only running the 2x120s on it due to space constraints/noise issues, so I ordered the thinner rad operating under the assumption it will make better use of the air flow that I have AND free up some space in the case (CM ATC-111).

And one crazy idea (sleep deprivation induced I am sure!), probably not even worth the work/hassle, but I do have a lot of stuff laying around and was thinking of trying to squeeze in my BIM2. If it's dumb, just say so :) No comments on my cool diagram tho :beer: ...just thought it would definitely aid in explaining it!

I was wondering if there would be any benefit/harm to running the CPU/one pump on the BIP2 and the GPU (+ possibly NB) on the BIM2/2nd pump. They wouldn't quite be 2 separate loops since both would be returned to the separate ports on the Swiftech res. Note red/blue != hot/cold, but the separate loops to each rad/pump.

http://home.austin.rr.com/rossmisc/images/crazyloops.gif

Basically one wye coming off the res out to feed both pumps and the 2 returns to the res being the 2nd "wye" themselves. I would think one pump would be enough for the CPU/BIP2 and the 2nd for the GPU/NB/BIM2, but with them mixed together like this at the res, I just dunno. Maybe the rads would need to be parallel to reduce restriction? I guess if that were the case, I wouldn't bother at all...just happen to a have BIM2 sitting here and started thinking. What do you think of this plumbing nightmare? Not worth the hassle? Sure would be interesting and something to look at :D

I plan to start off with a "normal" series WB loop, the pumps in parallel and just the BIP2, but if I get bored one day, who knows. Worth a shot or a total waste of time?

Some pics of the two Swiftech pumps (note the big "Delphi" cast on them):
http://home.austin.rr.com/rossmisc/images/mcp-350.jpg
http://home.austin.rr.com/rossmisc/images/mcp-3502.jpg

[crimedog]

incase anyone else is interested in this pump, here's a link http://www.sidewindercomputers.com/swmcpu.html
quite an amazing pump! wish it was offered with 1/2" od barbs and cost a lot less :p

i don't think there's a point of having two seperate loops (unless it makes routing easier) if you're sharing a small resevoir, the water from the two loops should go to the same temperature. i'd just run the pumps in series myself, might even get higher flow rates.

[Ross]

Yeah, I got mine from sidewinder. DD had them listed before anyone I think, but I wasn't sure if they were in stock or not. Hehe, if they cost 1/2 the price, it'd be great, but one would probably be sufficient for just a CPU loop. Of course you could have the 600/650 for the same price, so I suppose it would be a preference of fitment/noise over flow for which one you spend ~$80 on.

i don't think there's a point of having two seperate loops (unless it makes routing easier) if you're sharing a small resevoir, the water from the two loops should go to the same temperature. i'd just run the pumps in series myself, might even get higher flow rates.
I figured as much, but thought it might be a cool idea. I don't think the plumbing would be difficult (unless the rads were parallel), but there certainly would be more of it. As for parallel/series on the pumps, I guess I can/will try both and see if there is any difference at all in temp. Unfortunately I don't have any test equipment to see actual flow/pressure, so the only difference I would be able to note is actual change in CPU temps...

[Voodoo Rufus]

You're running the pumps in parallel correct?

Be interesting to see if series or parallel would work better for your loop since you have 2 rads and 3 blocks.....

Why do pumps have to be so expensive these days!!!

[Ross]

I am planning on running them parallel, but the loop(s) I posted above were just an idea that I had, not really the plan. I just wanted to see if anyone thought it might be an interesting idea or not.

I plan to run the one BIP2, probably all 3 blocks (CPU/GPU for sure) and both pumps, which I will try both in series and parallel to see there if any measureable temp difference. Unless series makes a big diff, I will probably keep them parallel since the plumbing would be a little "neater" the way I have it figured.

Guess I'll just start putting stuff together on Tues. and see what I end up with ;)

[Cathar]

Why do pumps have to be so expensive these days!!!

More like how come pumps are so cheap in the USA compared to the rest of the world...

You guys don't know how lucky you've got it.

Pumps bought in the USA cost one half to one third as much as buying them in most any other country. One can only but wonder why...

[Voodoo Rufus]

Well when you put it that way, I don't feel so bad dumping $80 on a pump. Still a chunk of dough though. :(

[Ross]

I haven't priced pumps outside the US, so I didn't know it was that bad elsewhere.

That's quite a great thread you've got over there at Procooling, thanks so much!! I already had the 2 MCP350s when I read your post about them this morning, so I felt a little better about making the change ;) What do you think about that loop setup I have above...might work well, might work worse or not even worth the effort?

[Achilles17]

Pumps bought in the USA cost one half to one third as much as buying them in most any other country. One can only but wonder why...

Thats only cuz the USD is so amazingly bad that you guys get such a good exchange rate. If the dollar was back to where it was in Clinton's time, it might be a LITTLE cheaper than in other places, but certainly not as cheap. I mean, if say a a D4 or MCP600 costs 80 euros in the EU, thats almost $120 these days. So why not just order it from the US? If the dollar was around the same price as the euro, which is what the euro was designed to be, then it wouldnt be so rediculously cheap here.

[Alacritan]

I wouldn't use a Y-fitting for the pumps' intakes. You won't have enough flow to either pump, most likely causing them to cavitate. What you should do is run everything in series. Water temperature doesn't vary much with each pass, so don't worry about water heating up after passing over your CPU. It doesn't. Just put everything in series. The DDC (MCP350) can definitely handle it, infact, that's really where it excels. It's better in highly restrictive systems than the MCP650 or MCP600. If you had a less restrictive system (fewer blocks) the MCP650 would actually be the better pump. So definitely put everything in series. So long as the reservoir is the first and last object in the loop, cavitation shouldn't be an issue.

[Ross]

You won't have enough flow to either pump, most likely causing them to cavitate
That makes sense...I didn't think it would really be a problem though. I could very easily swap so the 2 single ports on the res are the intakes for each pump and make the return on the single port. I don't know how that would effect the step-down that is supposed to reduce cavitation/air entrapment problems. I would imagine that it would work just as well in the opposite direction with two pumps pulling water out.

http://www.swiftnets.com/assets/images/products/MCRES-525/MCRES-525-BLACK-BACKGROUND.gif

So definitely put everything in series. So long as the reservoir is the first and last object in the loop, cavitation shouldn't be an issue.

First/last in line where the pumps are concerned? They will be. If I run them in parallel, they will be as above, but with the intakes on the 2 separate ports as I just noted. If I run them in series, the probably won't be next to each other in the loop, but one between the res/rad and the other between the last WB/res.

[Ross]

Well, I finally got around to getting the 775 WB in, along with the BIP2 and *one* of the MCP350s. I wanted to get WCing on this proc ASAP, but I haven't ironed out the logistics of how to route everything...it may just be too awkward running the res.

Anywho, I wanted to report on the MCP350: DEAD SILENT. I mean the only way to tell it's on is to feel it. With the case wide open and nothing on but the PSU and pump (leak testing), I turned the adjustable PSU fan as low as it would go and heard nothing but the PSU fan. You can *barely* hear the pump with your ear 1" away...it's awesome. It does complain pretty loudly when air runs through it, but once it's bled you'll never even know it's in your case, even if you leave it open.

I am at this very moment looking at 35C idle temps @ 4.0GHz/1.45Vc (3.2 540). Load is 41-42C. I was able to hit this OC on air (120mms) waiting for my WB, but idle was 48-49C, load in the 63-65C range and ambient was cooler by 4C as well.

If don't decide on a larger case, I may just add other 2 blocks first and see what happens on the one pump....it's seems pretty stout. 2 may be overkill, but are sure to work really, really well!

[found404]

If the pump is a silent as you claim, then it makes for a verytempting upgrade, too bad i can't find it in the uk yet. Probably import it.

Thanks for the info

[Ross]

I'd videotape it or something so you can hear (or not as the case may be) for youself, but it is very, very quiet...much quieter than the 600. I am sure as more people start getting them installed/reviewed, they will back up my findings.

It looks like more vendors are starting to get them in. Places that I looked at when I ordered mine that didn't have them then, now do. Hopefully they'll be finding their way across the "pond" soon for you ;)

[Ross]

If anyone is still following this thread, see here: http://www.ocforums.com/showthread.php?t=351928