flow pattern I'm planning on setting up my second watercooling system (my main pc is going to get a GPU & NB block added), my old system is being put into my older pc, and I'm thinking about flow pattern Right now, I'm going to use these parts: Danger Den RBX Danger Den Maze 4GPU Unknown chipset block (Danger Den Z Chip?) Heater core (big SOB, dual, maybe triple 120mm fans) and most likely a Hydor L30 with an Innovatek Tank-O-Matic my question is, whats the best flow setup? I was thinking pump to rad to RBX inlet, then from each outlet to GPU & NB blocks, then bringing flow back together before the res.... like this (shows off awesome paint skills) also can anyone reccommend a decent chipset block for NF2 Ultra-400?
Don't split the circuit. Differences in lenght, constrictions etc. will cause differences in flow which come back to bite you on the nose when the two channels merge again. Also, a split will halve the flow for each channel. If you really want togo that way however (I mean, free will and all that), don't merge the two channels but make them exit in the res seperately. It will act as a buffer in merging two different flow speeds together.
The flow in Coolmiesters shiney new rig is splitt from a WW into the GFX and NB and I don't think he's having any real trouble. I've also had it recommended to me on here recently - I'm torn as how to proceed now too... Nexxo - How's it going?
The advantage of splitting the flow is that it reduces the overall restriction of the loop, thus increasing the overall flow rate. As such, each of the two secondary blocks will get more than half of the original flow while the cpu block and the rad will get slightly more flow. Now the difference may not be huge, but there probably will be a difference, particularly if using a pump such as the L30, which doesn't have a lot of head, and as such will benefit more than an mcp600 for example with the outlined configuration as opposed to an all inline configuration. 8-ball
Also, as nexxo has mentioned, the IDEAL approach is to merge the two lines into the res. Unfortunately, this would require some modification of your tank-o-matic, but it is the best option. 8-ball
8-Ball - So if I was using a multi ported bay res and an eheim 1250 this type of setup would work nicely? P.s. How are you? Good to see you're still around on the boards. cderalow - sorry for the semi thread hi-jack!
Cheers mate. I've been really busy with work at uni, and will be till around june, but then I've finished, and I'm having a gap year. I'll also be buying a milling machine, so I can start making waterblocks and the like. If all goes will, I'm gonna try and put a range of products together and start selling them. Good to be back though. It's strange really, as forums tend to come and go as far as I'm concerned, yet Bit-tech is the only one I keep coming back to. No-one takes things too seriously (it is a hobby after all!) and the community spirit is very strong. Anyway, back on topic. The 1250 has almost identical characteristics to the L30 mentioned above, so the same logic applies. The flatter the pumps p-q curve (in other words, pumps with high free flow rating but low head) the greater the benefit of reducing the restriction to flow. I wouldn't recommend using something like the 1250 in a serial configuration with lots of blocks, as this tends to be a very restrictive configuration. Waterblocks tend to be more flow sensitive than radiators, hence if I were setting up a rig with a dual outlet cpu block and also wanted to cool the gpu and northbridge, I would favour a config with two parallel channels between the cpu block and the res. In each of the two parallel paths, I would have either the nb or gpu block, and then a single 120mm rad, such as a BIX or thermochill 120.1. This would be an alternative to running a dual 120mm rad inline with a everything else. As mentioned, the flow through each parallel path would be half that going through the cpu block, but since the overall restriction has been reduced, the overall flow rate will have increased. As a result, the flow through each rad will be slightly more than half what it would be in a serial configuration. Equally, since rads are less sensitive than waterblocks to changes in flow rates, it is likely that the benefit in performance at the cpu block, particularly if using a flow sensitive block such as the white water, will outweigh the loss in performance at the rads due to lower flow rates. Does this make sense? That said, I haven't proven this theory with actual results, but I'm hoping that I will be able to test some of these ideas during my gap year. 8-ball
Glad you're well mate! I look forward to seeing what the blocks are like you turn out, if fact I'd say you can put me down for one now once you're past the prototype stage. Anyway. I have a BIX2 (i got it cheap secondhand) where would you put that in such a loop as discussed above?
Since it is preferred to have the rad before the pump, I would say put your BIX2 between the pump and your cpu block, then split to the nb and gpu blocks then run the two paths into the res. Hope that makes sense. 8-ball PS, I wouldn't get too excited about any blocks I make this early. It'll probably be a while off. But I'll post up as soon as I get anywhere.
Hey JetsetJimbo, how's chills? Things here are going, but slowly. I have reached a stage where I need to wait for some parts to return for Metaversa, and that is always frustrating. Nearly there though... back to topic: whatever you do cderalow, keep the circuit short and plan carefully...
