The MO-RA3 family of external radiators is a product of Germany's Watercool. It is available in 360mm and 420mm models. The former natively supports nine 120mm fans; the latter natively supports nine 140mm fans (18 on the Pro, which supports fans mounted on both front and back). At a mere 9 fins per inch, it does not require high-pressure fans to maximize performance, relying instead on sheer size.
Other large external radiators include the Alphacool NexXxoS and the Phobya G-Changer Xtreme NOVA.
The radiator has no power-consuming elements itself. It provides 6 G¼ ports, of which 4 come plugged. There are three on the bottom, and three on the top.
|Model||Weight empty (kg)||Dimensions (mm)||Capacity||Pipe rows||Cu (m)|
An Arctic P-14 PWM PST RGB 140mm fan requires a maximum of 0.17A for the fan and 0.20A for the LEDs (all at 12V, so 2.04W and 2.4W respectively, 4.44W total). They provide daisy chaining:
|P-14s||Fan watts||LED watts||Total|
The Noctua NF-A14 is 0.13A (fan only), 1.56W. There is no daisy chaining.
Pairing 9x P-14s on the front with 9x NF-A14s on the back yields a maximum 54W of 12V power draw, exactly 4.5A. Coincidentally, this is the exact amount of 12V power provided by the 3 1.5A 12V pins of a SATA power connector.
It would be possible to locally provide power via a 12V AC adapter. 12V/5A adapters can be easily acquired, and would be sufficient to deliver the necessary 54W (5A at 12V is 60W). But such an adapter is likely to be significantly less efficient than a high-grade PSU. In my workstation, I've got an EVGA Supernova T2 Titanium power supply, which will be far more efficient than any dinky AC adapter.
Desiring to source the most efficient power possible, I'll run a lengthy PCIe 6-pin cable out the back of my machine, capable of carrying at least 75W on its three 12V pins (6.24A). I'll then use a trivial adapter to take that to SATA or Molex, and thus be able to power nine PF-14s on the front, and nine NF-A14s on the back. I do not intend to put any pumps on the MO-RA3, as I already have three D5s in my workstation. My longest PCIe cables are 75cm long, with a 30cm PCIe-to-Molex adapter. This means I can't use the MO-RA3 more than about a meter away from my workstation. If I wanted to run it three meters away, I'd need to use external power (and cut longer tubes).
I'll need to run two tubes out to the MO-RA3, but beyond coolant and power, what is necessary? The RGB and PWM of fans must be managed. If I have any temperature or flow sensors on the MO-RA3, they'll need hookups. It ought be possible, however, to control all this from a device local to the MO-RA3. Many people use Aquacomputer products such as the Splitty in this role, and run data over a USB or Aquabus cable. I'd instead like to use a SBC with 2.4GHz 802.11n. I will likely break the shield off a CODI6 to get the PWM+RGB headers.
There are some subtleties here.
The Arctic P-14 blundered IMHO with its daisy chaining. For both DRGB and PWM, you have:
fan -> wire -> female -> wire -> male
but every PWM fan hookup ever made, and every DRGB hookup ever made, offers male and wants female (i.e. header *pins*). Admittedly, this is a weird departure from the norm (you typically want your charged elements to be female, so they don't go stabbing charge into hands, legs, metal, etc.), but it is what it is. This means that if you want to extrude the female receptacles (which you do), it's going to have a minimum of two wires showing (the wire to the fan, and the wire to the male). Furthermore, you don't get the full useful length of the wire. This can be worked around by putting gender changers on the terminating fans' *male* connectors. Then, bring other fans' *male* connectors out, hooking up to the (unadapted) *female* receptacles in cascade, i.e.:
- fan1 -> wire -> female1 -> wire -> male1 plugs female2 -> wire -> male2 plugs gender adapter plugs source
- fan2 -> wire -> female2
- female1 is unused
the "gender adapter" here is just a 4-pin jumper.