Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 38,850
Have you thought about not using a radiator at all? Normally, it wouldn't make sense to simply heat up a tank of water and then let it cool during off hours. In the context of building out cooling infrastructure which includes over 70ft of tubing, however, it might.
With a 1200W continuous load, it takes 30min to heat 5gal of water from 23.9C (75f) to 50C. With a 60gal fish tank and a single 3090 gaming load (assuming 600W of GPU + CPU heat), you're looking at 11-12hrs of non-stop gaming to heat that fish tank to 50C. You could then let it cool by itself overnight.
If you do want to stick with your remote rad setup, just run a water-to-water heat exchanger next to the PC. Run your favorite coolant in the PC side of the loop and then run anything you want (even biocided-tap water) for the remote side. You won't have to worry about special tubing or coolant this way.
I haven't fully decided on the heatsink side yet. I had thought of a few different options.
- (Cheapest) A couple of large MORA style radiator assemblies with fans on full blast, out of ear-shot.
- (Riskiest) Homemade PVC tube cooling tower venting outside through a dryer vent (like the old school "bong" coolers from 20 years ago)
- (Most expensive) Water chiller(s)
The last option is really appealing to me due to performance. It would be nice to keep the loop temp at ambient (or maybe even between the dewpoint and ambient somewhere, if I can control it tightly enough to avoid condensation)
The challenge here is that compressors get damaged if they have to be turned on and off all the time, which is why most of them have a ~3min timer that won't allow them to kick on again until 3 minutes have passed after last turning off.
So, I'd need enough coolant in the loop that at max heat output from my system, the loop temp won't change more than my desired temperature level in 3 minutes. I'd probably set that temp change to ~1°C
I have a 1200W PSU, but I rarely if ever max it out, and even if I do, not all of that load is going into the water loop. I think a likely maximum heat load into the loop between the CPU and GPU is probably about 850w.
Assuming my coolant Cp is the same as water, I'd need about 7 gallons to keep temperature increase to 1°C in 3 minutes.
My early stage plan is to use a 6 gallon glass carboy as a reservoir. I'd either build a custom enclosure/stand and insulate it, or use an off the shelf home brewing insulating wrap and carboy stand. One thought was to maybe place it in a medium sized trash barrel, and fill with expanding insulating foam around it.
The last gallon would have to come from the fluid in the tubing and loops, but with the length of tubing I am considering I don't think that would be a problem. If my line length is 120ft in total (60 ft each way) and I choose a 0.5" ID tubing, that right there winds up being 1.22 gallons. The cooling blocks in the chiller(s) likely will add some volume too.
Two loops (sharing the same reservoir) would come in and out of that carboy. One to my office and thus the PC GPU and CPU blocks and back, and one to the chiller(s) and back.
I'd have to drill a couple of holes using diamond hole saws in the bottom of the carboy, and insert barb fittings that would then go to the pumps that go to the PC and chiller(s) respectively.
I'm thinking of reusing my dual EK D5 top and pwm pumps for the long line to the PC, and just using a single D5 for the Chiller(s)
I'm including a plural s on the chillers as I am not entirely decided there yet.
My favorite potential plan thus far includes a small like 1/10 hp or 1/6 hp chiller to handle idle and near idle use and then a big 1/2 to 1 HP chiller that kicks in only when needed (long encodes or gaming runs and the like)
I'm hoping that if I use an Aquaero controller to control the flow on each of the two loops, I can keep the flow really slow at idle which means, the small chiller brings the coolant down to target temp, and the big chiller does not trigger unless the little one can't keep up, at which point the flow increases, the little chiller is unable to hot target temp before it gets to the big chiller and it kicks in.
This is going to be a little iffy though, as it depends on pretty accurate temperature measurements, and I don't know how accurate the thermal probes in the chillers are, and how much they drift over time. I'm certainly don't going to be bringing them to a NIST certified calibration house to have them calibrated every few months
If this doesn't work, I'm going to have to default to setting the big chiller a degree or two higher than the small chiller. It would be sub-optimal, as this way I don't get the coldest temps when I need them the most, but honestly, at slightly sub-ambient, I don't really think it is going to make a difference. I'll be cool enough for a good overclock/boost clock with a comfortable margin, and I'm not getting down to LN2 extreme overclock temps either way.
Even in the humid summer, with the dehumidifiers and AC going, my office is usually under ~74°F and an RH of 45%.
That gives me a dew point of about 10.7°C. I'd need a decent safety margin, as I don't want to risk condensation, but if I can hit 15°C it would still be pretty awesome.
Either way, these are LONG term plans. I'm just in the scheming phase right now. It may happen, it may not happen, but if it does, we are talking "the future" and not the immediate one.
Last edited: