Window AC chiller box/phase change build

[technogiant]

I'm toying with the idea of adding peltier/tec cooling to the cpu. It would have a 3 fold benefit, add heat load to the heatpipes and prevent freezing, further cooling on the cpu and add heat load to the system to help prevent floodback issues.

Drew I've seen you comment on peltiers before in that they don't perform so well at sub zero temps....with my load temps of +20 to +30 on the cpu do you think I'd see benefit from them?

My one concern with additional peltier cooling would be my cpu temps may go below chamber temps and so cause condensation (with chamber temp of -22 perhaps a little higher with added peltier load maybe cpu temp wouldn't approach this).....don't know if that would be an issue as humidity must be low in there....it's been sealed for almost a week, there is no air getting in due to slight positive pressure and little if any getting out....balloon is still inflated after all this time.....I've enough desiccant in there to draw out 200cc of water which has had ample time to work.

At the moment I only have an arctic freezer pro 7 with 3 heatpipes....just using it for experimentation, if the heat pipe issue was sorted I'd get one of the big dogs.

Additional Comment:

Idea toyed with......peltier ordered and enroute



Uploaded with ImageShack.us

Also bought some copper flat bar to sandwich the peltier as it will overlap the cpu.

The peltier unit is rated at 228w on 15.2 v that equates to 180w on a 12v supply....think this will be enough for my q6600.....using a power supply calculator at my overclock it is meant to draw 200w...but I think the lower temp I'm using will mean there is less "leakage" in the silicon and so lower power usage....not used a pelt before...hope what I've bought is alright...perhaps I should have waited for a reply....I'm an impetuous child..lol

[Orthello77]

TG its been a while since i used Tecs, more than a decade i would suspect so excuse me if i'm a bit rusty. The max rating will be at a 0c delta between hot side to coldside i believe, eg no temperature shift. Also the cpu calculator may be calculating the maximum TDP of the cpu so that could be high for day / day usage. I would say in gaming etc you will see some drop, but running say intel burn test maybe not. It really depends how hot your cpu is going to get under normal conditions. I would start of at a lower voltage and try to take benefit of the temperature drop and then ramp it up again. Obviously for the Tec every watt less it has to shift from the CPU is going to help it keep the temperature delta higher.

I hate to say it, as i love your impulsiveness (thats me to a t) but you would get better temps i believe by water cooling the cpu only. Later on may you could do both, water cool the peltier . What you may find is the heatsink getting overloaded at a point with the CPU and Tecs heatload combined. It may cope with it - but at a raised hotside temperature and that could gobble up your gains from the peltiers hotside / coldside delta. Liquid cooling this will alleviate that. Peltiers do drop off in how much they can shift the temperature (the delta) as the hotside temperature drops unfortunately , colder is not always better with Tecs. Henceforth if you do get to water cooling the cpu, the peltier with its reduced efficiency at subzero temps may be of little benefit to the load it adds to the radiator system. You may have to try it and see if you go that route.

Try it and see , its a hard one to predict.

[Drewmeister]

I agree with everything Orthello said. The heatsink will have to dissipate upwards of 200w x 2 = 400w. I honestly don't see that happening. You're still using a water filled heat pipe... my guess is the water vapor would condense and freeze further out from the heat source. It seems simple enough of an experiment though. Give it a shot and see what happens!

Btw.. the frost point inside the box will be much lower than the air temp. I used to test -70c phase on uninsulated mobo's in a -20c chillbox.. never had any issues.

[technogiant]

Appreciate your views Drew & Orthello, I think I may have been hoping for a little too much with the peltier...... At idle my cpu temp is about +15c above chamber temp. I think the peltier being of a similar power to the cpu will add a similar heat load as the loaded cpu does, that is about +50c above chamber temp....so the hot side at idle will be about 15+50= 65c above chamber temps so with a chamber temp of -25c thats about +40c on the hot side of the pelt.....should translate to an idle cpu temp given the 67c temp delta of the pelt....40-67= -27c
When the cpu is loaded it adds about +37c to the idle temp give a loaded cpu temp of about +10c and a hot side pelt temp of 77c.

That's using temps obtained with my current heat sink which is little more than a stock intel one. I'm hoping using the pelt will allow me to use a much more powerful heat pipe based heat sink which could draw temps down another 20c

Think it all depends on a couple things, firstly will the pelt stop heat pipe freezing....you could be right there Drew, it may just freeze further along the pipe but I may be able to prevent that with fan control. Certainly more chance of being able to control this given the added peltier heat load rather than depending on cpu idle heat production.
Second would be again as you mentioned Drew the heat sinks total capacity, not a problem with a solid heat sink, it will just work at a higher temp.
But with heat pipes they have a limit as to how fast that water can recirculate, too much load and it will all be vapor with no liquid left to cool. Don't think condensing it will be a problem because of chamber temps, but the rate it flows back may be limiting, orientation of the heat pipes may allow gravity to assist in that although I'd guess it moves predominantly through capillarity in the sintered copper matrix.

On that note though provided the heat pipes don't freeze and if my calculations are anywhere near correct with the solid heat sink then the max hot side pelt temp will be +77c I would imagine that a heat pipe heat sink is designed to deal with +77c temps without overloading, and in all probability the hot side will be a lower temp with the more powerful heat pipe sink...perhaps +57c which would be about the optimal operational temp I'd imagine, and would give a loaded cpu temp of -10c.

While it may sound strange, the idea of this is not just about cpu temps, although I am hoping it will give a significant improvement. It's about allowing me to use heat pipe coolers, my choice of solid coolers is very restricted particularly of high performance ones.

The other reason is floodback, atm when my system is idle I've taken to running furmark in the background as I get concerned when the temps start to drop to -28c.

So I'm using my expensive gpu set up as a heater!!

I had thought of adding a separate heat source to the chamber to protect against this but that is really quite wasteful.

I thought a peltier which I don't intend putting any control on, it will just be strapped to 12v, would be an ideal solution, improving cpu temps, adding heat load at idle and hopefully allowing me to use heatpipe heat sinks.

Just a further query...does "speedfan" operate in terms of fan control at sub zero temps from the internal cpu sensors? or would I need to use some external temp probe and control set up to get fan control at these temps?...an external control may even be preferable as I could attach the probe to the extremities of the heat sink to make sure it didn't drop below zero.

Additional Comment:

Well my peltier unit arrived today, just got to wait for the copper flat bar to sandwich it now.

In all it is going to have a thickness of 1.3cm, I guess I'm going to have to adapt my heat sink mounts to accommodate that?

Was thinking of putting threaded rods through the mount holes and securing with nuts and non conductive washers....Is that the usual way?....I don't have a back plate, is that a concern provided I'm careful with the screw down pressure?

Additional Comment:

I'm thinking of powering the peltier from two pcie 8pin cables I've already sealed into the chamber.
I've no concern about power draw I've got a 1200w psu.
The cables are on separate rails and I see that each 8 pin connector has 3x +12v (yellow) cables and 5x 0v ground cables (black).

I'm not sure how much current each individual cable can carry but was thinking of connecting one or more of the yellow cables from each rail together and the same with the black cables, so to spread the load onto both rails and perhaps on to more than one cable from each rail.

Is it alright to connect same voltage cables from different rails together like that? (I always power my graphics cards with one cable from each rail to spread the load, but not sure if its alright to directly connect the cables?)

[Orthello77]

I'm optimistic the peltier will give you some benefit just as long as the hotside temp is maintained and doesn't gobble up all your gains. The heatsink will surely will be above zero so that could get rid of the heatpipe freezing issue or it could just help it but not entirely stop the freezing inside the pipes like Drew mentioned.

Unsure about mixing the rails, technically i think it should work as essentially its like running batteries in parallel (and that works) but it wouldn't hurt to get a second opinion on that one.

The peltiers need good pressure to work effectively , if you can't use a backplate then just load the tension evenly as possible, i never did use one but a torque screw driver is recommended for mounting peltiers to get the right tension.

Unsure about the speedfan at subzero too mate ... sorry i'm not much help today .. i've only used it in positive territory (that will change soon i hope lol). Speed fan can be operated manually eg set % of maximum rpm or it can work of CPU or GPU temps and variate to your custom set profile. Not sure how it will work if i can't make sense of the CPU or GPU temps though - maybe placing the probe to ensure it doesn't drop below zero (i've never used it with a probe senser but that would be ideal).

[technogiant]

Thanks Orthello, there is certainly no instruction book for this sort of thing.
As I said, I do power my individual graphics cards across the two rails...not sure if that is effectively the same as directly connecting the cables...I would imagine so...I mean the cards effectively have three power supplies with the pcie slot and two connectors.....I doubt these are all kept separate....just don't know for sure.

I don't really know how I'm going to manage the fan speed, the built insensors just don't operate well at below zero......its a shame you can't set fan speed according to %age load utilization of the cpu or gpu.....that wouldn't need temp sensors......maybe that's an idea for you coders out there, cpu sensors are notoriously inaccurate at lower temps.....come on guys you know you can do it...lol
But as it stands I'll either have to just set manually...but one setting wouldn't suit my needs across all power needs.
The only other way is to use a secondary probe/sensor....which would be good attached to the heat sinks extremities to prevent it going below zero, but how I get the probe to talk to the pc and regulate fan speeds I have no idea.....I'm hoping that the pelt will load the heat sink enough so I can run fan at 100% without freezing...but just crossing fingers on that one.

Additional Comment:

Just found this in wiki about psu power rails

"These groups are the so-called "multiple power supply rails". They are not fully independent; they are all connected to a single high-current 12 V source inside the power supply, but have separate current limit circuitry."

So I think it's okay to connect them as they come from the same source, if they were from separate sources I'd have been concerned that voltage ripple would have at times caused reversed current flow, but as they are from the same source I think it will be okay.
I think using one yellow/black cable from each rail will be fine, looking at the size of the cables that feed the pelt unit two would be more than adequate...also the psu standards say each cable should be able to handle 18A....there will only be about 10A per cable....ummm perhaps I'll use 2 from each rail to be on the safe side.
Each 8 pin connector is 150w so two of them will supply the pelt's power needs.
Something I've never understood about 6 pin and 8 pin pcie connectors....6 pin are rated at 75w and 8 pin at 150w yet the additional 2 pins on the 8 pin connector are just another 2 black/ground cables...doesn't make sense....unless its something to do with cable sensing and switches the current control circuitry to deliver 75w for the 6 pin and 150w for the 8 pin...in which case Id have to connect all the black cables.

Additional Comment:

I've had a few answers back now that are indicating heat pipe and vapor chamber designs may be good down to about -30c.
Msi have said their gtx 680 is good down to this level apart from a bios problem with the fans.
Arctic have indicated their cpu coolers are good to this level and now prolimatech have also indicated that although they use water that theoretically their product should be good to -30c.

Perhaps they are using low enough pressures in there to lower the freezing point of water after all? Didn't find that with my arctic freezer pro 7....but that is an old heat sink now...perhaps they have changed designs?

[Orthello77]

TG thats great knews on those answers back. Good that you can just pick something off the shelf like the msi card too. You can only hope the designs have changed , you have to wonder about their testing, i would say MSI would be rock solid though as they cater for the LN2 people and would have tested their card in extreme conditions.

[technogiant]

Yeah....It would be nice if true...but I'm far from convinced yet...particularly considering my practical experience with my arctic freezer pro.

I found that graph of water phase change vs pressure and temperature I mentioned previously.



Uploaded with ImageShack.us


As you can see the pressure that is required for water to freeze at -30c is 1mbar....and at that point water does not actually have a liquid phase, it just sublimes directly from a solid to a vapor and back again.

So theoretically I can't see how they can possibly work at -30c, there would be no liquid phase to allow the water to flow back to the heat sink base.

Additional Comment:

Found a nice little article over at "Tom's" about pcie power connectors. I happy I know whats going on there now. Two 12v and two ground cables from each of the two power rails will be more than sufficient.

http://www.tomshardware.com/reviews/...e,3061-12.html

[Drewmeister]

The pressure drop inside the heat pipe won't matter, the freezing point will still be at or near 0c. Would be nice if you could change the working fluid.. unfortunately that's not an easy process and could very well end up losing capacity.

Using software to control the hs fan on the tec might be a bit risky. If the pc hangs or crashes it could be the end of your cpu or worse. Best to use a separate control for that, preferably a dual or triple stage digital t-stat with thermocouple probes.

Not sure about that 228w tec providing 180w at 12v, might actually be more like 150w or less. Can you disable cores if needed for testing? Hope it works out.

You'll need higher mounting pressure as orthello mentioned. Ideally you would have a strong top hold down plate.. not always an option with a heat pipe cooler though. Maybe combine thermal epoxy with the wimpy stock mounting system.

You can put the floodback worries to rest with a simple test run. Run the chillbox down to max low end temp. A rotary compressor will bounce or shake around.. might even hear a popping sound. That a/c has a large accumulator and also two dips in the suction line to trap and meter off liquid. I really don't see it being an issue unless you plan to cycle the unit at -20c or something.. which of course your not. Best to test it out though.

[technogiant]

Yes, I must admit I have no confidence in the range of operation down to -30c that some of the mrfs have stated.....It just doesn't make scientific sense.

I hope I haven't chosen a pelt that is too small, I was trying to balance between enough power and not excessively heat loading both the chamber and heatsink.

It may be okay, I know at lower temps the cpu silicon will "leak" less current and so use less power than will be indicated on psu power calculators.....but how much would be pure speculation....just be a matter of testing it out....I'll use my solid heat sink first with fan set to 100% in bios so as to not introduce too many variables.

Here's a graph of what I'm talking about with 3770k



Uploaded with ImageShack.us

You can see that power use goes up exponentially....the graph only goes to +30c, looking at other sites the 3770k at the lower over clock of 4.5GHz under ambient cooling draws 244watts, so by dropping temps to +30c they have reduced the power draw by 50% to 115w.

So if this translates to the 200w psu calculator quote for my q6600 at its current oc and voltage (keeping in mind that being oc'ed to that level with ambient cooling temps must be in excess of +70c) then at the +30c which are my current loaded temps I'd think the cpu is drawing max 100w, and that's at my current temp, the pelt and better heat sink will drop that further...hoping for -10c loaded which will reduce cpu power draw even lower and well within the pelts capabilities.

And here's a graph for overclocked prestler which shows similar findings...so it's not just an Ivybridge characteristic.



Uploaded with ImageShack.us

The next size up pelt I could find was 400w which I felt was massive overkill.

I can't alter the number of cores in the bios that I have seen......I'll just sit in the bios hardware monitoring and monitor temps with cpu idle to start with to see what happens.

I'll have to see what I can do about clamping pressure, but think my hands will be pretty much tied by heat sink design on that one.

That's good to know about floodback......there are a few suction line "U" shaped dips in the system, which as you said must help.

I have noted some popping noises though, but generally when I have restarted the compressor after a shut down....I have been leaving it for at least the 3 minute minimum that the instructions recommends though...perhaps I should leave it longer as its operating at lower temps than usual.....would you recommend a minimum time before restarting the unit?

[Drewmeister]

I strongly advise against cycling the compressor at those temps, if you have to, then give it at least 5min off time... preferably 10. The problem is that all the liquid ends up in the evap and then gets sucked right into the accumulator upon startup. Chances are it will be ok, just keep a close eye on it.

I still think the q6600 at 3.9-4ghz is going to be putting out well over 150w. Years ago we needed to tune phase units for 250w to handle a Q6600 in the range of 4.5ghz. The cpu may have been more around 225w.

Looks like you'll need around 300psi of pressure.
Here's an excellent guide on TEC's by Ultrasonic2. http://www.tecpeltier.com/viewtopic.php?f=4&t=19

[technogiant]

I have no intention of cycling....this was just on a few times when I had to shut down for some reason and restarted a few minutes later so I hope that hasn't caused any compressor damage.
Normally it runs smooth as silk.....I'll just leave it for a longer period before restart next time if I have to shut down mid session for any reason.

Thanks for the link to that tec info.......

"Q. So how much Qmax do i need?

A.
This depends on many factors, but mainly the heat load your trying to cool and the temperature your trying to achieve. For a starting point you should have at lest twice ( two times for the yanks) the Qmax than the load your trying to cool. So if you have a CPU load of 200 watts you'll want to have at lest 400watts of Qmax . doing so will get your load to a proximately ambient. This is not core temps but cold plate temps or water temps (in a water to water chiller). For those of you who want cooler temps than that i would suggest you shoot for 4 times the amount of Q than you have in load ie 800Qmax. "

That has kind of put any hopes of using the tec I've got to bed....and it says you shouldn't join multiple rails on your psu which I was planning.....all getting a little too much trouble for the benefit now tbh.

I may actually revert to my original plan and get that Zalman all copper air cooler:-

http://www.zalman.com/eng/product/Pr...ad.asp?idx=167

The reason I steered away from it was because of its size and mobo clearance, but was thinking if that was a problem I could always use the copper bar I have to raise it slightly higher sort of like 5mm shims.

Additional Comment:

I think for now I'm just going to enjoy gaming on my current set up as is for a while...its working well.

I'll up grade my platform mobo/cpu/ram in the next few months...don't think I can hang on till Haswell, might just wait a while and see if a cooler stepping of IvyB comes out....but tbh that won't force my hand either way as I can get a solid copper Zalman 7500cu to cool the cpu.

Decision time will come when I upgrade my gpu's, if they don't work because of heatpipes/vapor chambers then that will force my hand to add a water cooling loop...which if I have to do for the gpu's then I may as well do for the cpu also.

Having said that with my 2x gtx460's 2048mb versions running at 980/1960/2000 on a 1920x1200 monitor....that probably wont be needed for a while...lol

But anyway thanks everyone for the input and support, particularly Drew and Orthello...couldn't have done it without ya.

[Orthello77]

Re the peltier, sometimes better ideas come along, like the portable A/C i bought which i knew by the time it was shipping i was not going to use because the window A/C was more suitable. I think in time to come you will inevitably watercool your components as that's the next big decrease in temps which is not too hard to achieve - its just $$.

I'm hearing what you are saying about upgrading, theres not a lot of point for that res, i run 2560x1400 with the two 1000mhz 580s and really theres not a single game that i would tangibly benefit from an upgrade at this point. The IvyBs don't really excite me either, ever since i learned about their thermal interface beneath the cap , i'd rather buy a sandybridge , maybe they will fix the interface in time or maybe it cannot be fixed (its a smaller die). We really need some hardcore nextgen games to come out and really force the issue.

Glad to be of help ! the stuff i have relooked at and the new ideas just over comments from you and Drew has really helped my system too. I'm really pushing for a this weekend testing so fingers crossed i'm in negative territory then too :-)

[technogiant]

Your right, a water loop would be the logical next step, but think I will only take it if forced to by the gpu heat pipe/vapor chamber issue, it is only $$ but tbh following my gaming session last night...only "hard reset" on steam...but still gpu temps using active vsync maxing out at +5c on gpu1 and not even registering above 0c on gpu2 and cpu temps not even moving coretemp numbers which are locked/bottomed out at 9,5,5,9 (actual temps below this)....chamber temps were bottoming out at -30c!! in periods of inactivity.....think that's good enough for me and keeps things a lot simpler and perhaps more reliable as a result, plus keeping the cheap/simple upgrade path that was my original intention with this project.....I really think barrring any future problems....it's pretty much job done, and I'm over the moon with the results.

[Orthello77]

WOW .. -30c chamber temps .. nice one TG .

Some sweet GPU temps too. Theres certainly something to be said for air cooled parts, resale is easier for one, in NZ its very hard to sell waterblocked cards at present as the market here is small for watercooled components.

Its great your finished, wish i could say the same, i glued in extra pieces to the resoiver last night making it two compartments (what a mission) silicon went over the evapourater by accident so out came the craft knife to scrape it off fin by fin lol. Just got electrics and top viewing lid to sort out and i'm ready :-)

[technogiant]

Yeah those gpu temps were while using nvidias active vsync, so gpu usage was only 30-40%....but that's how I game....I mean I don't play furmark..lol

I can sympathize about the silicon accident, luckily I missed the evap and just threw mine all over the acrylic case... ;-/ lol

On the subject of upgrades what do you think about IvyB, I know it's thermals aren't as good but when using extreme cooling methods as we are is that really an issue? it does have slightly improved ipc compared to sandyB

[Orthello77]

I don't much like what i've seen of Ivy Bridge so far, it has a poorer thermal interface compared to sandy bridge for minimal IPC gains. Infact air cooled at least the sandybridges seem to clock comparably higher enough so to make up for the IPC loss and at lower temperatures. I think the only saving grace is they Ivb produce less wattage , although maybe thats not overclocked, i really need to do more research about the leakage , suffice to say what i've read isn't to great when its comes to the overclocking aspect (air cooled at least). Drew may know how they compare to sandy bridge with exotic cooling ?

I was thinking the other day about the Asus Maximus V which has a watercooled channel for the mosfets etc but in a typically dumb move asus have not but a G1/4 thread on the waterblock / heatsink , just 3/8 inch barbs so that kills that idea for me, i run with 1/2 " and i'm not too happy about reducing the flow of the whole loop just for the mosfets.

It really stinks they make their best GPU cards with air coolers too .. its like they cater to a certain extent for the hardcore market but then stop short ....

So i'm thinking now to stick my upgrade back a bit as nothing too exciting there atm. Was looking at Sandybridge on that Asus board.

[Drewmeister]

IB scales very well with temp but the larger performance gains over SB don't appear till the more extreme temps, like cascade and LN2. One can nearly max out a SB on cold water... fully on SS cooling, whereas IB would need LN2. SB will outperform IB for your guys applications. At a minimum, a large SS(sub -60c load) unit would be needed to make much if any difference with IB. Cascade and up is where it's at though.

That's unfortunate about the tec plans.. would have been something if you could have pulled it off with -20c air. I do think you could have run it off a single rail with two independent molex connectors. It probably would have worked ok at very low load with that setup. Although interesting and fun.. probably not worth the time and effort. I'm sure you're within 100-200mhz of max cpu at that temp range anyways.

Those Zalman coolers are excellent.. I had a similar one on my s939 setup with a high cfm fan ...great performance. Cool how they clamp and machine all the fins together at the base. Hopefully it clears everything on your board without a shim... maybe you could just bend some fins if needed.

I'm pretty confident you'll max out any heat piped gpu in that chillbox short of hard modding it.

Orthello.. I'm guessing that mosfet block is probably a straight through design and wouldn't effect the flow rate much if any. If it is restrictive then you could also just add a tee and balancing valve if needed.

TG, thanks for sharing your project with us. It's not every day one gets to be part of something so unique like this. I really enjoyed helping with and seeing this wonderful build progress to completion.. I've learned much too. You are certainly a very creative and innovative person. Absolutely amazing job and results!

[technogiant]

Thank you both, Drew and Orthello...you flatter me with your comments.....till the next time TG signing out.

Additional Comment:

Okay.....so I've tried another type of cpu cooler that combined conductive cooling with heat pipe cooling......this sort of design was the most likely to work as the conducted heat should have thawed out the heat pipes when load increased.
While it worked to a point in that the idle temps didn't go through the roof it didn't perform well at higher loads...I think the problem is that the heat pipes are tuned to operate with in a certain temp range and I'm outside of that.

Will that isn't a huge problem for cpu cooling as I can use a solid heat sink it is a problem for future gpu upgrades as they all use heat pipes or vapor chambers.

I also tried direct conductive cooling of the gpus by running hundreds of 1.5mm wires from the gpu heat sink to the evap rad....and while it did work it gave 20c worse temps than standard air cooling.

So having explored several avenues I'm going to have to include a water loop in the chamber.

I could probably get away with just water cooling the gpus as my temps are good on the cpu with a solid heat sink but as I'm going to have to watercool my gpu's when I upgrade them I may as well also do the cpu.

In fact I'm going to start with the cpu and use my old hardware to get things right first.
So I'm looking for a waterblock to cool socket 775 thats also compatible with socket 1155 for when I upgrade and will be okay with methanol mix....also needs to be a match for my rad which uses G1/4 fittings.
This is my first go at water cooling so I'm going to need lots of help...especially as its non standard down to -30c and in the confined space of my chill box.

[Orthello77]

TG , you have a good radiator for the job in the GTX360 you have - may as well get use from that - i think its rated 1200watts at 10c delta with moderate air flow. Pair that with say a MCP655 pump should do the job more than well. I would use 1/2 inch ID piping but 3/8 is also fine and easier to handle / bend. I have used Primoflex clear tubing , its excellent at subzero temps , does not crack etc but will go very rigid with the colder liquid, but returns to flexible again at room temps i have found. All items can be found from FrozenCPU , but you may find closer to home also.

Mount your pump at the bottom of the loop if you can - easier for priming. I would use a small resoiver for bleeding , its really difficult with a T line.

Fluid mix good to -31c and further is 40% methanol, 30% glycol , 30% distilled / demineralized water. I would recommend this - have not seen any sludge using this mix. Use a pump filter. I have used the koolance inline filters.

Clamp all joins !! zip ties / twraps do work well if you run out of clamps. I have also expoxied the barbs to blocks to be sure.

Anyway .. bound to be lots more you need to know over time .. glad your upgrading and look forward to the results !!