Dual Reserator 1 PLUS issues

[Candybar]

Hi all,

i've got 2 Reserator v1 plus unit's in my loop, but this is not going very well.

Story:

I have added a second Reserator because i was able to pick it up cheap and it looks very awesome, it would also be nice if my temps will drop a few degrees.

I have added the 2nd Reserator behind the 1st Reserator , the problem now is that the water drops almost to the bottom on the first Reserator and the second one fills up until the top.
Also the flow-rate dramatically dropped.

i also have a extra pump in the loop (Swiftech 355) with extra XSPC top.

Loop: 335 pump --> HDD block x2 --> 240 RAD --> CPU -> MOBO -> 6990/6970/6970 in serie -> 360RAD -> Reserator 1 --> Reserator 2 --> back to the pump.

Is this because the reserator's have air hole in the top?

Is it better to make a serial setup?



Or is it better to create 2 loops?:
Loop 1:
6990+6970+6970 on 1 Reserator + Pomp and 360Rad

Loop 2:
CPU + Mobo + HDD's --> 2nd Reserator + 240 Rad. Can the pump of the Reserator handle this?

System Specs:

1x HD 6990 (Water)
2x HD 6970 (Water)
AMD FX 3850 @4.7 Ghz (Water)
Asus Crosshair Formula IV (Water)
2x Aqua Computer dual HDD Drive bay (Water)

Thanks Guys!

 

[GTXJackBauer]

Welcome to OCFs!

There's lots of restriction going on in the loop. One being the MB added to the loop unless you're extreme overclocking, the HD's and the second reservoir. Adding a second reservoir won't drop your temps what so ever. Your loop will reach a equilibrium. I would take the second reservoir out. If your pumps are the same I would buy a dual top as that will enable them in serial. Too many unnecessary things added to the loop. Maybe you were going for the looks but looks and bling can impede the loops performance.

What radiators and fans are you using? What are the FPI on those rads and RPM on those fans? What are your temps on those components? What is your delta temp?

 

[JeremyCT]

Are the Reserators sitting on the same surface (level with each other)?

Have you confirmed that the pump in each unit is the same (they do sometimes get replaced), functional, set at the same speed (the V2 has an adjustable speed pump, not sure about the earlier ones), and both are turned on?

The second Reserator really won't help much, especially if they're close together. It'll probably take a solid 30-60 minutes for the water in your loop to reach equilibrium after a change in load though. That's ... impressive.

Pushing that much water around definitely has an adverse effect on flow rates though, I know my flow rate is way better without the Reserator in the loop.

 

[Candybar]

Welcome to OCFs!

There's lots of restriction going on in the loop. One being the MB added to the loop unless you're extreme overclocking, the HD's and the second reservoir. Adding a second reservoir won't drop your temps what so ever. Your loop will reach a equilibrium. I would take the second reservoir out. If your pumps are the same I would buy a dual top as that will enable them in serial. Too many unnecessary things added to the loop. Maybe you were going for the looks but looks and bling can impede the loops performance.

What radiators and fans are you using? What are the FPI on those rads and RPM on those fans? What are your temps on those components? What is your delta temp?

Thanks for welcoming me

Yes the loops has a many blocks/Restrictions, but is this MB Block that bad?:

I thought because the CPU is water cooled you would do good to also cool the MOBO because of the missing airflow without CPU cooler.
Yes the HDD blocks are not very necessary, i might take them out of the loop.
I have only one 355 pump, and the 2 pumps inside the reserator's.
The reserator is more than a reservoir, it should also lose some heat right?

The 360 Radiator i use:
http://www.samenkopen.net/action_product/625299/971194
The 240 Radiator:
http://www.watercoolinguk.co.uk/p/Black-Ice-GT-Stealth-240-Radiator_153.html
Fans (6 on the 360 rad with a shroud):
http://nl.hardware.info/productinfo/17628/tacens-aura-fan-120mm#tab:overzicht

Temps without the 2nd Reserator:
CPU 30/40 IDLE, stressed 52.
GPU's, arround 40 IDLE, and on load (Mining some crypto currencies) with undervolting around 75/80. This is the "Delta Temp" and will not rise any further.

In this stage the system uses 880Watts from the wall.

Without undervolting the water become's to hot and the temps will rise to 95+.
In this stage the systems uses 1100 watts from the wall.

Are the Reserators sitting on the same surface (level with each other)?

Have you confirmed that the pump in each unit is the same (they do sometimes get replaced), functional, set at the same speed (the V2 has an adjustable speed pump, not sure about the earlier ones), and both are turned on?

The second Reserator really won't help much, especially if they're close together. It'll probably take a solid 30-60 minutes for the water in your loop to reach equilibrium after a change in load though. That's ... impressive.

Pushing that much water around definitely has an adverse effect on flow rates though, I know my flow rate is way better without the Reserator in the loop.

The Reserators are on the same surface/level of each other, they are slightly higher than the pc, about 20 centimeters.
The pumps are identical and has not been replaced. And turned on ofcourse .

 

[GTXJackBauer]

Yes the loops has a many blocks/Restrictions, but is this MB Block that bad?

Yes, it is not needed if you're not doing overclocking but just leave it for now.

I think you need to read up on water cooling and start over. Your Rad FPI look to be high while only pushing 1200 max PRM through them which won't cut it not to mention having them positioned at a higher level than your reservoir. I am sure they are holding a good amount of bubbles in them. Have you bleed your loop?

You need more radiator surface to cool your setup. You have 4 GPU chips and CPU to cool. Try doing 120.2 per chip should improve your temps substantially.

CPU 30/40 IDLE, stressed 52.
GPU's, arround 40 IDLE, and on load (Mining some crypto currencies) with undervolting around 75/80. This is the "Delta Temp" and will not rise any further.

What is the Delta Temps? DO you know what Delta Temps means? You're getting worse temps than air cooling imo.

I have only one 355 pump, and the 2 pumps inside the reserator's.

You're now saying you have 3 pumps? 1 external and 2 internally in a reservoir?

 

[Candybar]

Yes, it is not needed if you're not doing overclocking but just leave it for now.

I think you need to read up on water cooling and start over. Your Rad FPI look to be high while only pushing 1200 max PRM through them which won't cut it not to mention having them positioned at a higher level than your reservoir. I am sure they are holding a good amount of bubbles in them. Have you bleed your loop?



You need more radiator surface to cool your setup. You have 4 GPU chips and CPU to cool. Try doing 120.2 per chip should improve your temps substantially.



What is the Delta Temps? DO you know what Delta Temps means? You're getting worse temps than air cooling imo.



You're now saying you have 3 pumps? 1 external and 2 internally in a reservoir?

The Radiator is positioned lower than the Reservoir, in the bottom of the case
The Delta (Maximum change in temp) is now about 20/25

A Zalman Reserator cooling tower has a build in pump: ( i have 2 of them)

Reserator:

Inside Pump:

Im not necessary looking into the lowest temps, im happy if i can run my GPU's arround 60/70 degrees at full load, main thing is the system to be quiet.
I have the 360 radiator running with 6 of those fan's in push-Pull mode with a shroud between the push fan's and the radiator.

I will post some pictures soon to make it more understandable, tomorrow i will try running the reserator's parallel .

 

[JeremyCT]

Your temps are horrible for water cooling. Ditch the second Reserator (if I had cash I'd offer to buy one, I miss mine), eliminate the unneeded blocks, and add more rad like Jack suggests.

I'm not a big fan of the "120.2 per chip" rule personally. By that logic, you'd be looking at 120x10 worth of rad. However, a system that draws 1100 watts under load probably needs more than 120.5. If you want quiet, you need big radiator area.

The HD blocks are entirely unnecessary, and the MB block isn't doing you much good either. What does this monstrosity look like? If there's any airflow over the HDs and the CPU area on the MB, both of those blocks can be eliminated.

 

[marjamar]

I'd like to see some pictures of your setup myself. I have a EK mosfet/VRM water block in my loop. I move 75 to 80 gph though it at 100% pump speed. No one here has ever answered my question about what is a good standard for flow rates, so I can only guess based on how my rig cools, whether or not that is typical or not. There are a bunch of factors that go into making a good loop. I'm not saying that as one who has, rather one who struggles with it. I have seen a much improved loop on my rig however, so I am learning good things out of the bad things anyways.

So, post a few pics so everyone can have a look-see and do some better "guessing" to give you some help with.

-Rodger

 

[JeremyCT]

Rodger-

Some basic guidelines on flow.

0.5 gpm = bare minimum
1.0 gpm = this is a good minimum for performance water cooling
1.5 gpm+ = excellent, probably overkill

 

[Candybar]

 

[GTXJackBauer]

I'm not a big fan of the "120.2 per chip" rule personally.

I disagree. That rule generally applies to low FPI or quiet rig setups as well as moderate to high OCs. 120.1 per chip is for 30 FPI radiators and screaming fans. It's been like this for years and always has.

Rodger-

Some basic guidelines on flow.

0.5 gpm = bare minimum
1.0 gpm = this is a good minimum for performance water cooling
1.5 gpm+ = excellent, probably overkill

+1

Jeremy's on the money. It's just that simple. You can also do the math if you'd like on each added restriction in the loop if you don't have a flow meter.

Here's some more info on flow and restrictions.

Source

Source #2

 

[Candybar]

Have added some pics, i know it looks a bit messy but whatever.

 

[GTXJackBauer]

Have added some pics, i know it looks a bit messy but whatever.

Seems the links to your pics are broken. Use this forum to host the pictures.

 

[JeremyCT]

Those are way too close together to add much of any additional cooling. You need to have space around these for natural air convection. You've essentially just increased the volume of your loop.

 

[JeremyCT]

I disagree. That rule generally applies to low FPI or quiet rig setups as well as moderate to high OCs. 120.1 per chip is for 30 FPI radiators and screaming fans. It's been like this for years and always has.

Well, no, not exactly. We used to determine needed radiator size by estimating thermal load and working from there. I guess that fell out of favor. Too complicated for the nooblings I suppose.

 

[GTXJackBauer]

Well, no, not exactly. We used to determine needed radiator size by estimating thermal load and working from there. I guess that fell out of favor. Too complicated for the nooblings I suppose.

Its better this way in laymans terms and to simplify it than to get down dirty and nerdy to look at every little spec. Those that want to go that route can. That rule of thumb pretty much came from the experts, Martin and Con to name a few if I recall correctly. If I recall correctly, Martin was the one who said 120.1 @ high FPI or 120.2 @ low FPI to simplify things. I don't understand what's wrong with that rule of thumb. You still base the amount of heat generated by heat surface as well as FPI count.

You hear the 120.2 rule of thumb more often nowadays because a lot watercoolers want their rigs to be silent while getting great temps + OC. Whats the point of spending so much on a loud setup. To each their own I guess.

 

[JeremyCT]

I don't understand what's wrong with that rule of thumb

This guy has "5 chips". I don't think he'll need 120.10 to get "cool and quiet". That's where that rule gets a bit screwy.

Also, my rig has 2 chips. 120.3 worth of rad. Very very quiet, dT under max load of 7 deg C.

It just doesn't work well for all cases, so I favor a more customized approach I guess. If you're going to get into this sort of cooling, "spitballing it" by throwing stuff together and seeing if it works is a bad approach IMHO. I prefer a much more measured and deliberate sort of approach.

 

[GTXJackBauer]

This guy has "5 chips". I don't think he'll need 120.10 to get "cool and quiet". That's where that rule gets a bit screwy.

Also, my rig has 2 chips. 120.3 worth of rad. Very very quiet, dT under max load of 7 deg C.

It just doesn't work well for all cases, so I favor a more customized approach I guess. If you're going to get into this sort of cooling, "spitballing it" by throwing stuff together and seeing if it works is a bad approach IMHO. I prefer a much more measured and deliberate sort of approach.

That's just a way to get it started. Not saying its finalized or go grab them now. Of course they need to read up and do their homework. I always tell folks to read up and check out martin's page.

I don't need to explain myself but you should know better when he has 2 6970s and a 6990. Over 1000w+. A good high FPI 120.3 can get anywhere from 250w-350w @ 1800-2200 RPM give or take. Make it now 120.6 and I believe its still short. He goes 120.9 and he's getting good temps, ambient temp permitting.

We'll continue this tomorrow. The party bus is about to leave. (My little bro is driving to a family Xmas eve party) Merry Christmas btw.

 

[m0r7if3r]

I'm not a big fan of the "120.2 per chip" rule personally. By that logic, you'd be looking at 120x10 worth of rad. However, a system that draws 1100 watts under load probably needs more than 120.5. If you want quiet, you need big radiator area.

I'm not sure where 120.2 per chip came from, back when I was in every thread, my rule of thumb for recommendations was 120x1.5 per CPU, 120.1 per GPU, and round up...which actually works out to exactly what you recommend!

I disagree. That rule generally applies to low FPI or quiet rig setups as well as moderate to high OCs. 120.1 per chip is for 30 FPI radiators and screaming fans. It's been like this for years and always has.

I'm not sure about how long it's been like that...I don't think it was going around when I joined the forum, more on that to come.

Well, no, not exactly. We used to determine needed radiator size by estimating thermal load and working from there. I guess that fell out of favor. Too complicated for the nooblings I suppose.

To be fair, when you're getting going, especially if you're trying to do it fast, there's a LOT to digest, so having some simple rules that can be used as a starting point isn't necessarily a bad thing...I'd go so far as to say it's kinda necessary.

That rule of thumb pretty much came from the experts, Martin and Con to name a few if I recall correctly. If I recall correctly, Martin was the one who said 120.1 @ high FPI or 120.2 @ low FPI to simplify things. I don't understand what's wrong with that rule of thumb. You still base the amount of heat generated by heat surface as well as FPI count.

You hear the 120.2 rule of thumb more often nowadays because a lot watercoolers want their rigs to be silent while getting great temps + OC. Whats the point of spending so much on a loud setup. To each their own I guess.

Martin's never really been active on this board (he was mainly XS, little OCN)...so if it was him it didn't start here. I may have actually been the one to start it, initially I was saying 120x1.5 per heat source, but I revised to 120x1.5 per CPU and 120x1 per GPU in light of the triple SLI setups that came about around the GTX480/GTX580 era.

This guy has "5 chips". I don't think he'll need 120.10 to get "cool and quiet". That's where that rule gets a bit screwy.

Also, my rig has 2 chips. 120.3 worth of rad. Very very quiet, dT under max load of 7 deg C.

It just doesn't work well for all cases, so I favor a more customized approach I guess. If you're going to get into this sort of cooling, "spitballing it" by throwing stuff together and seeing if it works is a bad approach IMHO. I prefer a much more measured and deliberate sort of approach.

Having a rule of thumb does give you a good place to start from. I think using it as an absolute would be a horrendous idea, you'd end up with a lot of weird rigs. That said, I think the rule might need revising, since I'm guessing it's still stuck in the high-TDP CPU era, back when a stock CPU was doing 120W and an OC took it to 200-300W.

Stock CPUs now have a TDP topping out around 80W, and they only go to about 150W or so under OC (someone correct me if I'm wrong on that). The flip side to that is that because the lids are now all glued instead of soldered on, thermal transfer out of the core is worse, so you end up needing more cooling to keep things in check. If that changes, or with the next process shrink, I think that CPU cooling may become largely trivial, since there will be so little heat to cool (note: pure speculation on my part here).

 

[GTXJackBauer]

Well said Mor, well said. I totally and understandably agree with everything you said. Cheers, now I need to get out before I drink to much before the party.