T-Balancer and Watercooling

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
azadragon
Posts: 5
Joined: Fri Jul 01, 2005 10:13 am

T-Balancer and Watercooling

Post by azadragon » Fri Jul 01, 2005 10:31 am

Hi ppl :D

I am thinking of getting a T-balancer to automatically control the fans of my system. These will include one Papst 4412 F/2 GL to act as a case exhaust fan and a total of 6(!) Papst 4412 F/2 GM to be installed in a tripple black ice extreme in push-pull configuration.

As you may imagine the GM's will generate a lot of noise and need to be regulated. Withought beeing positive, I can say that for my cooling needs the fans will never have to run at full speed but lets say at 10V maximum (in terms of voltage as i am not too used to PMW). Now, I have the following questions:

1) Can the TBAN handle such a load? (if for example i use splitters for my rad fans and end up with three fan conectors and use the last one for my exhaust fan?)

2) Are these fans PMW friendly? IS there any need for attentuators?

3)Also i am not planning to use any sort of sensor that comes with TBAN; i plan to use my mainboard sensor via MBM5? Has anyone tried that? Does it work well? (Also gonna use the T-balancer LCDC plug-in)

Thank you for your time in advance.

AndrewC
Posts: 174
Joined: Wed Apr 02, 2003 7:16 pm
Location: Mobile, AL

Post by AndrewC » Fri Jul 01, 2005 11:33 am

I don't know much about the T-Balancer, but would suggest you only use 3 fans not 6. Choose either push or pull not both, because the cooling gain for both is negligible. Here's some testing from pro-cooling to prove it. http://phaestus.procooling.com/shrouds-BIX.GIF

azadragon
Posts: 5
Joined: Fri Jul 01, 2005 10:13 am

Post by azadragon » Fri Jul 01, 2005 12:43 pm

Thats ok i am thinking of using t-balancer with 3 fans anyway so my questions remain :)

AndrewC ta for your answer can you give me the link where u got the table i am intrested in reading more and the search doesnt work.

AndrewC
Posts: 174
Joined: Wed Apr 02, 2003 7:16 pm
Location: Mobile, AL

Post by AndrewC » Fri Jul 01, 2005 12:49 pm

Pro-Cooling Radiator Thread
It's a long discussion on ideal radiators, with a focus on quieter cooling.

teejay
Patron of SPCR
Posts: 749
Joined: Mon Aug 02, 2004 2:23 am
Location: The Netherlands

Post by teejay » Fri Jul 01, 2005 11:42 pm

I believe the T-Balancer can handle 15W per channel, but you should check the specs on their site to be sure. The 120mm Papst fans are OK with the T-Balancer and don't really benefit from attenuators in my experience (read about it here [SPCR forum link]). However, 6 of these fans will be very very audible, even at 5V.

The T-Ban should work fine without sensors an I have tried it with MBM5 and that works nicely as well. However, by relying on the mobo sensor you loose the biggest advantage this thing has: the ability to function independent of any controlling software. No matter what happens on the software level, your rig is always cooled because the T-Bal has its own autonomous sensor grid and controlling logic (mmm, that sounds a bit too much like marketing-speak... NM). You can always install one or a few sensors (analog ones are definately preferred), but I don't really recall if you can use MBM values alongside sensor input. Anyway, my advice would be to use T-Balancer sensors, not the mainboard ones.

azadragon
Posts: 5
Joined: Fri Jul 01, 2005 10:13 am

Post by azadragon » Sun Jul 03, 2005 6:56 am

Ok, I ve done a bit more reshearch and now I can say why i wouldn't prefer the t-balancer's sensors as well as some questions :)

Reasons for using MBM5 sensors:

1) Spaghetti effect. I like clean installs and these sensors will have a very bad effect on the tideness of my case.

2) Ill be using a matrix orbital and LCDC. Currently there is no WORKING plugin for LCDC and T-balancer which means that there is no way od displaying the temps on my lcd screen. On the other hand, by using MBM5 you can have all the info you need, and you can get the fan speed from T-balancer by connecting some tacho outpouts to your mobo.

3) My on-board sensors are very accurate. Not to say that these come with the TBAN are not but placement will be inevitably worse as the onboard sensors are very close to the cpu.

Now:

1) Teejay from what ure saying i guess that ure implying that there is a potential danger if ure using mbm5. Are u suggesting something like MBM5 crashes and TBAN loses its input or get an altered one? What happens in that case (if for example mbm5 crashes?)

2) Can mbm5 sensors assigned to all and every channel (i.e. work the exactly same way as digital and analog sensors of TBAN)

3) What happens at start up? The tban wil remain with no input till MBM5 is initiliased. Will that force TBAM to adapt a default mode and run the fans at 100% till MBM% kicks in?

4) I know that MBM5 must be ruunning for the TBAN software to get the values but does the TBAN software must be running as well alongside?

I 'd like to thank everyone for answering my irritating newbie-type questions :D

teejay
Patron of SPCR
Posts: 749
Joined: Mon Aug 02, 2004 2:23 am
Location: The Netherlands

Post by teejay » Sun Jul 03, 2005 9:48 pm

azadragon wrote:Spaghetti effect.
I hear you. Clean installs & T-Balancers happen from time to time, but they do require a black belt in cablegami.
azadragon wrote:1) Teejay from what ure saying i guess that ure implying that there is a potential danger if ure using mbm5. Are u suggesting something like MBM5 crashes and TBAN loses its input or get an altered one? What happens in that case (if for example mbm5 crashes?)
(...)
3) What happens at start up? The tban wil remain with no input till MBM5 is initiliased. Will that force TBAM to adapt a default mode and run the fans at 100% till MBM% kicks in?
No, I did not mean that it could be hazardous to your equipment: the one time a few of my sensors failed (they got disconnected) the T-Ban set all fans that were configured for those sensors at 100% speed. I'm guessing the same will happen at startup.
azadragon wrote:2) Can mbm5 sensors assigned to all and every channel (i.e. work the exactly same way as digital and analog sensors of TBAN)
(...)
4) I know that MBM5 must be ruunning for the TBAN software to get the values but does the TBAN software must be running as well alongside?
I don't know... but I'll check this evening since you got me kind of curious. Since integration with MBM5 happens at the software level and the T-Ban only "sees" the MBM sensor if you start the T-Ban software after MBM, chances are that you need to have both apps running... will get back to you.

azadragon
Posts: 5
Joined: Fri Jul 01, 2005 10:13 am

Post by azadragon » Mon Jul 04, 2005 12:04 am

teejay wrote:will get back to you.
Excellent, i'll be waiting. Thanks a lot for your help I really appreciate it.

teejay
Patron of SPCR
Posts: 749
Joined: Mon Aug 02, 2004 2:23 am
Location: The Netherlands

Post by teejay » Mon Jul 04, 2005 12:12 pm

OK, did some testing. The bad news up front: when MBM5 is closed while the T-Balancer is configured to use it, the T-Ban software assumes the "sensor" temp has dropped to 0 degrees Celsius and lowers the fan speed accordingly, regardless of the actual CPU temperature :shock: that alone would prevent me from ever using this combination. I tried creating a custom response curve that would set the fan speed at 100% for 0C and 30% for 1C, but apparently you cannot create descending curves. Another problem is that you need to have the T-Balancer software running alongside MBM for all this to work, otherwise the fan will just spin at 100%.

The good news: if you do not assign a physical sensor to a fan channel, the connected fan will spin at 100% speed until the T-Balancer software is loaded alongside MBM5. Also, as long as you define a suitable response curve, the controller reacts very quickly to temperature changes (which is the way you would want it to). Trying to maintain a target temperature of say 55C (the alternative of using a response curve) proved to be impossible: even though gain and fan adjustment speed were configurable, I could not find a setting that changed fan speed slow enough so as not to drop the fan speed to 0% when the temperature dropped 1 degree yet fast enough not to bake the CPU when the load suddenly goes up. I could not find a hysteresis setting for the MBM sensor setup, the way the regular sensors use.

Having tested all this, I stand by my original recommendation: the T-Balancer works beautifully with its own sensors. That way you don't need any software running and your fans are automatically adjusted. For me, the MBM combo is not reliable enough to trust my hardware on. With analog sensors the T-Balancer really shines, even though it still has its flaws.

Have you considered using Speedfan and a motherboard that allows fan speed adjustment?

Becks
Posts: 76
Joined: Mon Mar 01, 2004 11:19 pm

Post by Becks » Mon Jul 04, 2005 2:48 pm

Well I just have to say that IMO making the fan speed a function of the cpu's temp is the wrong way to do things, making it a function of the water temp is much more predictable and easier to read.

My wced setup has been going fine since the tbalancer came out basically... water says <33C giving cpu temp at max ~43/44C. Fans are generally <500rpm.

azadragon
Posts: 5
Joined: Fri Jul 01, 2005 10:13 am

Post by azadragon » Tue Jul 05, 2005 1:46 pm

Thanks anyone again. It seems that for the setup i have in mind i have the following options:

1) Go without the T-BAN and auto-regulate my fans via mobo with no additional software.

The mobo i'll be getting (ABIT AN8) supports automatic voltage regulation for up to 6 fans. This is done by specifing to 'hot points' one low and one high and matching this with a reference temp (e.g. CPU temp). I GUESS that the curve between the two points is linear and i have no idea how well feedback works in such a system.

Pros:
i) Less roucources taken (no additinal software)
ii) All information can be displayed in my matrix orbital
iii) Less cables, no spagetti and a free usb port
iv) Cheapest option

Cons:
i) Curve defined only by 2 points
ii) Fans can only be reulated by 5-12V
iii) Not sure about the efficiency of the feedback mechanism (Does any1 have experience with mobo controlled fans???)

2) Go with the TBAN and use MBM5 sensors

Pros:
i) Efficient regulation
ii) All information can be displayed

Cons:
i) Danger of MBM5 crashing
ii) More resources and ports
iii) Noisy startup

3) Go with the TBAN and use TBAN's analogue sensors

Pros:
i) Efficient regulation and max quitness
ii)Prolly the safest option
iii) Moderate resources

Cons:
i) Spagetti (bad)
ii) Info cannot be passed to matrix orbital (VERY bad)
iii) Disparancies between analogue TBAN's sensors and mobo sensors



So what do you suggest? Please feel free to suggest an alternative to the above as well. I think I am a bit inclined to the first choise and I'd appreciate if somene with relative experience could day a bit about how efficient mobos are in regulating fans. Also du you think tha speedfan can help in any case?

Thank you.

teejay
Patron of SPCR
Posts: 749
Joined: Mon Aug 02, 2004 2:23 am
Location: The Netherlands

Post by teejay » Tue Jul 05, 2005 1:59 pm

An option you missed (already pointed out by Becks): get the T-Ban with one or even several water temperature probes. You can also add a flow meter for really nifty measurements :)

Does your LCD support JaLCD? Apparently the T-Ban supports this, but I am an absolute LCD n00b...
iii) Noisy startup
You're kidding, right? At least I hope it's very low on your "cons" list :D Since the T-Ban gives all fans 100% fan speed just after boot, you'll have a noisy startup anyway. If not already obvious: I'd suggest the T-Ban with sensors & water temp probes.

Post Reply