LSD - need an opinion...

[bilbat]

Finally have the MOBO on its way - decided on a DS5 as it seems to be a DQ6 w/no CrazyCool, and one less pair of eSATA cables. Also have 40 & 30mm fans coming for (respectively) the north & southbridges, will rubber mount about a sixteenth or a thirty-second off the heatsinks. Going into a CM Cosmos 1000 w/five fans - lots of general airflow... My question, if it were you, which way would you mount the fans. I'm undecided - if I 'suck' the sinks, I'll be blowing the heat out into the case; if I 'blow', I'll probably get better heat transfer, but the heat bill be blowing across the board. ???

I intend to photograph and document everything carefully as a sort of 'how to' for air cooling a GB MOBO...


Bill

[Lsdmeasap]

Re: LSD - need an opinion...

I would point them down or across (For the NB if you can). Sucking would likely not cool as well. SB cooling is really not needed unless you plan to pump some crazy volts into it.

I saw your other comment as well about the heatpipe, I would recommend against lapping the heatpipes unless you plan to screw them back down. And if you do plan to use screws, be very careful not to crack the NB.

[bilbat]

Re: LSD - need an opinion...

Gotcha, thanks.

Already have the nylon 4-40's you recommended in the 'sticky'; the same 'thermal probes' you noted there actually work pretty well as a torque wrench Actually have pretty much experience in 'board surgery' - Allen-Bradley output board used to cost $650, & the triacs that always blew were $1.29 at Radio Shack. Also, remember hanging a lot of h'ware onto my Kaypro, which I think had only a two layer board, and I seem to recall being pretty thin & flimsy.

Intend to do some experimenting first by putting various sized measured blobs of Artic silver on some blocks of plexiglass & torquing 'em together to watch how they spread...

Figured as long as I was putting fans on the board, I might just as well cool everything the factory thought required a heatsink; I don't plan to really 'crank it' (intend a 9550 @ 3.5), but will be running 3 RAID pair (including 4 Velociraptors) through the SB and it can't hurt reliability: I only build a new workstation once every six or eight years!

Thanks again;

Bill

[Lsdmeasap]

Re: LSD - need an opinion...

Sounds like some good plans! And it should last you a while if you keep up the upkeep on things.

Just be careful when tightening down that NB. Here is a nice review on Paste application techniques >>>

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=170&Itemid=1&limit=1&limitstart=3

[bilbat]

Re: LSD - need an opinion...

Thanks LSD, you just cost me $7!!!

Really good article (& I must have read ten of 'em by now...); the best line: "Fixing the first problem required some wet-sand paper and elbow grease (that's a metaphor for hard work, so don't go searching the web for another product)." Also contained a pointer at the end leading to a good 33 product TIM comparo - wound up buying something I'd never heard of: TIM Consultants T-C Grease 0098. Best thing - for once, for a small item, the shipping was less than $2 - why can't everyone do that?! Ordered a six dollar cable from NE along with some other stuff - when checking out, it automatically generated a seperate order # and charged me six and a half for shipping! %$#*&!

[VorLonUK]

Re: LSD - need an opinion...

Thought I'd give some NorthBridge Feedback...

I'm using a Zalman ZM-MFC2:- ::: Zalman, leading the world of Quiet Computing Solutions :::
and have carefully mounted all the temperatures sensors (which I've checked for reasonable accuracy).

Out of the 4 sensors, one is affixed to the Northbridge Heatsink of my P35C. The maximum temperature I've seen (and on a hot day - room ambient @ 30C) and whilst looping 3dmark Vantage is 60C. Obviously taking into account Core to Heatsink and Heatsink to Sensor thermal inadequacies could easily push the NB realistically up by another 10C. However, this still indicates that the large NB Passive Heatsink Gigabyte use is well up to the job and far better than the solutions of old (and new) of a smaller physical Heatsink utilising a small cheap and cheerful fan.
Even those who choose to upgrade the fan to something of better quality will always be hampered at some stage by dust.

Personally I'd take a fully passive solution any day. You can't beat it for quietness and reliability. Obviously if Performance is a must, then you have no choice than to add some active intervention and sadly all the problems that go with it too.

[bilbat]

Re: LSD - need an opinion...

Glad to hear you're happy with the MFC2 (lotsa people on NE reviewing claim to have fan 'revving' problems) - I plan on using one to do my six case fans (in a Cosmos 1000): two PWM exhaust in the top, daisy-chained to channel 4; one exhaust at the back on channel 3; one intake in the floor on channel 2; and two sucking through the HD cage Y-cable to channel 1... Temps: channel 1 in the HD cage; channel 2 in the intake airstream for room ambient; channel 3 suspended in the middle of the case; channel 4 in the exhaust stream at the top. Case has foam filters in the bottom intakes to help keep the dust out, & all four MOBO fans hooked up to MOBO monitoring...

[VorLonUK]

Re: LSD - need an opinion...

Glad to hear you're happy with the MFC2 (lotsa people on NE reviewing claim to have fan 'revving' problems) - I plan on using one to do my six case fans (in a Cosmos 1000): two PWM exhaust in the top, daisy-chained to channel 4; one exhaust at the back on channel 3; one intake in the floor on channel 2; and two sucking through the HD cage Y-cable to channel 1... Temps: channel 1 in the HD cage; channel 2 in the intake airstream for room ambient; channel 3 suspended in the middle of the case; channel 4 in the exhaust stream at the top. Case has foam filters in the bottom intakes to help keep the dust out, & all four MOBO fans hooked up to MOBO monitoring...

Well the "problem" with the ZM-MFC2 is that i uses PWM on it's 3 pin outputs (x3) as opposed to the ZM-MFC1 (and i believe the plus too) that I can confirm use Linear Voltage regulators (7805's varying the ground pin, so they can never go any lower than 5 volts).

PWM (output from the controller itself) is the norm for most of the fan controllers out there (and we are talking about the standard 3 pin type here - NOT 4 pin). But rather than sending a PWM control signal (if you will) to the fan which has also has a 12 volt supply (aka 4 pin pwm), the controllers output is an actual "switched" "DC" (squarewave) output directly to the fan.
This PWM "switched" output can "Conflict" with the switching frequency of some of the DC Brushless fans on the market, causing them to rev up and down.(Brushless fans for those who don't know, use an IC to switch between poles - so that the fan may turn).
I have found this to happen with the Noctua fans, whereas the bulk I have tried have been fine. The problems usually occur too at low voltage, rather than at mid to high control.

I did design and fabricate and add on for the Zalman ZM-MFC1, that took the standard controller and turned the two 5-0-12volt switches to 1-1-1 logic. The circuit could discriminate as to what was being switched (ie as you only had a varying 3 state voltage level) and then inturn that switched through 74HC logic to output to variable DC voltage reguators that could be preset by the user.
I never used it in the end, but I still have the completed PCB, with all the components fitted.

[bilbat]

Re: LSD - need an opinion...

Well now, that makes sense! If you recall, I had spoken in another post about having thought hard about the possibility of a 'beat frequency' problem when using SYS_FAN2 w/a PWM motor, as the supply is PWM'd; I concluded that, with the motor's PWM switches "locked on", it shouldn't present any kind of difficulty. However, a 'brushless' motor (i.e., a motor that is essentially wound 'inside-out' with the coils on the stator and the magnets on the rotor) switches three (or five or seven, depending on how many poles it has) times per revolution, based on the rotor position as indicated by a rotary encoder, and the RPM. This won't pose a problem for my case fans, as none of them are brushless, BUT, now I'm concerned about my 40mm NB fan, which IS a 'Maglev' brushless design (Sunon KDE1204PKV3); guess I just have to wait & see...

[VorLonUK]

Re: LSD - need an opinion...

This won't pose a problem for my case fans, as none of them are brushless

Really BilBat?

I have used, worked on and seen many applications that still use a proper DC motor ie with brushes, however I have yet to see a "Case" fan (the common type we associate with PC's etc) that uses Brushes.
All the DC fan's i've seen (PC related) have been the Multipole Brushless type, ie Stator/Hall Effect feedback.

[Lsdmeasap]

Re: LSD - need an opinion...

I cost you $7? Do you mean saved? How?

Ohh, I see! Well If I was you I would call up Newegg and ask about that. I do know one item may be $-3 shipping, but once you order 2 of anything there it will be $7-8 Shipping. Even if one item is free shipping, once you add something that is not you drop the free or cheap shipping as it both goes into a normal 3 day UPS box

Call them up and speak nicely, you may get a full shipping refund. They do alot of nicer things over the phone compared to a email about it.

[bilbat]

Re: LSD - need an opinion...

I am so glad I stumbled across this forum whilst evaluating MOBOs - I swear, every single day, I learn something new (and, generally, useful)! My perspective is from industrial apps (and, maybe, a little bit of RC car and helicopter experience) so I have always assumed that if a fan doesn't say 'brushless', it isn't - 'cause brushless usually isn't cheap. I never thought of Hall switching for the commutation, but once you pointed it out, it's obvious, and a brilliant solution - you already have a rotating magnet - why not? It's got to be even cheaper than an old-fashioned brushed motor, as you eliminate the brushes, springs, brush holders, and commutator. Have plenty in the 'junk-pile' - now, I've gotta rip one apart - just to appreciate the engineering:)