4850 with Zalman VF900?

T

thefurion

n00b
Joined
Jun 5, 2008
Messages
54
Anyone know if the VF900 is able to be correctly mounted on the 4850's? I had a 7900gt before the 4850 and had the VF900's but would be nice if I can reuse it considering they were very nice :D. I hear that all/most fans that can fit in the 3800 cards are able to be used on the 4800 series cards

The VF900 seems to be mountable on the 3800's..
http://www.hardforum.com/showpost.php?p=1031667263&postcount=19

Thanks for the reply ^_^;
 
mm thanks for the reply :D. My asus card idles at around 50-55 degrees so I am guessing replacing the fan + some AS paste will only make that number go down. This is good news.. I have something to fiddle around with when I get back home today.
 
The 4850 uses the same mounting holes as the 3800 cards, so every heatsink that fits the latter will also fit the former.
 
thefurion said:
mm. I opened up my 4850 today and it seems like I do not have enough little sinks to cover the PWM's :(. I guess I will wait off until I find more RAM sinks...
Click to expand...

There's no proof that the PWMs actually need sinks. Just point a fan at them if you want. It's better than the insulation they get with the stock cooler.
 
I would highly recomend heatsinking the PWM on 4850 ! I installed a Accelero S1 on my 4850 and added some AKASA VGA ramsinks on the PWM. The PWM sinks run burning hot, actually my card would crash of overheating before I installed a 120mm fan on the Accelero.

Some people have just sawed off the 4850 reference coolers PWM copper heatsink and reused it, works well I hear.
 
aldamon said:
There's no proof that the PWMs actually need sinks. Just point a fan at them if you want. It's better than the insulation they get with the stock cooler.
Click to expand...

That's right, no proof as of now but the pwm get hot as hell. Are you willing to take the risk? I wouldn't.
And as always, good airflow in the case is the key for a cool computer (no pun) and can save you from a lot of problems.
 
^ Yep I never even tested the card without heatsinks on the PWM:s, might work, but did not want to risk it with a new card.
 
Wolfkin said:
That's right, no proof as of now but the pwm get hot as hell. Are you willing to take the risk? I wouldn't.
And as always, good airflow in the case is the key for a cool computer (no pun) and can save you from a lot of problems.
Click to expand...

sam0t said:
^ Yep I never even tested the card without heatsinks on the PWM:s, might work, but did not want to risk it with a new card.
Click to expand...

I'll just repeat what I've said in other threads: Just because something gets hot to the touch, or hot in your opinion, doesn't mean it needs a piece of metal stuck to it. The PWMs get hot in stock cooling. If the PWMs were in any danger of meltdown, they would not be insulated and sealed in with crappy TIM, sandwiched in with a horrible stock cooler, and still covered by warranty. The companies selling heat sinks for every little component are leveraging your insecurity.

My 4850 works fine with no PWM sinks (1.37V, 840/1100), just like my 8800GT and my X1900XT did and my 7900GT did and my 7800GT did and my 6800GT did. See the pattern?

BTW, ASUS agrees with me. Their 1GB 4800 has zero PWM/RAM cooling sinks:

http://www.overclock.net/hardware-news/347038-vr-zone-asus-hd4850-1gb-glaciator.html
 
While I respect your "experience", those are dangerous words.

The thetaJA of FETs used on that board are about 40 degrees C per Watt. Therefore if the local switcher has about 90% + efficiency...on a 150W card, you still are burning up ~15W in the FETs. If you have 6 on a board...you have about 2-3W per FET or a 100+ temp rise in the junction temperature of the device. Considering the average PC case has about 40 deg C air (unless you go nuts on fans)..you have about a 150 degree C junction (which is the limit). While this won't kill parts quickly...it sure will accelerate them towards death. Also, the hotter the parts..the higher the rds(on)...and thus higher ripple will result to the RAMS/GPU and thus potentially making your board unstable or not able to overclock as well as it could.

If you remove the stock heatsink and chose not to put sinks on the PWM's, at least for the love of your card, get a fan on those FET's (which happens indirectly with most coolers). It will reduce the effective thetaJA by a factor of 2 with good airflow and thus keep your FETs in happy land.

I mean...you don't need a seatbelt until you get in an accident...no?

Edit to your edit...there is still forced airflow over the FET's.
 
Trepidati0n said:
If you remove the stock heatsink and chose not to put sinks on the PWM's, at least for the love of your card, get a fan on those FET's (which happens indirectly with most coolers). It will reduce the effective thetaJA by a factor of 2 with good airflow and thus keep your FETs in happy land.

I mean...you don't need a seatbelt until you get in an accident...no?

Edit to your edit...there is still forced airflow over the FET's.
Click to expand...

I see what you're saying but the tiny heat sinks everyone is buying have very little surface area and will reduce temps by a few degrees, tops. That kind of reduction is just not significant on something like a PWM, which as we've seen run fine in the stock cooler at high temps. I've still seen no evidence that running with naked PWMs on the 4850 is any more dangerous than using the stock cooler. In fact, I've seen no evidence yet that the temps go up at all, air flow or no air flow. We have to assume from looking at the stock cooler that they weren't getting any air flow or cooling before. Regular case airflow is most likely better than that shit sandwich of a stock cooler.

And to address you final point about my edit, yes there's airflow with the ASUS cooler, but I bet everyone here has a spare 80mm fan laying around. That's better than buying and sticking little heat sinks on the card. If a little residual air flow is good enough for ASUS, it should be good enough for us.

Also, going back to the original post, wouldn't the VF900 provide just as much residual air flow as the ASUS cooler?]
 
why don't you guys just get the artic cooling new vga heatsink for the 4850 & 4870 ???
 
123Lanoix said:
why don't you guys just get the artic cooling new vga heatsink for the 4850 & 4870 ???
Click to expand...

Because my Sytrin VF1 Plus has worked on 4 cards now and I enjoy keeping my money? I'm sure theinfurion feels the same way about his VF900.
 
No, the VF900 does not provide as much residual air flow as the Glaciator heatsink because the fins are much more tightly packed, especially at the base of the heatsink where they converge.

I used to have a VF900 on my 4850 and the temps very decent, but not great. I then put an Accelero S1 on it and the temps are excellent. I run the Accelero S1 passively since there's enough air flow coming from my side case fan. It loads at about 60 degrees.
 
i ran my 4850 for a week without ram or pwn heatsinks just with a waterblock on the gpu. never saw any instability and was oc'd to 675/1100. that being said i did end up putting some of those shitay zalman heatsinks on the chips but i dont see how they could really reduce the heat being generated that significantly maybe 5C tops.
 
i can get 47c idle and 60c unload with stock cooler just by adjusting the fan speed.
 
I have a VF700 Alcu that I put on my 4850 as I had it laying around. Runs better than stock but not much ... but definitely quieter.
 
I most likely will not OC my card that much so I will probably just attach the VF900 on it tonight. It looks like the 1 gig asus has no sinks for the RAM either? o_O.. I got enough little sinks for the RAMs so I will be putting them on..

edit: One thing I noticed is that my VF900 uses a 3pin connector while the on board fan is a 4 pin connector. I'm just gonna attach it to my PSU and sacrifice the adjustible fan speed I guess..
 
So in a well vented case ramsinks are not needed on the VRM's? I would put them on the memory though.
 
Word of caution, I think the quality control on VF900s is poor. Some people get great temps, others do not ( like me). I tested the heatpipes by putting the sink on a hotplate with a thermal pad for contact and the pipes did not get hot as quickly as they should. Tried the same on my new Accelero and the pipes heated up within seconds.
 
zod96 said:
So in a well vented case ramsinks are not needed on the VRM's? I would put them on the memory though.
Click to expand...

GDDR3 has never needed RAM sinks. I think people use them because they look cool.
 
Wait you don't need to put any ramsinks on the memory? Or the VRM's?
 
So did anyone find out if we need for sure ramsinks on the VRM's? I do think that even adding a ramsink to them, would probably only drop them by a few degrees anyway. When I had my 4870, I tried using ramsinks with the vrm's and without. The difference was like 5 degrees at most.
 
my 4850's with a v900 load temp is 62C idle is ahh. 35-43C.. right now its 43C but it doesnt say its at 150mhz like it should be at idle it says 625 (not sure why dont really care)

i put the mc14 heatsinks on everything on the card.. it was a real pain too (had to cut them bend them..) and one of the vrm heatsinks just fell off and hit the bottom of my case ;/ so i guess ill just leave it there and not reapply that one seems fine.. took me forever to clean the bottom of the ramsinks i guess i could have skipped it all but makes me feel better at least and looks cool hehe
 
You must log in or register to reply here.
Back
Top