| mobyd |
Given that the biggest problem with some of the more powerful chipamps (LM4780, TDA7293 etc) is getting a lot of heat away from a smallish object in a hurry, has anyone tried any of the commonly available CPU coolers typically used on Pentium class processors. Some of these boast impressive specs (0.3degC/W)
but I wonder what their bulk heat moving capacity is really like.
M |
|
|
| pinkmouse |
Did you search before you posted? I think not! ;)
There is lots of discussion on the use of CPU coolers for chip amps here. |
|
|
| Bazukaz |
| Cpu cooler is good , especially with a fan used. a fan can be set to low rotation speed to decrease noise. |
|
|
| Vikash |
| quote: | Originally posted by mobyd
Given that the biggest problem with some of the more powerful chipamps (LM4780, TDA7293 etc) is getting a lot of heat away from a smallish object in a hurry |
Is it? I wouldn't say heatsink requirements are exceptional by any means. Many people use just large blocks of aluminium/copper or the case itself if it's of reasonable thickness.
As pinkmouse has said, you should find plenty of discussions on CPU coolers. IMO, certainly no need to add a fan into the mix. |
|
|
| Bazukaz |
| A fan dramatically increses power dissipation of the heatsink.For myself , i don't like fans for noise , but a thermo-relay could be used , so fan turns on only on high output levels. |
|
|
| jackinnj |
| you can use an LM3524 (SG3524) as a temperature related PWM fan speed controller -- National has an application note on their website -- |
|
|
| sagarverma |
i use a diminishingly small heatsink on my tda7294 amp(that small heatsink is called flower),the heat sink has small 12v dc brushless fan,the whole cooling setup only four or five times the size of ic..the setup keeps the ic cool....its small but very unsafe bec if the fan fails the chip will burn out in few sec bec the heat sink is too small to sustain the heat transfer even for few mins.
moreover the heatsinking req. of the chipamps is not that much critical to be met by using fans and it can be easily achieved by using big heatsinks. |
|
|
| Bazukaz |
| Most of modern ICs have thermal protection , don't know about TDAs , but LMs really do. |
|
|
| Leolabs |
| The old TDAs do not have a good one,although in their datasheets,their stated that they have.:( |
|
|
| zenon |
| DC fans can easily run at lower voltages then they are rated, for my PC I have two 12v fans wired in series so they run slower (6v across each) and the noise dropped from quite loud to inaudible. The noise all comes from the speed.. slow those fans down and they will be silent. |
|
|
| merlinx76 |
| My sub amp has 2 fans (which i slowed down) my source is my computer which has 2 fans, the computer on the bench behind me is even louder than mine, the fan for the dehumidifier in the next room is probably louder still. I despise fans... i want to be rid of them all... dont use fans :xeye: :xeye: |
|
|
| Vikash |
| quote: | Originally posted by merlinx76
i want to be rid of them all... dont use fans :xeye: :xeye: |
:up:
Why people want to use fans is beyond me. To each his own I guess. Just use a bigger heatsink and take the fan out of the equation unless size is that much of a constraint. What am I missing? Sounds to me like it's a case of doing it for the sake of doing it. I'll say no more on the matter ;) |
|
|
| Bazukaz |
| Fans are the option if you want to push the IC to its limits.With passive cooling it is hard to archieve less than 70 Degs Celsius(and sometimes even this) , if the amp is multi - channel and powerful.For myself , i don't like fans too.But they can be an option for high power applications. |
|
|
| Nordic |
| Alternatively drop the voltage to the chip until it has suitable power levels for use with whatever heatsink you have.... |
|
|
| EWorkshop1708 |
| But the hard part remains.................how to mount the chip to the CPU cooler......... |
|
|
| dnsey |
| quote: | | But the hard part remains.................how to mount the chip to the CPU cooler......... |
Not too difficult - use the double-sided adhesive pads or epoxy adhesives made for the job:) |
|
|
| karma |
or use a metal drill bit and drill and tap a hole to screw it down;)
makes it really tight:D |
|
|
| Bazukaz |
| quote: | Originally posted by EWorkshop1708
But the hard part remains.................how to mount the chip to the CPU cooler......... |
Drill a hole into the cooler and fix with a screw.Use mica insulator with thermal paste.
Silicone pads have poor thermal transfer. |
|
|
| triden |
sidenote:
What I did was make a temperature sensor applied to the heatsink that is combined with an opamp and comparator to turn a fan on when a certain temperature is reached. This way the fans only run when the half to. The circuit is fairly easy, but I havnt tuned it to operate best yet. |
|
|
| Nordic |
The aluminium part is very easy to drill with any metal drillbit.....
Also, try to get rubber grommets to attatch the fan to the cooler they kinda look like 2 chesspawns stacked bottom to bottom...
Even quiet fans transmits lots of vibration, which is of course easier to detect in a PC as the whole case resonates with it...
I thought my specialy imported, custom made Yate Loon fans were quiet, until I put them on grommets. |
|
|
| ! |
Fans are taboo because so few know how to implement them well. Common mistakes are made such as thinking use of a small fan is better, or a thin one, or trying to do some complex thing which is entirely unnecessary.
Basic tips for easy fan use:
- Keep RPM under 1,000, by combination of fan model and voltage level
- use a thick(er) fan whenever possible and at least 60mm, preferribly 80mm.
- Thermal control is not important. Proper fan implementation will be silent, and choosing a quality fan means it will run for 20 years. A fan can run at a fixed speed with no good reason to get more elaborate in most cases. Do not be deterred because someone tries to make things more difficult than they are, some "hypothetical" ideals are more hypothetical than realized, benefit. The one issue with fans is use in very dusty environments, particularly if used for whole-chassis cooling or with fine-tinned 'sinks, it will require a periodic maintenance (clean out dust).
- Yate Loon are often low quality fans. They require relubrication and are often so off-balance that they need special isolation. Many off-brand fans need this, but most good quality major brands don't. Wwhat any fan needs is an isolated power source to keep the noise off the chip power rails.
The benefits of a properly implemented fan are real, particularly when it is set up to move air through whole chassis, not just in/out of the 'sink. Only using it for the 'sink cooling is a bit of a waste of the secondary benefits.
- Lower chip temp
- Cooler running transformer
- Cooler running capacitors, diodes, etc. (lifespan increase)
- Smaller heatsink
- Smaller chassis
- Did somebody say Class A? More amp options easily attainable.
- Lower total amp cost
It's your amp, I'm not trying to convince anyone to use a fan but for all the very real benefits it is seen far too seldom, and so for a moment I took devil's advocate position even though my next amp probably won't have one. |
|
|
| Nordic |
| Don't forget that some of the world's top fans comes from the exact Yate Loon factory floor under other names... you get what you pay for. |
|
|
| dfdye |
| Just another note to add: DC fans can crank out a ton of noise due to brushes. If properly designed, your amp should be immune to this, but AC fans will minimize that noise since they don't have brushes (woohoo!). If you can use an AC fan, I would highly recommend you use these over DC fans in noise sensitive applications. Again, I am not saying that DC fan noise will goof up an amp, just that you can eliminate this noise source quite easily and avoid the possibility (and it means you don't have put in a separate DC psu just for fans) |
|
|
| Spasticteapot |
| quote: | Originally posted by dfdye
Just another note to add: DC fans can crank out a ton of noise due to brushes. If properly designed, your amp should be immune to this, but AC fans will minimize that noise since they don't have brushes (woohoo!). If you can use an AC fan, I would highly recommend you use these over DC fans in noise sensitive applications. Again, I am not saying that DC fan noise will goof up an amp, just that you can eliminate this noise source quite easily and avoid the possibility (and it means you don't have put in a separate DC psu just for fans) |
A wall-wart can be had for next to nothing these days, and it's often not hard to strip them down and attach them to the power transformer's primary.
I, however, offer a quieter, better-performing solution:
Watercooling.
It's been in use for years with PCs; why not give it a shot? A simple waterblock can be made out of a bit of copper scrap with some grooves cut into it; thin copper pipe makes for a dandy high-flow waterblock. |
|
|
| dfdye |
| quote: | Originally posted by Spasticteapot
I, however, offer a quieter, better-performing solution:
Watercooling. |
Overkill, but I agree it is really cool! (Edit: Sorry for the pun. :rolleyes: It was totally spontaneous and I didn't even notice it until I posted.)
Also, I was more referring to electrical noise generated by brushes than audible noise, just for the record. I think my post was a little ambiguous on that. Sorry for the confusion. |
|
|
| ! |
| quote: | | Don't forget that some of the world's top fans comes from the exact Yate Loon factory floor under other names... you get what you pay for. |
Which specific brands were you thinking of? Yate loon does not make Delta, NMB, Sunon, Panasonic, Papst, Nidec, Sanyo, Comair. These are the majority of the best names. It is sometimes true that you get what you pay for but the fan market is so varied, it's quite easy to get a top-quality fan as a surplus part for quite cheap.
| quote: | | DC fans can crank out a ton of noise due to brushes. If properly designed, your amp should be immune to this, but AC fans will minimize that noise since they don't have brushes (woohoo!). |
No typical DC fan you'd see [used in an amp, or a computer, or most any small fan without need for high-torque], has brushes. There is no need for AC fans and they might be a hard choice in some cases because the amount of control (easily obtained) over their speed is quite limited. If your use makes the default speed of an AC fan acceptible, it would do fine but for electrical or directly audible noise reduction/prevention they aren't necessary.
| quote: | | If you can use an AC fan, I would highly recommend you use these over DC fans in noise sensitive applications. Again, I am not saying that DC fan noise will goof up an amp, just that you can eliminate this noise source quite easily and avoid the possibility (and it means you don't have put in a separate DC psu just for fans) |
It's quite simple to isolate a fan and provide good control over it's speed simultaneously. Put it behind a linear regulator (such as LM317) in current-limiting mode. Current limiting mode is superior to voltage limiting mode due to the way the fan would pulsate with only voltage limitation, ie- it allows a lower quiet RPM IF you are trying for the lowest RPM possible. If you have another need then an appropriate filter can be used, per application, but the main thing is to never put one directly parallel to either (or both) of the supply rails used for other sensitive components. |
|
|
| dfdye |
| quote: | Originally posted by !
No typical DC fan you'd see [used in an amp, or a computer, or most any small fan without need for high-torque], has brushes. |
Touche. I guess I am still thinking old school.| quote: | | It's quite simple to isolate a fan and provide good control over it's speed simultaneously. . . |
Yea, but with an AC fan all you have to do is plug it into the power before the transformer. You should be able to buy the speed/size that you need before hand and avoid speed control issues altogether. I know you can use a celing fan speed control from the hardware store if you really have to adjust the speed of the fan, but I have no idea how noisy those are.
Anyway, I'm not trying to start a debate, just adding my suggestion. Besides, I hate fans! :D |
|
|
| ! |
| quote: | | Yea, but with an AC fan all you have to do is plug it into the power before the transformer. You should be able to buy the speed/size that you need before hand and avoid speed control issues altogether. |
In practice the issue is most often the speed control, that one often cannot choose (find) an AC fan that has sufficiently low RPM. Take an 80mm x 25mm fan, one of the most common sizes... where can you get one that is spec'd for 800 RPM? Key is implementation, that a fan can be used but to be used in such a way to avoid the most common detractions- noise, short(er) lifespan and shorter service intervals. Towards this end, getting a fan to turn as slowly as (reliably) possible is a good goal. AC fans just don't facilitate that.
| quote: | | I know you can use a celing fan speed control from the hardware store if you really have to adjust the speed of the fan, but I have no idea how noisy those are. |
There are many types of AC fans, with a ceiling fan you are starting with a different motor and a different RPM range. It's also bulky, more expensive than just throwing a 50 cent regulator plus 5 cent resistor in. Ac fans don't really have any benefits over DC fans in most low-flow situations, rather than the more obvious benefit that they're used where there is no DC supply at all to step down from. If you put an AC fan in a case, if you didnt isolate it somehow too, then you have that noise on the AC line local to the amp.
| quote: | | Anyway, I'm not trying to start a debate, just adding my suggestion. Besides, I hate fans! :D [/B] |
Normally I wouldn't mention fans at all, but in this thread, it only seems appropriate as they are on the CPU coolers and without a fan, a typical CPU cooler is a bit marginal in size unless the builder only wants a lower wattage chipamp. Funny how we always assume if a chip can do 60W then everyone "should" be designing towards 60W rather than the target use. How often do you hear anyone claim it might be ok to use a 50VA transformer and 2200uF caps? |
|
|
| dfdye |
| quote: | Originally posted by !
How often do you heard anyone claim it might be ok to use a 50VA transformer and 2200uF caps? |
Without revisiting the AC/DC fan issue (about which I am sure we can respectfully disagree), I coincidentally am using 2200uF caps in my LM3886's, Also conicidentally, I am powering this righ with two 80VAC torroids that I am probably going to drop to one if all tests work out right. But this is more than slightly off topic, so I will elaborate in other threads where appropriate! :D
Oh, and to get back on topic, I have both of my chips cooled by old school Pentium 133 passive heat sinks (you know the funky little blue ones?) They are more than adequate without any active cooling, even when running the amp full out. I am honestly not seeing why you are referring to these as being marginal. What CPU coolers are you using that aren't hacking it running passively? I figure if those old simple sinks can hack it than anything that is available today has to have a much higher capacity. If there is a design that you are using, have used, or have tested that is not sufficient, please let us know so that it can be avoided.
David |
|
|
| ! |
| quote: | Originally posted by dfdye
Oh, and to get back on topic, I have both of my chips cooled by old school Pentium 133 passive heat sinks (you know the funky little blue ones?) They are more than adequate without any active cooling, even when running the amp full out. I am honestly not seeing why you are referring to these as being marginal. What CPU coolers are you using that aren't hacking it running passively? I figure if those old simple sinks can hack it than anything that is available today has to have a much higher capacity. If there is a design that you are using, have used, or have tested that is not sufficient, please let us know so that it can be avoided.
David |
There are a lot of funky blue heatsinks out there in the wild, I'd have to see a picture of the 'sink with something common beside it for scale. Even then, It goes back to what I wrote about output, that some aren't going to run at near 1/3 output. If a 'sink is designed to be passive, it is an exception in today's CPU 'sinks and may be more appropriate (if large enough), since it was originally designed for the task, tends to have lower density fin/pin grouping. Where it is mounted (entirely internal to a case with low passive ventilation perhaps) is another variable. Even whether the chip uses the isolated casing and the 'sink can be mechanically 'sunk to the metal case/suppport or must be further isolated from chassis ground instead, can matter.
How hot do you want to let a chip get if it can be avoided? A bit of subjectivism there since some take the textbook "it says X degrees so I'm under that" and others hit some lower figure whether valid or arbitrary, or at least think they do but I don't recall too many people reporting back temp measurements taken of their chipamp cased, in use, rather than only what it sounds like. Most electronics that fail early due to heat, will of course be working until they fail else they wouldn't have been turned on for the use. Even a low quality 20 year old amp might work today if it's failure points are addressed but until a design has aged, it is more difficult to know if a $200 (random figure) amp would work 5 years or 35.
Essentially too many variables for a cut-n-dry "cpu 'sink passive is ok" declaration, rather than ignoring the 'sink was from a computer and just looking at it's C/W as implemented. Each builder can do it themselves and decide. |
|
|
| dfdye |
| quote: | Originally posted by !
How hot do you want to let a chip get if it can be avoided? |
The manufacturer data sheet is always a good place to start. (hint: P17 of the LM3886 sheet is almost entirely dedicated to the subject. Here's the URL to get you started. http://www.national.com/search/sear...keywords=lm3886 )
BTW, pictures of the dinky blue heat sink in comparison to some household objects are on their way. Also, even with tight fin spacing, modern CPU heat sinks work rather well as passive coolers as long as the fins are vertical and convection is allowed to do its thing (Yea Physics!!)
As always, YMMV, but chipamps are really rather easy to cool. Get a big hunk of aluminum and you are 9/10ths of the way there. . . .
Why are we arguing about this again? :scratch: If you get thermal shutdown, get a better sink. If not. . .
Still waiting for the sinks that you have used that don't work. Pictures aren't really necessiary, just descriptions will be fine. |
|
|
| ! |
Of course, a place to start but never the final answer as they are always max values, (typically) towards a guaranteed lifetime projection that is all too often, far far below the desired lifetime of the device. There is no way I'd settle for near 165C on an LM3866, nor anywhere near the max datasheet temp for any other parts in the case such as caps, diodes, transformer, etc.
| quote: | BTW, pictures of the dinky blue heat sink in comparison to some household objects are on their way. Also, even with tight fin spacing, modern CPU heat sinks work rather well as passive coolers as long as the fins are vertical and convection is allowed to do its thing (Yea Physics!!)
As always, YMMV, but chipamps are really rather easy to cool. Get a big hunk of aluminum and you are 9/10ths of the way there. . . . |
Big hunk, yes, especially when mounted to more metal. Relatively speaking they are not hard to cool, but like any easily cooled device there is still the possiblity to choose an insufficient 'sink that results in hotter running, shorter life.
| quote: | | Why are we arguing about this again? :scratch: If you get thermal shutdown, get a better sink. If not. . . |
This is where we disagree. I'm not interested in running any device so hot it (has potential to) shuts down to save itself from destruction. If there is any remote doubt I measure the temp. Of course the die temp cannot be taken so the casing thermal figures are used to inflate the extra-case temp.
| quote: | | Still waiting for the sinks that you have used that don't work. Pictures aren't really necessiary, just descriptions will be fine. |
Nothing fancy, they were typical midsized socket 7 'sinks. Chip was not shutting down to prevent self destruction, but I didn't "try" to force that with increased volume either. We just have completely different perspectives on heat removal, I don't "redline" anything unless it's a real drag race. |
|
|
| Variac |
| more interesting IMHO is whether you could cool a large MOSFET with a CPU cooler. Then you could truly make a mini Aleph! |
|
|
| Spasticteapot |
| quote: | Originally posted by Variac
more interesting IMHO is whether you could cool a large MOSFET with a CPU cooler. Then you could truly make a mini Aleph! |
You can dissipate over 150 watts of heat with a basic-model waterblock.
On the topic of CPU heatsinks:
1. These things dissipate TONS of heat. A Pentium-4 heatsink, which usually has tons of fins and is made of copper, will cool a gainclone just fine. They dissipate huge amounts of heat. (Heck, Zalman even makes passive CPU coolers.)
2. Get a good one, if you can. A Thermalright XP-120, while pricey, can dissipate very large amounts of heat when actively cooled; Zalman HSF's (get one for Socket 370 or 423, so it'll be cheap) do a nice job passively.
3. Use a fan. A good (120$) 120mm Panaflo brushless dual-ball-bearing fan running at 7.2 volts produces nearly no noise, and lots of airflow.
4. Use a discrete (wall-wart?) power supply.
Stop being ninnies. Just find something between 7 and 9 volts with 200ma or more of current. Dead easy.
5. Watercooling is easy.
All you need to do, really, is cut some channels into a bit of copper sheet about 1cm thick. Cover it with a piece of polycarbonate or plexiglass to which you attach the in and out barbs, and use a gasket cut out of some gasket rubber from the hardware store. Then, wire up an aquarium pump (10$ for a cheap Via Aqua), a small plastic tub for a resivoir, and some copper tubing or an old heatercore from a car as a radiator. It's not hard to do.
6. Submersive cooling:
Canola oil conducts electricty almost as badly as air. Just fill your case with the stuff, and throw in a small pump and some copper fins for heat dissipation. Presto! Noiseless, cheap liquid cooling! (Plus, you never have to worry about leaks).
7. Heatpipes:
For all you ninnies who are afraid of filling your computers with water or cooking oil, heatpipe-based CPU coolers work very nicely at dissipating heat. The trick is finding one.
8. Zalman heatpipe cases:
Although I'm not sure if they're made anymore, Zalman made a case which could passively cool a pentium 4 based system by using the case as a gargantuan heatsink. You could cool about a dozen gainclones with ease if you spaced 'em out right.
I hate to say this, but y'alls are WAY behind the world of computing. Small (computer) heatsinks can dissipate huge amounts of heat. |
|
|
| Variac |
canola oil doesn't conduct heat close to as well as water does..not even close.....
Its easier than you state to make a waterblock. Grey Rollins cools his Aleph with his MOSFETs bolted to a 1/8" copper plate to which he has soldered a length of 1/2" home copper pipe. Then pumps water through the tube to the radiator.
The problem in a computer is that there is a cpu chip putting out a moderate amount heat from a very small area. It needs to be moved as efficiently as possibly away from the chip. This is different that a big audio amp which generally is giving off a lot more heat but from a less concentrated source.
In the case of a chip amp , the amp is running in class AB which doesn't give off that much heat... A reasonably sized heatsink easily can handle that amount of heat, so why complicate things with a fan or with water cooling? Possibly because you need a powerful chip amp in a very small case, but that's not normally needed. So I think thats a big point that people are making here. Water cooling is just silly in this case- unless you want to because well, you want to !- which is just fine. In fact the image of 4 square CPU heatsinks arrangened to look like a 4 barrel carb appeals to me! :D
In an amp generally the heat isn't as concentrated in such a small area, the noise of a fan and a pump are more of an annoyance, because, the point of amp is good sound with a decent noise floor.
So I dispute that the computer folks are so advanced, they just have different requirements, and noise isn't generally as much of an issue.
So now , if you want to discuss a large class A amp like a Pass Labs Aleph then you have cooling problems. They give off hundreds of watts of heat, really, like 600 watts! and just the heatsinks can cost hundreds of dollars.
With them fan and or water cooling could be a big help. With water cooling it is important to remember that once you have the heat in the water you have to get it out of the water. For this you need a radiator- a radiator with at least as much fin area as you would have if you just had fins on the amp itself. The cooling setup for a computer isn't even close to what is needed.
Soo, it could work if you have a place for a 2' by 2' radiator, and the radiator, tubes, pump and maybe a fan for the radiator all cost less than the couple of hundred the fins would cost. And the fan isn't noisy. The advantage is mostly that is is really cheap to get a radiator in a junkyard, and the block can be cheaply made as I mentioned above..and maybe a quiet pump is available at a surplus place.
If you make the radiator big enough, and the tubes also, and mount the radiator above the amp, then the hot water wll rise into the radiator without a pump and then circulate down when it cools, a thermo-syphon. With a big radiator and enough ventilation the fan might not be needed .
But you either hang the radiator above your amp (only if you aren't married) or hide it in a box which probably requires a fan, so you still have an incredible amount of space wasted- maybe more than the heatsinks would have used if you had just used them....
|
|
|
| Spasticteapot |
| quote: | Originally posted by Variac
canola oil doesn't conduct heat close to as well as water does..not even close.....
Its easier than you state to make a waterblock. Grey Rollins cools his Aleph with his MOSFETs bolted to a 1/8" copper plate to which he has soldered a length of 1/2" home copper pipe. Then pumps water through the tube to the radiator.
The problem in a computer is that there is a cpu chip putting out a moderate amount heat from a very small area. It needs to be moved as efficiently as possibly away from the chip. This is different that a big audio amp which generally is giving off a lot more heat but from a less concentrated source.
In the case of a chip amp , the amp is running in class AB which doesn't give off that much heat... A reasonably sized heatsink easily can handle that amount of heat, so why complicate things with a fan or with water cooling? Possibly because you need a powerful chip amp in a very small case, but that's not normally needed. So I think thats a big point that people are making here. Water cooling is just silly in this case- unless you want to because well, you want to !- which is just fine. In fact the image of 4 square CPU heatsinks arrangened to look like a 4 barrel carb appeals to me! :D
In an amp generally the heat isn't as concentrated in such a small area, the noise of a fan and a pump are more of an annoyance, because, the point of amp is good sound with a decent noise floor.
So I dispute that the computer folks are so advanced, they just have different requirements, and noise isn't generally as much of an issue.
So now , if you want to discuss a large class A amp like a Pass Labs Aleph then you have cooling problems. They give off hundreds of watts of heat, really, like 600 watts! and just the heatsinks can cost hundreds of dollars.
With them fan and or water cooling could be a big help. With water cooling it is important to remember that once you have the heat in the water you have to get it out of the water. For this you need a radiator- a radiator with at least as much fin area as you would have if you just had fins on the amp itself. The cooling setup for a computer isn't even close to what is needed.
Soo, it could work if you have a place for a 2' by 2' radiator, and the radiator, tubes, pump and maybe a fan for the radiator all cost less than the couple of hundred the fins would cost. And the fan isn't noisy. The advantage is mostly that is is really cheap to get a radiator in a junkyard, and the block can be cheaply made as I mentioned above..and maybe a quiet pump is available at a surplus place.
If you make the radiator big enough, and the tubes also, and mount the radiator above the amp, then the hot water wll rise into the radiator without a pump and then circulate down when it cools, a thermo-syphon. With a big radiator and enough ventilation the fan might not be needed .
But you either hang the radiator above your amp (only if you aren't married) or hide it in a box which probably requires a fan, so you still have an incredible amount of space wasted- maybe more than the heatsinks would have used if you had just used them....
|
First, I was suggesting the waterblock for a class-A amp. Those put out tons of heat.
Secondly, the "copper tube" method is much less efficient than using channels in the copper itself. Look at Danger Den's or Little River's waterblocks; they all work this way. (That, and they're easier to make; all you really need is a drill press and some time.)
Secondly, radiators don't need to be as large as you think. An actively cooled 8" radiator can dissipate the heat produced by a dual-core Opteron processor; I've seen watercooling setups crammed into Shuttle XPC's the size of toasters.
Lastly, the "oil submersive cooling" attempt works very well for one simple reason: You've got so darned much of the stuff. Having copper strips with half in the oil and half out will dissipate a lot of heat. (I've seen top-of-the-line systems cooled in this manner with NO FANS AT ALL, including on the graphics card.)
Also, I've yet to see a good "thermal siphon" watercooling system. A cheapie pump is the way to go. |
|
|
| Variac |
Clearly we agree that this makes the most sense for Class A
I'm pretty sure the method I mentioned will be easier for some to make waterblocks. Perhaps your way will work better for others. Again I am only quoting someone, but he has in fact made waterblocks in this way and has used them and they work very well on an Aleph class A amp by his report.
Ok an 8" square radiator WITH a fan can cool a dual core CPU
A dual core CPU doesn't put out anywhere near 600 watts obviously, 150 watts max I'd think, correct me if I'm wrong...Then we're talking about a 16" x 16" radiator AND still using a fan for 600 watts.
The point is that you can move heat around, but at some point you havee to get rid of it. Insteat of a radiator I have heard of people using buried pipe or water from a fishpond.
A case filled with oil will need to radiate the heat to not eventually get hot. In the situation you described, the outer case is acting like a large heat sink, Admittedly if the computer is used for a limited time then the oil would have time to cool before overheating while it is switched off.
|
|
|
| Spasticteapot |
| quote: | Originally posted by Variac
Clearly we agree that this makes the most sense for Class A
A case filled with oil will need to radiate the heat to not eventually get hot. In the situation you described, the outer case is acting like a large heat sink, Admittedly if the computer is used for a limited time then the oil would have time to cool before overheating while it is switched off.
|
That's what the copper fins are for; they'd act as a radiator.
Secondly, although CPUs only put out about 150 watts maximum, you must consider that you can actually disperse a lot more heat if the temperature difference is greater; a computer CPU will often run at 45 degrees tops, while a class-A amp can do a minimum of 60. Also, it's traditional to use a heatercore from an old car (VW's are supposed to work nicely); these dissipate monstrous amounts of heat, and are cheap enough that two can be used.
An intercooler or chiller is a possible option, too. Peltiers, providing you don't saturate 'em (read: use MASSIVE waterblocks) can help dissipate more heat; a chiller on the resivoir can also help. |
|
|
| Nordic |
Being quite a PC modding fan, I have done a fair bit of reading into the topic, and basicaly its not just the rate at which water can absorb heat but also the rate at which it will transfer it, thats important. I saw a number of side by side tests with difirent liquids, and waters wins outright.
PC modding is a great source for nicely packaged small radiator if you do feel that way inclined, alternativly hit a scrapyard and get a Camry's heater radiator. |
|
|
| ! |
Humans, it seems, are nothing if not ironic. On the one hand we have the limits of passive cooling, and on the other people contemplating oil and radiators. I'm sure they could be done well, too, but a lot of unnecessary expense and addt'l size. Within the expected heat production of (most) chipamps, we reach a point where it requires very large passsive 'sinks for high output, or a fan, but never is the heat generation so high as to need anything more unless you have an extreme environment. That includes Class A. I'm sure if someone tried to make some odd thing, 5 bridged output in an old shoe, there'd be a need for something special to cool it but I'm about to get back to a point I made earlier;
The benefit of the fan is not just to cool the chip more, it's to cool the whole amp case some. Submersing the whole amp in oil doesn't seem like a real solution as much as a novelty and while water-cooling can be an interesting hobby, who is going to water cool the rest of their amp? The fan would need be pushing or pulling on the case frame for best benefit and if that is done the radiator is mostly just an impedance to the flow rate and another fine-finned thing for dust to get hung up on. Fans can be cheap, easy to implement, and pretty low maintenance or they can instead become an undesirable burden. The best fan implemenation is one where you forget there's even a fan in the amp, except perhaps to swap a filter panel every couple years or so. |
|
|
| Nordic |
Well considering the lenghts fanatics go to to isolate their amps from vibration, I reccon a fan is not the way to go....and I guess even watercooling would add quite a bit of vibration due to the flow of liquid in the pipes....
I tried using a fan with my first GC, gave up eventualy as I could always tell by the sound if the fan was on. |
|
|
| tiltedhalo |
| Then the fan was not properly implemented. |
|
|
| Bazukaz |
| quote: | Originally posted by Nordic
Well considering the lenghts fanatics go to to isolate their amps from vibration, I reccon a fan is not the way to go....and I guess even watercooling would add quite a bit of vibration due to the flow of liquid in the pipes....
I tried using a fan with my first GC, gave up eventualy as I could always tell by the sound if the fan was on. |
Please explain , why is vibration an issue with amplifiers ? |
|
|
| lbreevesii |
| look at the design of arctic cooling brand cpu coolers, particuarly the one designed for the athlon64 bit chips. It(among others) mount the fan with rubber grommet peices to isolate vibration. Look around on sites like silent-pc for links to makers of vibration isolating grommets and the like. |
|
|
| Bazukaz |
| Running the fan at lower voltage (6-8 volts) would decrease fan noise by much. |
|
|
| lbreevesii |
That it will. Keep in mind the cfm airflow at full speed as it relates to how much cooling you need though. Slow it down by too much and you won't get enough airflow to do the job.
Also keep in mind that different fans are better made(and thus quieter) than others. for pc use 30db is considered somewhat high but is still not very noticable. 25 is fairly low, and 20-23 is very low. |
|
|
| Bazukaz |
| Ball bearing fans should be quiter than sleeve bearing ones. |
|
|
| lbreevesii |
generally they are. There are also ceramic and even better, maglev fans on the market.
IMHO the best fans around are made by a company called acousti-fan. They ring in at 21db and 25cfm of air for the 80mm size(standard pc case fan size).
They also have another series of fan that pushes 20cfm at a virtually silent 15db. |
|
|
| Nordic |
| If people are putting amps on spikes just to protect it from the soundwave's vibration's I think short of magnetic levitation there is no way you can imlpement a fan not to impart vibration several orders of magnitude more than what you'd expect to bounce back from the wall. |
|
|
| ! |
Much misleading infos accumulating.
Note that Nordic wrote "lengths FANATICS go to to isolate their amps". Tubes are very important to isolate. Earthquakes are too. The amp builder that puts cone feet on an amp for any reason other than it's looks, is 100% delusional about it helping reduce vibrations. It has been scientifically proven beyond a shadow of a doubt that this does not reduce vibrations at all. Now, "IF" those cone feet were set atop a softer surface like rubber, or maybe unfinished pine, balsa, etc, to the extent that the surface under the cone points was significantly flexing, it would then help. Problem is, seldom is this done in conjunction, instead they are used as if magic bullets on regular hard surfaces. Not that it's a good practice to have excessive vibrations in any gear, but it is relative.
Acrtic Cooling has to use isolators because they use such **** fans. Reducing fan vibration is, first and foremost, always choosing a well balanced fan. In particular some things increase the odds of that- choosing a top name brand, not a thin fan (over 15mm thick), and not one of those very rigid frames (like the clear LED types). Running the fan at the optimal (lowest reliable) voltage does a LOT to further reduce any minor vibrations that remain though they might not have been significant enough already.
One need not keep in mind the cfm airflow, quite the opposite. You WANT to slow it down the maximum amount possible. These are not CPUs, they don't have 100W/cm² and only making a poor choice of heatsink (metal portion) or fan (too small/thin) will require unduely raising the RPM. 30db of actual noise is not so noticable, but a fan rated as 30db is rated as such in "free air", not implemented in a chassis where turbulence has a very significant noise production.
For our uses, there are very very few (if any) fans that start out with low enough noise levels. A fan rated as 22db free air, will be too loud for most discriminating listeners and needs voltage or current reduction towards that end.
Ball bearing fans are not quieter. There are some very poor quality sleeve bearing fans that wear down fast from their bearing and lack of lubrication but considering only good quality fans (all else equal except the bearing type), sleeve bearing fans are quieter. There are so few though that are very good, most will require lubrication every few years, less often with lower RPM.
Ceramic and maglev are just marketing tricks. No better than good quality fans, perhaps better than the lowest quality fans but that's not a quality level one wants to compete in. Their designs are meant to maximize profit, not result in a good fan.
Fan tech has been mature for over 50 years with minor exceptions such as synthetic lubricants, lower noise ICs, and better metal fabrication for the balls & races. There is no new special fan that is as good as a top brand (hence why they are the top brands) of majar fan manufacturer's products, not fan relabelers who just throw their own label on and hype marketing buzzwords. |
|
|
| lbreevesii |
What matters is the quality of the bearings the resulting airflow and noise levels.
AC does not use S*$@ fans. Do you have experiece with them personally? I own multiple cpu AND gpu coolers from them and they run wonderfuly and quietly. They use the rubber mountings to reduce extra vibration. Mabye thats why I can't hear it running. As a matter of fact their gpu's are incredible. Their bearings are some of the smoothest i've seen.
the thing is that in my experience its mostly good quality fans that are using ball bearing fans, and decent quality, but cheaper fans using sleve fans.
Another reason why I recommended acousti-fan is that they not only include a connector with a resistor to cut the voltage, but they also have a thermocouple to further reduce fan speeds when they are not needed. |
|
|
| ! |
| quote: | Originally posted by lbreevesii
What matters is the quality of the bearings the resulting airflow and noise levels.
AC does not use S*$@ fans. Do you have experiece with them personally? I own multiple cpu AND gpu coolers from them and they run wonderfuly and quietly. |
Yes I do, and no they are not good fans. Average, perhaps. Even their MTBF ratings are lower than any decent brand. You write about quality of bearings but there's nothing about their bearing particularly noteworthy, they're just commodity grade fans.
They can run fine when new- just as any cheap fan can. Remember that this particular use (chipamps) is not same thing as CPU cooling since a CPU will be technologically old in a few years but a good amp should be expected to have a lifespan of a few decades. Even so, better quality fans than the AC are expected to last over 1 decade easily, but theirs, maybe 4 years if you're lucky. They have a warranty for 6 years on some but they are taking advantage of something with it- that not a lot of people are going to pay as much shipping and the time to replace one fan under warranty when a new fan can be delivered for near same cost and no time spent on it.
| quote: | | They use the rubber mountings to reduce extra vibration. Mabye thats why I can't hear it running. As a matter of fact their gpu's are incredible. Their bearings are some of the smoothest i've seen. |
Nonsense, there is absolutely nothing different about their bearings. That "extra vibration" is due to their poor balance. Take off the rubber and run that fan at same RPM as a better quality fan and you'll see for yourself that it is true. AC sells an "image" of low noise, and takes median quality fans then runs them at low RPM. That's fine really, it'll work but better to just use a good quality fan at low RPM instead and it will be quieter and longer lasting as even AC admit when they publish lower MTBF ratings for their product.
| quote: | | the thing is that in my experience its mostly good quality fans that are using ball bearing fans, and decent quality, but cheaper fans using sleve fans. |
Often true, except that there are some quality sleeve bearing fans too, particularly those by Papst, Panaflo, (a few) Sunons, Comair. Far too many sleeve bearing fans were just cheaper to make though, and guess what? Some of AC's fans are those cheap sleeve bearing fans.
| quote: | | Another reason why I recommended acousti-fan is that they not only include a connector with a resistor to cut the voltage, but they also have a thermocouple to further reduce fan speeds when they are not needed. |
The key is to start out with the BEST quality fan instead. Anything else is just going to cause more problems in the long run. Further, even with the fans having inline resistor and thermal control, it's threshold/response is too high for a chipamp. Lower heat density and ability to use a larger heatsink in the chipamp means a lower fan RPM is possible. |
|
|
| lbreevesii |
| Have you delt with their gpu coolers? no rubber mounting there and they run smoother than butter. |
|
|
| dfdye |
| I just remembered that I promised pictures of what I was using so that we could all be on the same page. Here is a picture of one of the heatsinks I use in front of my laptop and with a coin for size comparison. I use one of these per chip and I haven't had any problems so far. |
|
|
| dfdye |
Here is what I do to them before I mount them on the chips! :D (different sink)
[edit] Forgot to mention that these will probably be lapped pretty well before use just cause I always do that sort of thing. Not that it will really make a difference since I have to use insulation pads anyway. . . :D [/quote] |
|
|
| dfdye |
| Here is the latest batch that I have prepped for future use. The big one will probably use 2 chips in a stereo amp. The holes were pre-drilled for mounting to the CPU, but the are spaced and sized perfectly to mount two of my Linkwitz Lm3886 boards. I was thinking of making an iPod amp, and this would be perfect. |
|
|
| noodle_snacks |
| quote: | Originally posted by lbreevesii
AC does not use S*$@ fans. Do you have experiece with them personally? I own multiple cpu AND gpu coolers from them and they run wonderfuly and quietly. They use the rubber mountings to reduce extra vibration. Mabye thats why I can't hear it running. As a matter of fact their gpu's are incredible. Their bearings are some of the smoothest i've seen.
|
I have an AC graphics card cooler. It is actually fairly noisy compared to the rest of my case and makes a distinctive whirring sound. Of late the bearings have started to go, and i have only had it for about 3 months, the next video card i get will hopefully have passive cooling. |
|
|
| ! |
| quote: | Originally posted by noodle_snacks
I have an AC graphics card cooler. It is actually fairly noisy compared to the rest of my case and makes a distinctive whirring sound. Of late the bearings have started to go, and i have only had it for about 3 months, the next video card i get will hopefully have passive cooling. |
you might be able to quiet it some by lubing it, and if the case is well ventilated, reducing it's RPM some but of course then temp measurements of the card are evermore important.
| quote: | originally posted by dfdye
I just remembered that I promised pictures of what I was using so that we could all be on the same page. Here is a picture of one of the heatsinks I use in front of my laptop and with a coin for size comparison. I use one of these per chip and I haven't had any problems so far. |
Choosing the appropriate heatsink can vary per application. If your amp stays cool enough, great. That 'sink is not big enough for the maximum possible output of a (LM3886 for example) chipamp. It is not necessary to design for max power and certainly nobody is compelling you to go deaf listening to one at that level, but the size 'sink that works for one person may be not enough or the opposite, overkill for someone else.
The spec sheet or app notes (I forget which) for LM3886 or Overture amps lists some possible output levels, and they reach a few dozen watts and specify heatsink C/W ratings which are lower (better) than the 'sink you show, enough different that I don't need to know the exact 'sink spec because it's not even close to their recommendations... but again, their recommendations are going to be for max possible heat at rated wattages of the chips which the amp may not even physically be able to produce in a given configuration even if cranked up 1/3 volume continually which it probably isn't. |
|
|
|
