Don't know if this works and hope no one laughs, here is a basic concept.
An externally hosted image should be here but it was not working when we last tested it.
What devices are you looking at, as they may be available in newer more thermally efficient packages?
For FETs etc there is a variety of new packages designed for heat removal from the die, for IC's and drivers QFN packages with thermal pads are becoming more prevalent.
Unfortunately the newer thermal packages are mostly SMD. The point of these packages is to create the lowest thermal resistance between the die and the outside world where the heat can be removed.
PCB layout and orientation when completed can have a big influence on the thermal requirements of a design.
International Rectifier - The Power Management Leader - DirectFET Home Page
TI DualCool™ NexFET™ Power MOSFET Technology - TI.com
http://www.fairchildsemi.com/collateral/new/100027-001_DualCool_Packaging_Product_Overview.pdf
http://www.cirrus.com/en/pubs/appNote/AN315REV1.pdf
PCB Thermal stuff.
EMS007 Designing In Thermal Management
http://pcdandf.com/cms/images/stories/mag/0608/0608pcdm_digital.pdf
For FETs etc there is a variety of new packages designed for heat removal from the die, for IC's and drivers QFN packages with thermal pads are becoming more prevalent.
Unfortunately the newer thermal packages are mostly SMD. The point of these packages is to create the lowest thermal resistance between the die and the outside world where the heat can be removed.
PCB layout and orientation when completed can have a big influence on the thermal requirements of a design.
International Rectifier - The Power Management Leader - DirectFET Home Page
TI DualCool™ NexFET™ Power MOSFET Technology - TI.com
http://www.fairchildsemi.com/collateral/new/100027-001_DualCool_Packaging_Product_Overview.pdf
http://www.cirrus.com/en/pubs/appNote/AN315REV1.pdf
PCB Thermal stuff.
EMS007 Designing In Thermal Management
http://pcdandf.com/cms/images/stories/mag/0608/0608pcdm_digital.pdf
Why ???.
Removing encapsulation will render the device photo sensitive and liable to damage to the bonding wires.
Just mount the standard device directly to a decent slab of copper which is then thermally coupled but electrically isolated from a standard large heatsink.
Fan cool as an extra if you have to.
Be sure to lap all surfaces before assembly to ensure best thermal coupling.
Eric.
Dumb proposition from the outset.
Removing encapsulation will render the device photo sensitive and liable to damage to the bonding wires.
Just mount the standard device directly to a decent slab of copper which is then thermally coupled but electrically isolated from a standard large heatsink.
Fan cool as an extra if you have to.
Be sure to lap all surfaces before assembly to ensure best thermal coupling.
Eric.
Just the other day I viewed a video (on YouTube?) where I guy was making a demo heat pipe from soldered copper tubing with a bottom reservoir partially filled with acetone. This should make a very nice phase change medium, because the boiling point is 134F (56.5C). The acetone won't heat above the boiling point until all of it has boiled away, and this provides a nice constant temperature sink for heat (phase changes suck up a lot of heat).
The problem with something that uses a lower boiling temperature fluid is that (a) these are not widely available to the public and (b) you will need a refrigeration system to cool down the condensing end of the heat pipe because indoor summer room temperature will be too close to, or above, the boiling point!
You could build a completely passive system using acetone, place something to transfer heat out of that system (like a heat sink) in to the environment, making sure to locate that where the acetone vapor will rise to.
-Charlie
The problem with something that uses a lower boiling temperature fluid is that (a) these are not widely available to the public and (b) you will need a refrigeration system to cool down the condensing end of the heat pipe because indoor summer room temperature will be too close to, or above, the boiling point!
You could build a completely passive system using acetone, place something to transfer heat out of that system (like a heat sink) in to the environment, making sure to locate that where the acetone vapor will rise to.
-Charlie
@marce - I have built numerous custom cooling solutions for PCs, most of them passive or semi-passive - including a fluid cooling system, and a sub-zero system. You obviously know your stuff and I see your points - but this is about an audiophile device, which means that numerous factors come into play, that are normally not a consideration - I listed them in my previous post. By vibration I meant vibration on microscopic level - either generated within the device, or carried through air/floor from the "outside world" (the speakers, most importantly).
The chip in question is the legendary LM3875; it does have a newer version (LM3886), but it comes in the same package, and I prefer the older one anyway
I like the idea of submersion cooling, but I think it may be difficult to find a fluid that: is neither too conductive nor too insulating, has both good heat conducting properties and low boiling point. And is reasonably accessible, stable, non-toxic and safe for handling and for the environment The best I could think of is 96% ethanol/4% water. This would be very weakly insulating, have reasonable cooling properties (worse than water), and a reasonable boiling point of 78C. It is also relatively harmless, although flammable.
I considered acetone, and to be honest, I would much rather use methyl acetate instead, because it smells a lot nicer Both have similar physical properties otherwise - including, unfortunately, very high flammability.
Phase-change coolant, while very efficient, presents a problem of variable pressure. An elegant answer to that would to leave some empty space in the container and evacuate air from it. This would reduce pressure, boiling point and the risk of explosion.
I'm waiting for a reply from the company that makes them If they could do a sample/short run for me at an acceptable price, I'll definitely go for it. It's not that expensive, apparently. I mean, we are talking tens of dollars for a small one, not thousands.
The chip in question is the legendary LM3875; it does have a newer version (LM3886), but it comes in the same package, and I prefer the older one anyway
I like the idea of submersion cooling, but I think it may be difficult to find a fluid that: is neither too conductive nor too insulating, has both good heat conducting properties and low boiling point. And is reasonably accessible, stable, non-toxic and safe for handling and for the environment
The best I could think of is 96% ethanol/4% water. This would be very weakly insulating, have reasonable cooling properties (worse than water), and a reasonable boiling point of 78C. It is also relatively harmless, although flammable.I considered acetone, and to be honest, I would much rather use methyl acetate instead, because it smells a lot nicer
Both have similar physical properties otherwise - including, unfortunately, very high flammability.Phase-change coolant, while very efficient, presents a problem of variable pressure. An elegant answer to that would to leave some empty space in the container and evacuate air from it. This would reduce pressure, boiling point and the risk of explosion
.Diamond heatsinks actually do exist (diamond composite, to be precise); they are just not very widely available. In fact, this is a part of the mysterious heat-sinking solution I mentioned in post#1
I'm waiting for a reply from the company that makes them If they could do a sample/short run for me at an acceptable price, I'll definitely go for it. It's not that expensive, apparently. I mean, we are talking tens of dollars for a small one, not thousands.
Last edited:
How about graphite? Should be cheaper than diamond, however more fragile. I was looking for different way to heat-sink transistors of a Class A amp. I've settled on heat-pipes.
There do exist graphite heatsinks (actual heatsinks; the diamond ones I mentioned above are more like heat-spreaders really). Apparently they have pretty decent thermal performance. Graphite is still somewhat electrically conductive, but nowhere near as conductive as copper, so it should be a good choice. Unfortunately, I haven't yet succeeded in finding a company that actually sells them...
Your amp looks kinda cool actually with all them coolers Although LM3875 dissipates relatively little power (some 30W p.c. at loudest setting) and normally only requires a moderate passive heatsink.
Your amp looks kinda cool actually with all them coolers
Although LM3875 dissipates relatively little power (some 30W p.c. at loudest setting) and normally only requires a moderate passive heatsink.
Last edited:
Silver is the highest heat conductivity metal, but the base plate of the chip is still just copper.
you also still have the interface between the chip and the metal.
One thing that comes to mind is low temperature soldering, using a metal similar to bismuth.
You solder your amp chip to a plate of silver, and then cool that silver plate with liquid or thermoelectric cooling.
This letting the chip vibrate itself sounds a bit bogus though, along the lines of wooden volume knobs.
There is no way to get good thermal conductivity without electrical conductivity. Diamond is your only material thats magic like that.
you also still have the interface between the chip and the metal.
One thing that comes to mind is low temperature soldering, using a metal similar to bismuth.
You solder your amp chip to a plate of silver, and then cool that silver plate with liquid or thermoelectric cooling.
This letting the chip vibrate itself sounds a bit bogus though, along the lines of wooden volume knobs.
There is no way to get good thermal conductivity without electrical conductivity. Diamond is your only material thats magic like that.
Last edited:
Beryllium Oxide
You have apparently overlooked Beryllium oxide (BeO).
Among non-metallic materials, BeO is second only to diamond in thermal conductivity, with k approaching 280 W/m/K. By comparison, metallic aluminum has a k of only 240 W/m/K, and copper slightly over 400 W/m/K. It is also a superb electrical insulator exhibiting vanishingly small dissipation factor, which makes it especially useful at gighertz frequencies. BeO is most commonly used for critical aerospace / military applications where electrical, thermal and mechanical performance are of paramount importance.
These BeO insulators would be perfect for an LM1875 or LM3886 based chip amp:
629879-01 BeO - 1.4" x 1" x 0.064" Beryllium Oxide Insulator
Not exactly inexpensive, but combined with a copper heat sink and appropriate thermal compound, they would be extremely effective.
You have apparently overlooked Beryllium oxide (BeO).
Among non-metallic materials, BeO is second only to diamond in thermal conductivity, with k approaching 280 W/m/K. By comparison, metallic aluminum has a k of only 240 W/m/K, and copper slightly over 400 W/m/K. It is also a superb electrical insulator exhibiting vanishingly small dissipation factor, which makes it especially useful at gighertz frequencies. BeO is most commonly used for critical aerospace / military applications where electrical, thermal and mechanical performance are of paramount importance.
These BeO insulators would be perfect for an LM1875 or LM3886 based chip amp:
629879-01 BeO - 1.4" x 1" x 0.064" Beryllium Oxide Insulator
Not exactly inexpensive, but combined with a copper heat sink and appropriate thermal compound, they would be extremely effective.
Did you happen to notice the text that says "Toxic" and "Not sold to individuals"???? Hello? This advice is just as bad as the guy who suggested a bath of mercury (which hopefully was in jest but you couldn't tell from the post...).
Has this thread morphed in to some kind of material fantasy trip? Where we combine our $5 chip amp with some unobtainium to get some kind of incremental (and illusory) gain in some performance metric? Beryllium Oxide? Diamond?
I'm kind of surprised no one suggested a bath of liquid nitrogen! At least you can buy that at your local compressed gasses retailer, is non toxic, would transfer a lot of heat... while it lasted.
Perhaps some practical advice would be more useful - like you cannot mate two solid surfaces (e.g. chip amp and heat sink) and get good contact without using something to fill microscopic voids and imperfections in both surfaces, e.g. a film of heat sink compound??? This alone can increase heat transfer, if that is an issue. This might be more useful advice to the person starting this thread...
-Charlie
CharlieLaub is correct, need to focus on obtainable materials, and not something that will kill you with 0.2PPM in air, or is too expensive.
For thermal compound, you can try DIY with diamond dust. Some info here.
As for heat-sink, all depends on what material you can get, higher thermal conductivity the better, unless you can get unobtanium for cheap, you are back to your staring point.
For thermal compound, you can try DIY with diamond dust. Some info here.
As for heat-sink, all depends on what material you can get, higher thermal conductivity the better, unless you can get unobtanium for cheap, you are back to your staring point.
I never said it was my intention to dissipate as much power as possible, but to dissipate it as EFFICIENTLY as possible - which is a different story altogether
Absolutely. May I suggest, deliberately misunderstanding you, that the most efficient way is the cheapest one, i.e. to use conventional aluminium heatsinks.
Which rules out the most efficient heatsinking materials. A proven solution against eddy currents is to drill holes or mill slits into the material.
Which means just the opposite of the above. Electrical conductors have no dielectricity and are most efficient in discharging static electricity.
I have lapped about 0.5mm from the bottom and the top of the chip so far and I haven't even cut through the black plastic yet. I will probably lap a bit more anyway, just to see what's really inside my amplifier (maybe a little imp
There is plastic on top and bottom? Well this may depend on which side is up for you, but please tell us you are not trying to improve the heatsinkability of the LM3875TF instead of starting by the non-isolated LM3875T.
Copper comes quite close to silver and is much better than the commonly used aluminium. It also has a better inner damping which should give “good vibrational properties” as required. Of course copper is a highly conductive material just as aluminium and was therefore ruled out from the start.
Liquid nitrogen is a possible solution it also decreases semiconductors noise what might be more important factor for some advance equipment that must operate at low temperature.
For home use regular heat sink is a perfect match imho. Otherwise tubes can be used instead they do like to be hot.
For home use regular heat sink is a perfect match imho. Otherwise tubes can be used instead they do like to be hot.
In my amp project, I will use copper heat spreaders with kapton film insulation between the copper spreader and the conventional extruded aluminum heat sink, and a thin layer of thermal compound between the chip and the spreader. The chip is clamped by an aluminum bar across the face of the chip, bolted to the heat sink without the bolts touching the spreader. Fairly conventional, no unobtainium, but fairly effective.
I see some good comments and some stupid comments here Guys, please keep in mind that I'm driven by the want to create something unique and new - something that is hopefully not worse, and perhaps better (sonically) than what we already have. Please don't take offence at some of the more outlandish ideas, as this is as much about discussing new ways of thermal managemant as it is about actually making a workable solution
@CharlieLab - you really hurt my feelings, suggesting that I would forget about such a basic thing as a thermal interface material I do know my cooling, just looking for something new this time
@michal - believe it or not, but I actually came up with the same idea independently I even found a micronised diamond supplier.
@suntechnik - Liquid nitrogen is not suitable for constant use, and is not suitable for use with amplifier chips in any case.
Oh, and I have not overlooked BeO Or BN or AlN, for that matter. It's just, like graphite heatsinks, they seem to only exist theoretically I can't find anyone who would be able to actually supply them. OCZ was intending to produce a carbon-nanotube based heatsink, they called it HydroJet. Lamentably though, they abandoned the project shortly after presenting the first prototypes 
Guys, please keep in mind that I'm driven by the want to create something unique and new - something that is hopefully not worse, and perhaps better (sonically) than what we already have. Please don't take offence at some of the more outlandish ideas, as this is as much about discussing new ways of thermal managemant as it is about actually making a workable solution @CharlieLab - you really hurt my feelings, suggesting that I would forget about such a basic thing as a thermal interface material
I do know my cooling, just looking for something new this time @michal - believe it or not, but I actually came up with the same idea independently
I even found a micronised diamond supplier.@suntechnik - Liquid nitrogen is not suitable for constant use, and is not suitable for use with amplifier chips in any case.
Oh, and I have not overlooked BeO
Or BN or AlN, for that matter. It's just, like graphite heatsinks, they seem to only exist theoretically I can't find anyone who would be able to actually supply them. OCZ was intending to produce a carbon-nanotube based heatsink, they called it HydroJet. Lamentably though, they abandoned the project shortly after presenting the first prototypes How about compressing powdered graphite in a cast to form a heat-sink? Partially embed the chip in it? Add some diamond dust to it? Still would be very fragile, unless you add some bounding agent that will not diminish thermal conductivity too much.
As for graphite heat-sinks, you probably can't find any because it's just not economically feasible to make and sell them, as there are cheaper alternatives.
That does not mean DIY is out of question
As for graphite heat-sinks, you probably can't find any because it's just not economically feasible to make and sell them, as there are cheaper alternatives.
That does not mean DIY is out of question
Did you happen to notice the text that says "Toxic" and "Not sold to individuals"???? Hello? This advice is just as bad as the guy who suggested a bath of mercury (which hopefully was in jest but you couldn't tell from the post...).
Has this thread morphed in to some kind of material fantasy trip? Where we combine our $5 chip amp with some unobtainium to get some kind of incremental (and illusory) gain in some performance metric? Beryllium Oxide? Diamond?
I'm kind of surprised no one suggested a bath of liquid nitrogen! At least you can buy that at your local compressed gasses retailer, is non toxic, would transfer a lot of heat... while it lasted.
Perhaps some practical advice would be more useful - like you cannot mate two solid surfaces (e.g. chip amp and heat sink) and get good contact without using something to fill microscopic voids and imperfections in both surfaces, e.g. a film of heat sink compound??? This alone can increase heat transfer, if that is an issue. This might be more useful advice to the person starting this thread...
-Charlie
Charlie,
While it is indeed hazardous in friable or powdered form, according to the majority of national and international agencies responsible for product safety, finished products containing sintered beryllium oxide (such as the insulators specified) present little, if any, potential for detrimental health effects when used in the intended application.
It would appear that the State of California is among the few agencies which have any restrictions on the handling of products containing sintered BeO. But then again, the government of California seems to feel that nearly everything is hazardous to ones health.
Interestingly, even the European Union has determined that sintered BeO should not be included in the RoHS directives.With regard to BeO insulators being “unobtainium”, shall I presume that you mean that the products manufactured by companies such as AAVID Thermalloy and which are available from Digikey, Allied, Mouser, etc. are unobtainable?
Over the years, I have ordered numerous BeO insulators from various venders without any difficulty whatsoever, legal boilerplate notwithstanding. Unfortunately, excessive prices and minimum purchases often apply, hence the recommendation of a surplus vender. In any event, my post was in direct response to the thread originators admittedly somewhat obscure requirements.
As such, I stand by my recommendation: If you wish to couple any semiconductor (even a $5.00 chip amp) to a heat sink with high thermal efficiency, beryllium oxide insulators are almost without peer.
regards,
Keith
Like all other currents eddy currents need a closed path. By drilling holes or making slits in the right places, those paths are interrupted and eddy currents can either not flow at all or need to use a longer path with a higher impedance which dampens them. The same principle is used in transformers where stacks of isolated sheets are used instead of massive cores to avoid eddy currents.
By the way, you may consider that eddy currents are the reason why aluminium cases have a shielding effect and that you actually could rather profit from eddy currents in your heatsink.
By the way, you may consider that eddy currents are the reason why aluminium cases have a shielding effect and that you actually could rather profit from eddy currents in your heatsink.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.