Kilowattski,
The figures you are using are for a heatsink with very thin metal layers unlike the proposed with up to 1mm thick layers. They aren't very big in area either. Because they aren't that big in area, they aren't spread apart very far to allow good cooling wirthout a fan. If they were the size he proposes, the fins would be a couple of cm apart at their edges. It does seem to be true that smaller homemade ones might be less efficient than larger ones by even less than their proportional size because the layers/fins are very close together close to the device. I think in larger sizes they would work verywell without fan cooling.
Actually, very few CPU heatsinks would work without fans, because they are designed to be used with fans!
The figures you are using are for a heatsink with very thin metal layers unlike the proposed with up to 1mm thick layers. They aren't very big in area either. Because they aren't that big in area, they aren't spread apart very far to allow good cooling wirthout a fan. If they were the size he proposes, the fins would be a couple of cm apart at their edges. It does seem to be true that smaller homemade ones might be less efficient than larger ones by even less than their proportional size because the layers/fins are very close together close to the device. I think in larger sizes they would work verywell without fan cooling.
Actually, very few CPU heatsinks would work without fans, because they are designed to be used with fans!
Magura!
O.K., but the main question is : which one is better heat-conveying, the aluminium filled epoxy or the silicon grease with screw?
kilowattski said:
well not neccesarily always on fans, put a thermistor on the Heatsink somewhere to turn the fans on (120VAC fan/thermistor bridged with a cap to prevent arcing) so they wouldn't always be noise 🙄
That really wouldn't work either as the nature of a Class A amplifier dictates that it dissipates power constantly even with no signal so the fan would be on constantly once the amp reached operating temperature. I also meant audible noise from the fan running not electrical noise.
kilowattski said:
I don't mean to be rude, but this is DIY right??? You weren’t supposed to go out and buy the Zalman ones...... And the physical size of the CPU coolers aren’t really what I would even consider using in a class A amp. 😉
The thing was to use the IDEA!!!
Upscale and use thicker fins! The fins on the Zalman ones is about 0.2 mm! I recommend 0.5 to 1.0 mm (thicker is harder to bend to shape but would be better). And the max distance between the fins on the Zalman is 3 mm at max (the end of the fins), a DIY one would use way much longer fins, and a lot more space between the fins. And the width (length of the bars claming the fins together) of the Zalman is about 4 cm (that would be the height if used without fans to get the air moving and the way we would use them). A DIY one would be best to have about 20 - 30 cm high, with fins of at least 10 cm (probably longer). And that is a LOT more area.
And it is like Variac says, CPU coolers are made to have fans on them, and thus have lots of heatsink area per volume to make it small. But remember, this is not the way of making the heatsink that does this, it is how it is implemented. Most CPU coolers are made the "normal" way, but you guys aren’t skeptical to using heatinks made by the same technique because of that.
So, forget that the example I gave you were from a CPU cooler! This way of doing it makes it possible to make as large (and efficient) heatsink you could EVER want, AND make it suitable for non forced cooling.
And about materials, copper IS a way better material here. First of all the lower thermal resistance. There is much higher thermal gradients here. It is important to get the heat efficiently spread.
Secondly aluminum (witch is about the only real alternative??) is not that easy to bend (especially in the sharpest bends). It forms cracks if bent to sharp.
But the price difference is rather big, so if you are able to get a LOT of more area by using aluminum that is probably the best.
One good source of copper plates is roof coverers (I don’t know the English word for it, but the guys that puts up coverings around chimneys and such with copper plates).
There are lots of sources for cheap (or FREE!) metal plates around. Your imagination is the only thing that sets the limits here 😀 And if the plates you are able to get doesn't have perfect surface or for some other reason is best not shown it is possible to make the heatsinks internal, or covered up or something.
This IS a great way of using simple materials to make really good heatsinks. No other way of doing it makes that many sources of materials usable! Used aluminum roof plates or whatever. LOTS of potential sources for materials here!
Anders
as for the claims by kilowattski of the heatsinks not working without a fan....its a real good ide our nowegian friend have come up with....for the cpu size, they sure need a fan, but for a larger scale they sure can run convection...just like any other heatsink.
About the choice of material:
Aluminium is the material of choice here for 2 reasons.
1) is only about a quarter of the weight of copper...considering the size in question, weight is an issue- if not for other reasons, then for the sake of price. A copper version of a useable size would cost a fortune.
2) You can anodize the aluminium black, that increases the efficiency of the heatsink.
It was mentioned that aluminium cracks when you bend it, that is not correct. There are some alloys that would crack, but they are few and expensive.
The main reason for any metal to crack around the edges when bent, is the bad surface quality on the edge after cutting. If filed or sanded, the edge wont crack...even the hardest alloys can be bent if done right.
Magura
About the choice of material:
Aluminium is the material of choice here for 2 reasons.
1) is only about a quarter of the weight of copper...considering the size in question, weight is an issue- if not for other reasons, then for the sake of price. A copper version of a useable size would cost a fortune.
2) You can anodize the aluminium black, that increases the efficiency of the heatsink.
It was mentioned that aluminium cracks when you bend it, that is not correct. There are some alloys that would crack, but they are few and expensive.
The main reason for any metal to crack around the edges when bent, is the bad surface quality on the edge after cutting. If filed or sanded, the edge wont crack...even the hardest alloys can be bent if done right.
Magura
Could a decent heatsink be made using short lengths of copper pipe stacked between plates of aluminum and soldered/welded into place? It certainly sounds like it would be cheap. The copper pipes should allow for good air flow because of their shape, and it should provide plenty of surface area. What do you thermodynamics experts think?
copper and aluminium is hard to weld together, its been done, but not an easy task, soldering is even worse.
You dont get much surface from a pipe, since each pipe takes up too much space.
If you wanna make a heatsink, look at the earlier postings in ths thread, especially the last mentioned type (the psalm book) is easy (as easy as it gets...that dosnt mean actually easy to do) and good/- cheap.
In the end of the day i wouldnt make my own heatsinks without acces to a sheetmetal workshop, it will most likely turn out as a lot of work and little satisfaction (bad looking).
Looking at the actual cost of a good extruded heatsink, a DIY heatsink have to be a project for the sake of making a heatsink or because you simply cant get what you need off the shelf (as i wrote earlier, i machined a bunch of heatsinks, but only because i couldnt buy something usefull).
Magura
You dont get much surface from a pipe, since each pipe takes up too much space.
If you wanna make a heatsink, look at the earlier postings in ths thread, especially the last mentioned type (the psalm book) is easy (as easy as it gets...that dosnt mean actually easy to do) and good/- cheap.
In the end of the day i wouldnt make my own heatsinks without acces to a sheetmetal workshop, it will most likely turn out as a lot of work and little satisfaction (bad looking).
Looking at the actual cost of a good extruded heatsink, a DIY heatsink have to be a project for the sake of making a heatsink or because you simply cant get what you need off the shelf (as i wrote earlier, i machined a bunch of heatsinks, but only because i couldnt buy something usefull).
Magura
I think akb's idea will work in aluminum or copper.
The copper makes the best heatsink- no doubt. With copper you can use thinner sheets and get more fins. I don't think the weight is such a problem. The plates could be cheaply cut by a sheet metal shop that makes gutters, heating ducts, etc. They can cut 20 at a time probably with the shears they have. The roofing supply shop is probably a good suggestion for getting the metal. Metal for the edges and transitions of roofs is called flashing in english. I think the aluminum will work well too. Normal aluminum won't crack, it is quite soft.
I think it comes down to what you can find at a good price. Unless you find it as surplus copper is very expensive.
Tpenquin: If you make the heatsink entirely from copper (the pipes and plates) you could solder it together, but you need a torch like used for plumbing. It's hard to get it hot because-well-it's a heatsink!!
The copper makes the best heatsink- no doubt. With copper you can use thinner sheets and get more fins. I don't think the weight is such a problem. The plates could be cheaply cut by a sheet metal shop that makes gutters, heating ducts, etc. They can cut 20 at a time probably with the shears they have. The roofing supply shop is probably a good suggestion for getting the metal. Metal for the edges and transitions of roofs is called flashing in english. I think the aluminum will work well too. Normal aluminum won't crack, it is quite soft.
I think it comes down to what you can find at a good price. Unless you find it as surplus copper is very expensive.
Tpenquin: If you make the heatsink entirely from copper (the pipes and plates) you could solder it together, but you need a torch like used for plumbing. It's hard to get it hot because-well-it's a heatsink!!
Making a psalm book of 300*100*1 mm copper sheets, lets say 50 pcs.....makes a 12.5 kg heatsink.....now thats (a wild guess) around 0.4-0.5C/W, meaning that you will need more than one for most amps. In the end of the day they would be fragile and heavy/ expensive.
For aluminium you can divide the weight by 2 since the heat transfer of aluminium is half of copper based on transfer to weight ratio.
Magura
For aluminium you can divide the weight by 2 since the heat transfer of aluminium is half of copper based on transfer to weight ratio.
Magura
Materials properties:
Emissivity (heat dissipated per surface area) is also higher for Aluminium than for Copper. You gain a further 10% by anodizing.
Petter
Emissivity (heat dissipated per surface area) is also higher for Aluminium than for Copper. You gain a further 10% by anodizing.
Petter
good point....aluminium for heatsinks in not limited sizevise matters are sure a win-win VS. copper
Magura
Magura
akb1212,
I did not take your post to be rude. I simply think this design will require a fan to function properly even for a large DIY version. I am simply not in love with this design, but thats just my opinion. I do not want to rain on anyones parade. For the record, my reason for not liking the design, is that I think it will be difficult to fabricate and then again there is the fan issue. Thats all. Again, I could be wrong not having tried to fabricate one, it is simply speculation. If someone proves me wrong, thats great as I will be happy to stand corrected. I am becoming intrigued by water cooling out of the desire to some day build a higher power aleph x of about 100 watts. C&H TECHNOLOGY, INC. makes some very efficient cool plates which according to their specs has a thermal resistance of .007 at a 1 gallon/minute flow for a 12" plate. I am trying to find out how much a 12" plate would cost. A simple PC liquid cooling radiator, pump and fan should be enough to cool the cool plate. Something like the Koolance Exos would be perfect. A smaller chill plate could bolted to a conventional heat sink to increade it's heat dissapation capabilities. The neet thing is that the cooling unit could be put in a closet or other room to cut down on fan noise. I just started looking at water cooling and I am starting to be intrigued with it's possibilities for cooling a higher power Class A amplifier where a conventional heat would not be able to handle the 400 to 700 watts or higher of heat dissapation.
I did not take your post to be rude. I simply think this design will require a fan to function properly even for a large DIY version. I am simply not in love with this design, but thats just my opinion. I do not want to rain on anyones parade. For the record, my reason for not liking the design, is that I think it will be difficult to fabricate and then again there is the fan issue. Thats all. Again, I could be wrong not having tried to fabricate one, it is simply speculation. If someone proves me wrong, thats great as I will be happy to stand corrected. I am becoming intrigued by water cooling out of the desire to some day build a higher power aleph x of about 100 watts. C&H TECHNOLOGY, INC. makes some very efficient cool plates which according to their specs has a thermal resistance of .007 at a 1 gallon/minute flow for a 12" plate. I am trying to find out how much a 12" plate would cost. A simple PC liquid cooling radiator, pump and fan should be enough to cool the cool plate. Something like the Koolance Exos would be perfect. A smaller chill plate could bolted to a conventional heat sink to increade it's heat dissapation capabilities. The neet thing is that the cooling unit could be put in a closet or other room to cut down on fan noise. I just started looking at water cooling and I am starting to be intrigued with it's possibilities for cooling a higher power Class A amplifier where a conventional heat would not be able to handle the 400 to 700 watts or higher of heat dissapation.
Here's a relevent amplifier-the Pass Aleph 2
It is almost exactly 30 cm in each dimension
As we all know, Nelson tends to overdo rather than underspec things. I think it gives a good idea of whats needed. In a while I will count up the fins! The output devices for this amp are about
16x13mm so the stack of plates can't be thicker than about 16mm
or 16 1mm sheets. There are 12 devices in this box, so another question is how many to put on each heatsink- probably 3, so we need 4 heatsinks
Looks like about 15 fins per each of 4 sides, each one about 50mm long. I can't tell if the fins are on the back also, could someone tell me? I'm assumeing yes.
Since our homemade heatsink fins are close together at the base lets say we have to make them longer-75mm, that makes each heatsink 150mm wide and 300mm tall so 2 of them will fit on each side of a 300mm square box. Each heatsink has 15 1mm fins. That leaves the front and rear of the box not requireing heatsinks?
If they aren't that pretty, they could be placed on the inside of the box with screen or mesh top and bottom of the box for ventilation.
So, an estimate IMHO is 60 plates, each 75mm x 300mm.
So go find some old aluminum gutters at the scrapyard!! Don't let your wife catch you taking them off your house!! 😀
It is almost exactly 30 cm in each dimension
As we all know, Nelson tends to overdo rather than underspec things. I think it gives a good idea of whats needed. In a while I will count up the fins! The output devices for this amp are about
16x13mm so the stack of plates can't be thicker than about 16mm
or 16 1mm sheets. There are 12 devices in this box, so another question is how many to put on each heatsink- probably 3, so we need 4 heatsinks
Looks like about 15 fins per each of 4 sides, each one about 50mm long. I can't tell if the fins are on the back also, could someone tell me? I'm assumeing yes.
Since our homemade heatsink fins are close together at the base lets say we have to make them longer-75mm, that makes each heatsink 150mm wide and 300mm tall so 2 of them will fit on each side of a 300mm square box. Each heatsink has 15 1mm fins. That leaves the front and rear of the box not requireing heatsinks?
If they aren't that pretty, they could be placed on the inside of the box with screen or mesh top and bottom of the box for ventilation.
So, an estimate IMHO is 60 plates, each 75mm x 300mm.
So go find some old aluminum gutters at the scrapyard!! Don't let your wife catch you taking them off your house!! 😀
Attachments
Variac said:
They can be about as wide as you please if you make the base thick enough. Lets say you want 40 instead of 15 finns, you just have to make the base 12,5mm thicker, that should do the trick.
This design truly is scaleable...im getting to the point of feeling like making one just to see what it looks like in real life 😀
Magura
Perhaps I misunderstand you, but I keep making the same mistake (if you are making a mistake). The edge of each sheet HAS to touch the back of the output device. The brilliant part of this design is that there is no traveling of heat through multiple layer boundries.
But thinking it through more: Practically, it probably is not a problem for multiple devices. You just stagger the devices in different positions on the "spine"
Mark
But thinking it through more: Practically, it probably is not a problem for multiple devices. You just stagger the devices in different positions on the "spine"
Mark
Variac said:
Exactly....all you gotta keep in mind is that the spine must be thick enough to transfer the heat. Lets say we bolt the spine fpr every 30mm, then the heat transfer the layers in between cant be much of an issue. Im not sure, but i guess the bigger versions ive seen of this type were welded. For a small version like this i guess id just weld the backside of the spine nice and deep, prior to making the mounting surface.
Magura
No, the point is we don't want to deal with heat transfer between layers. The design is not fully scalable. The spine can only be as thick as the combined width of the staggered devices.
The edge of each sheet must contact the back of at least one device for this to work optimally. Of course it will probably work fine if you add a few layers and count on some heat transfer, but this isn't optimal.
The edge of each sheet must contact the back of at least one device for this to work optimally. Of course it will probably work fine if you add a few layers and count on some heat transfer, but this isn't optimal.
- Status
- Not open for further replies.