...something fishy about half size heatsinks?
The idea that cutting a heatsink in half length will improve its ability to transfer heat, seemed wrong to me when it was first mentioned. It still seems wrong to me.
I just read the "thermal tutorial" pdf on http:Wakefield.com the Wakefield Thermal Solutions site to see if there was any merit to this idea. Since it is a very basic idea, a basic tutorial would show this as being a factor. It does not show this as being valid, nor a design factor.
In fact just the opposite is true. They show a "design selector graph" where the 3" standard height rating is used to determine the target thermal resistance for the system - the output of the graph is the required total "height" (long linear dimension) of the extrusion necessary. Longer/more is needed for lower thermal resistance/getting rid of watts!
There may be some confusion due to the idea that in natural convection an extrusion with many thin fins will start to have limitations on the air flow, so adding length beyond a certain point will begin to have reduced benefit since the air is not flowing. But that should not be a significant factor in the relatively short heatsinks we are using, and especially not if the fin spacing is not terribly close.
I say use a big chunk o' heatsink if you got it.
The F5 that I have had my hands on gets plenty hot with a heatsink that is about 10" wide on the bottom and maybe 12" high, about ~20 fins across, so maybe 0.5" centers on them... iirc that is about the size NP suggests for the F5 in the article??
Don't skimp on heatsinks unless you want that thermistor to drop your bias down out of class A due to the temperature rising... anyone checking their bias with the amp full hot??
_-_-bear
The idea that cutting a heatsink in half length will improve its ability to transfer heat, seemed wrong to me when it was first mentioned. It still seems wrong to me.
I just read the "thermal tutorial" pdf on http:Wakefield.com the Wakefield Thermal Solutions site to see if there was any merit to this idea. Since it is a very basic idea, a basic tutorial would show this as being a factor. It does not show this as being valid, nor a design factor.
In fact just the opposite is true. They show a "design selector graph" where the 3" standard height rating is used to determine the target thermal resistance for the system - the output of the graph is the required total "height" (long linear dimension) of the extrusion necessary. Longer/more is needed for lower thermal resistance/getting rid of watts!
There may be some confusion due to the idea that in natural convection an extrusion with many thin fins will start to have limitations on the air flow, so adding length beyond a certain point will begin to have reduced benefit since the air is not flowing. But that should not be a significant factor in the relatively short heatsinks we are using, and especially not if the fin spacing is not terribly close.
I say use a big chunk o' heatsink if you got it.
The F5 that I have had my hands on gets plenty hot with a heatsink that is about 10" wide on the bottom and maybe 12" high, about ~20 fins across, so maybe 0.5" centers on them... iirc that is about the size NP suggests for the F5 in the article??
Don't skimp on heatsinks unless you want that thermistor to drop your bias down out of class A due to the temperature rising... anyone checking their bias with the amp full hot??
_-_-bear
Last edited:
No, but the fact is that you dont get much improved cooling by using bigger than max optimal size
Then its better to use TWO heatsinks
And NO, if a heatsink is already on the small side, it should NOT be cut in two
I will be using 2x 300x200mm each channel/mono, rather than one 300x400
A picture of the heatsink in question might help
Let's consider that idea... what is the "optimal" size?
Is it the exact size that results in the "textbook" required deg/watt dissipation? Or should we upsize a % beyond that to cover the possible increase in ambient temperature (summer time?). How much beyond??
The reality, I think is that the limitation on most practical extrusions WRT length is not in the fins or the air movement, but in the inherent thermal resistance of the "back plate" related mostly to the thermal conductivity of aluminum itself and the thickness limitation from the extrusion process.
As a result, there are some companies that make heatsinks not by extrusion process but by staking or thermal epoxy fixing of fins of greater length (tallness) than extrusion permits AND much thicker back/rear plate than extrusion permits.
I have an example of just such a beast - it has something like a 7/8" rear/back plate and fins that are rippled and about 4 " high...
Cutting a properly sized heatsink into two sections and then placing half the thermal load on each is not the same as what I understood to have been suggested earlier... that seems fine.
_-_-bear
Is it the exact size that results in the "textbook" required deg/watt dissipation? Or should we upsize a % beyond that to cover the possible increase in ambient temperature (summer time?). How much beyond??
The reality, I think is that the limitation on most practical extrusions WRT length is not in the fins or the air movement, but in the inherent thermal resistance of the "back plate" related mostly to the thermal conductivity of aluminum itself and the thickness limitation from the extrusion process.
As a result, there are some companies that make heatsinks not by extrusion process but by staking or thermal epoxy fixing of fins of greater length (tallness) than extrusion permits AND much thicker back/rear plate than extrusion permits.
I have an example of just such a beast - it has something like a 7/8" rear/back plate and fins that are rippled and about 4 " high...
Cutting a properly sized heatsink into two sections and then placing half the thermal load on each is not the same as what I understood to have been suggested earlier... that seems fine.
_-_-bear
Tinitus and AndrewT are right. If you go to Rtheta you can use their online simulator. The right size is best and bigger is nearly no advantage. I think it was AndrewT or someone he was talking to on the John Curl Blowtorch thread that mentioned something worth mentioning here. If you strapped a mosfet to the Brooklyn Bridge you still could not dissipate much more heat than for instance the 8x6x2.5 sink that Nelson recommends for each device in this circuit. A hot spot occurs and it has resistance to spreading further out in the sink so you need to maximize the dissipation to air around that hotspot. Adding more length on the already cool end of a large heatsink helps zero.
I think you only need to register on the Rtheta site to simulate any of their extrusions and you can visually see proof of this.
Uriah
Edit. We were typing similar ideas at the same time. I agree with most of what Bear said there. To thin on the backplate and the heat cant get away fast enough because it quickly goes to only a few fins, path of least resistance, and to thick and you get a hotspot right over the mosfet that has no escape via a fin. Cutting in half helps if the sink is ALMOST enough for two mosfets because it allows extra surface contact with the air where the cut is performed.
Rippled is not very effective unless there is a fan involved.
I think you only need to register on the Rtheta site to simulate any of their extrusions and you can visually see proof of this.
Uriah
Edit. We were typing similar ideas at the same time. I agree with most of what Bear said there. To thin on the backplate and the heat cant get away fast enough because it quickly goes to only a few fins, path of least resistance, and to thick and you get a hotspot right over the mosfet that has no escape via a fin. Cutting in half helps if the sink is ALMOST enough for two mosfets because it allows extra surface contact with the air where the cut is performed.
Rippled is not very effective unless there is a fan involved.
Last edited:
So, let's use smaller heatsinks so the efficiency will be higher and temperature of the devices will be lower??
Please share your conclusions Sir Andrew?
_-_-bear
exactly so... but then the issue is NOT the heatsink's ability, but the resistance of the JUNCTION between the device and the heatsink!!
(that and the aforementioned ability of the back plate to spread heat - the thermal resistance locally)
A completely different issue...
You assume the end of the heatsink on the F5 is "cool"?
_-_-bear
Last edited:
It's a real pain in the A that they view the length, width, height differently - at least the thermal efficiency is reasonably logical.
For F5 (standard build) the per channel consumption is about 80 watts and temp rise approx 25*C so heatsink thermal rate design is 25/80 *C/watt (= 0.31 *C/w) - plenty of examples of different proportions, widths, lengths, etc. Unfortunately, 2 seperate heatsinks will quite often track at different temps if not tied together via a solid plate for good thermal conduction and alter the behaviour of those carefully matched Fets.
The "Conrad Heatsink" site's Technical section discusses just this very thing in practical terms - suggest a look,
Hopefully, this link will work http://conrad heatsinks
For F5 (standard build) the per channel consumption is about 80 watts and temp rise approx 25*C so heatsink thermal rate design is 25/80 *C/watt (= 0.31 *C/w) - plenty of examples of different proportions, widths, lengths, etc. Unfortunately, 2 seperate heatsinks will quite often track at different temps if not tied together via a solid plate for good thermal conduction and alter the behaviour of those carefully matched Fets.
The "Conrad Heatsink" site's Technical section discusses just this very thing in practical terms - suggest a look,
Hopefully, this link will work http://conrad heatsinks
Ofcourse not
Cooling will be better, but no by much more that by the air itself
Fore every type of heatsink you can draw a curve that shows the lowering of temperature fore every cm you ad to length
At a certain point the curve flattens
Which means that any material added above this point is waste
A good 300mm wide heatsink may reach this point at 200-250mm length
A 100mm wide heatsink reach optimal length at maybe 100mm length
I suppose it also has to do with how FAST the heat spreads
As said, if you mount a device on exstremely large heatsink, it may be gone long before the heat reaches the outer ends
In other words, its the area close to the device that is most important
And how big the effective area is, depends on type of heatsink design
Its really quite simple
I don't think we've got too much differential in the present view on heatsinks...
I am going to pick up an "el cheapo" non-contact thermometer - something I have always wanted and it is on sale at Harbor Freight dirt cheap now... going over there later this week and get me one!
Anyhow since it is spot reading, afaik, I'll be able to check the local temps across Rick's F5 heatsinks and see what the delta and absolute max values look like... I think I know based upon my hand, but i'll have some real data.
Let me stick my head out, so that someone can come along with one swift blow and dispatch me to the audio hereafter... I think that the temp along his somewhat oversized heatsinks will prove to be rather even, indicating that there is no "hot spot" and that this slightly oversized bit of recycled pop can is working in such a way as to keep the temp rise down, and that the 'extra' length/height (not fin depth -hee hee screwy nominclature, eh?) is working to good benefit!
_-_-bear
I am going to pick up an "el cheapo" non-contact thermometer - something I have always wanted and it is on sale at Harbor Freight dirt cheap now... going over there later this week and get me one!
Anyhow since it is spot reading, afaik, I'll be able to check the local temps across Rick's F5 heatsinks and see what the delta and absolute max values look like... I think I know based upon my hand, but i'll have some real data.
Let me stick my head out, so that someone can come along with one swift blow and dispatch me to the audio hereafter... I think that the temp along his somewhat oversized heatsinks will prove to be rather even, indicating that there is no "hot spot" and that this slightly oversized bit of recycled pop can is working in such a way as to keep the temp rise down, and that the 'extra' length/height (not fin depth -hee hee screwy nominclature, eh?) is working to good benefit!
_-_-bear
Hello
My experience with long heat sink a la 50Cm long or more ..
Better to cut it half and use a piece of copper 8-10mm plate under the power mosfet .
If you have a 50 Cm heat sink you can realize very easy the end of the heat sink tend to be mush cooler than in the mid section were the mosfet mounted .
Also I have very bad experience to extend the lead of the power mosfet by wire , especially strained wire .
I had some similar problem with my Aleph X . The heat sink on side was at least 8-10C degree cooler than were the mosfets were mounted .
I used silver plated Teflon wire just 2inch for every power mosfet to spread the heat evenly .
The sound of the amp become a disaster ,
A did com pair one side to another so I know what I write here .
After I trow all the extension wire out and I soldered back the mosfet direct into the PC board .
It was day and night the difference sound wise .
Probably if you use solid wire that is another cup of tea .
But for me no more wire between the power mosfet and the PC board .
Greets
My experience with long heat sink a la 50Cm long or more ..
Better to cut it half and use a piece of copper 8-10mm plate under the power mosfet .
If you have a 50 Cm heat sink you can realize very easy the end of the heat sink tend to be mush cooler than in the mid section were the mosfet mounted .
Also I have very bad experience to extend the lead of the power mosfet by wire , especially strained wire .
I had some similar problem with my Aleph X . The heat sink on side was at least 8-10C degree cooler than were the mosfets were mounted .
I used silver plated Teflon wire just 2inch for every power mosfet to spread the heat evenly .
The sound of the amp become a disaster ,
A did com pair one side to another so I know what I write here .
After I trow all the extension wire out and I soldered back the mosfet direct into the PC board .
It was day and night the difference sound wise .
Probably if you use solid wire that is another cup of tea .
But for me no more wire between the power mosfet and the PC board .
Greets
You mean like this 25 KILO heatsink
I will use it fore BA-2
Man, I thought I had big heavy ones fore my F5
But this one, its so heavy its insane
Why Ba-2 instead of F5 ?
Could a line input transformer with the capacity to be wired 1:2 ratio be installed before the F5 for an extra 3dB of gain? I think this should reduce the amps input impedance as seen by the preamp from 100K to 25K.
I was thinking of something like the Lundahl LL1540. This is 30dBU capable, and the HF roll-off should reduce the risk of HF oscillations.
Is my understanding correct? Any thoughts?
I was thinking of something like the Lundahl LL1540. This is 30dBU capable, and the HF roll-off should reduce the risk of HF oscillations.
Is my understanding correct? Any thoughts?
I could have posted "De (la) Puta Madre", but most will not know the (verbal and real) meaning either. 
("M de P" is from the lyrics of a bilingual song)
For real filth, visit the national archaelogical museum in Napels to read for yourself what Roman grafito artists were capable of.
A fat boy base plate is never the cause of hot spots.
Hot spots are by definition the effect of restricted heat flow rate level due to a Thin base plate in relation to a high dissipation level.
Thermal conductivity at ambient temp levels is around 235 W/k*m for Aluminum, for a 6"x12" (HxW) heatsink with a 0.2" base plate that becomes a limiting factor.
("M de P" is from the lyrics of a bilingual song)
For real filth, visit the national archaelogical museum in Napels to read for yourself what Roman grafito artists were capable of.
A fat boy base plate is never the cause of hot spots.
Hot spots are by definition the effect of restricted heat flow rate level due to a Thin base plate in relation to a high dissipation level.
Thermal conductivity at ambient temp levels is around 235 W/k*m for Aluminum, for a 6"x12" (HxW) heatsink with a 0.2" base plate that becomes a limiting factor.
Last edited: