Technical Information - Temperature Correction Considerations
Aavid use 75Cdegrees for delta Ts-a
Technical Information - Performance Factor Table
Aavid use Rth s-a proportional to sqrt(Height) go from height 0.25 to 1 to 4, shame 16 is not included in the table.
Aavid use 75Cdegrees for delta Ts-a
Technical Information - Performance Factor Table
Aavid use Rth s-a proportional to sqrt(Height) go from height 0.25 to 1 to 4, shame 16 is not included in the table.
Hmm, looks like Fischer claims slightly less than 0.3K/W for SK435 for a length of 150mm (6"). That might imply their data is for a temperature difference of even more than 75C. 80?
Depends on the operating temperature (plus for natural convection only)
Try this on for size, the German Way : www.alutronic.com/NeueHomepage/PDF/Alutronic2010_Standard_Profilkuehlkoerper.pdf
See PR171
(there's no international protocol/standard on the procedure of determining thermal resistance numbers, or delta-t)
Last edited:
This I do not entirely agree.
Even for my first Class A project (AXJ100), I have been able to predict MOSFET case temperature to an accuracy of 2°C. The data provided by the manuafcturers, be it heatsink extrusions, or insulation sheets, or else, are usually quite accurate and detailed. But thermal convection is not a subject describeable with one single equation. You just need to take time and read the technical sections of heatsinks etc. properly.
Of couse you can just over-dimension by a factor of two, and then you can hardly go wrong. Amongst others, big amp cases and heatsinks always look impressive.
Patrick
Even for my first Class A project (AXJ100), I have been able to predict MOSFET case temperature to an accuracy of 2°C. The data provided by the manuafcturers, be it heatsink extrusions, or insulation sheets, or else, are usually quite accurate and detailed. But thermal convection is not a subject describeable with one single equation. You just need to take time and read the technical sections of heatsinks etc. properly.
Of couse you can just over-dimension by a factor of two, and then you can hardly go wrong. Amongst others, big amp cases and heatsinks always look impressive.
Patrick
You cannot predict the climate. So, aircon excepted, a bit of margin isn't a bad thing.
They say it's warming.😉
They say it's warming.😉
You can predict how many degrees temperature rise above ambient very acuurately.
Whether ambient is 25°C or 35°C makes very little difference to the final results.
Patrick
Whether ambient is 25°C or 35°C makes very little difference to the final results.
Patrick
Sure you can, why i ordered a +40ft retractable harmonica swimming pool cover in the Czech rep. this weekend, at 20% discount till monday the 28th.
At the burden of wrenching 3000lb of polycarbonate panels and powdercoated aluminum frames together myself, me just saved myself $18K-$20K, Yeeha.
Just required firing up the cerebral department, a bit of reading, translating, and asking around.
Notta Badda for a couple of days labor, imho.
For some, the journey is often more interesting than the destination (the latter is merely the excuse)
If you're 50% mechanical engineer, as Patrick and me, throwing in plenty chunks of metal is sub-standard.
Folks who prefer it fast and easy should just follow up on Pops' advice from the DIY Aleph days : Never enough Heatsink
Your fingers will tell the difference.
Components may not appreciate too much warmth.
Cool is comfortable.
It leaves also a margin for Diyer's building errors. (heat transfer)
( Ambient differencies can be way more than 10°C)
Last edited:
Let's not argue over this for no reasons.
Jacco and I merely want to point out the possibility to predict the temperature rise.
If anything, it is a technical challenge more than cost saving.
If you feel more comfortable to leave a larger margin than to go into the theory in depth, it is also totally legitimate.
Regards,
Patrick
Jacco and I merely want to point out the possibility to predict the temperature rise.
If anything, it is a technical challenge more than cost saving.
If you feel more comfortable to leave a larger margin than to go into the theory in depth, it is also totally legitimate.
Regards,
Patrick
In that case, you should definitely read the entire Alutronic pdf i linked to.
Thermal resistance figures for forced convection heatsinks are determined by Alutronics at a heatsink temperature of 85C and 35C ambient (aka delta t of 50 C's )
I have made an experiment.
I placed a classA amp at the edge of a table. One heatsink outside.
The heatsink with nothing under it ran way cooler than the one on the table because natural convection was better.
For which one will you calculate the temperature rise.
Which numbers would you choose for air flow. Would you consider the feet's high?
I do not contest calculations can accurately predict results but these results will depend on hypothesis about initial conditions. These conditions vary for eachone.
That's why i do not bother too much these calculations.
The curves given by heatsinks providers suffice to me. And some margin.
In fact, i have other exiting priorities in the understanding of amplifiers theory.
I placed a classA amp at the edge of a table. One heatsink outside.
The heatsink with nothing under it ran way cooler than the one on the table because natural convection was better.
For which one will you calculate the temperature rise.
Which numbers would you choose for air flow. Would you consider the feet's high?
I do not contest calculations can accurately predict results but these results will depend on hypothesis about initial conditions. These conditions vary for eachone.
That's why i do not bother too much these calculations.
The curves given by heatsinks providers suffice to me. And some margin.
In fact, i have other exiting priorities in the understanding of amplifiers theory.
Last edited:
The one over the edge, which is the right one.
You are supposed to leave an air gap below heatsinks. Having their lower edge nearly flush with the surface below, chassis feet or no feet, basically 'short circuits' a major part of the heatsink, which is what your experiment demonstrated.
If memory serves, then the recommended spacing is equal to or greater than the depth of the heatsink itself.
Last edited:
> For which one will you calculate the temperature rise.
Free entry & exit.
But I doubt if you get more than 5°C difference at MOSFET case even if obstructed, unless you have very low fins and large depth, like 80mm.
The non-uniform temperature distribution of your heat source over the sink area makes more difference.
> Would you consider the feet's high?
"If memory serves, then the recommended spacing is equal to or greater than the depth of the heatsink itself."
This is indeed a very good answer.
And the longer the heatsink, the less this difference (entry losses between free and obstructed).
It is also know that I like using towers with long heatsinks.
😉
Patrick
.
Free entry & exit.
But I doubt if you get more than 5°C difference at MOSFET case even if obstructed, unless you have very low fins and large depth, like 80mm.
The non-uniform temperature distribution of your heat source over the sink area makes more difference.
> Would you consider the feet's high?
"If memory serves, then the recommended spacing is equal to or greater than the depth of the heatsink itself."
This is indeed a very good answer.
And the longer the heatsink, the less this difference (entry losses between free and obstructed).
It is also know that I like using towers with long heatsinks.
😉
Patrick
.
Last edited:
Can someone supply a picture or diagram of an ideal heatsink with the most efficient proportions.
Steve Slater's wonderfeet, Ø 1.5"/38mm, couple of years ago he sent me a box of them (i tend to buy in c to k numbers)
The bottom panels of the cases i construct have 0.2" minimum thickness, why cheapskate (at $3.70/lb).
Even without machined alloy rings inbetween feet and panel, regular heatsinks are raised well enough above the stand.
www.apexjr.com/images/APEXfeet1.jpg ($0.50 each/4, $0.43 for a box of 350)
The bottom panels of the cases i construct have 0.2" minimum thickness, why cheapskate (at $3.70/lb
).Even without machined alloy rings inbetween feet and panel, regular heatsinks are raised well enough above the stand.
www.apexjr.com/images/APEXfeet1.jpg ($0.50 each/4, $0.43 for a box of 350)
Attachments
Baseplate thickness ~ diagonal dimension across sink / 20
Fin thickness at root ~ fin height / 10
Fin thickness tapered to ~ 30% of root at the tip.
Fin gap ~ 5 to 10mm depending on sink height.
Fin thickness at root ~ fin height / 10
Fin thickness tapered to ~ 30% of root at the tip.
Fin gap ~ 5 to 10mm depending on sink height.
Simulating heatsinks
Go Here
R-Tools
Register
Once they let you in you can simulate to your hearts content. You pick the extrusion, tell it the size, tell it a lot of other stuff that is on your datasheet, then hit the simulate button and you will simulate the heat on all points on the sink. Nice thing is that once you put the data in you can change the size of the sink over and over looking for the perfect temp without changing anything else.
Uriah
Go Here
R-Tools
Register
Once they let you in you can simulate to your hearts content. You pick the extrusion, tell it the size, tell it a lot of other stuff that is on your datasheet, then hit the simulate button and you will simulate the heat on all points on the sink. Nice thing is that once you put the data in you can change the size of the sink over and over looking for the perfect temp without changing anything else.
Uriah