Wait a sec. Load watts is not fet dissipation watts, not even close.

fet amps x Rds(on) will give watts dissipation (when turned on fully).

IRFP480 Rds(on) is 0.27, and Pd(max) is 280W
So you not gonna get a single 480 to 600W of dissipation.

If the load is 30A, one 480 fet will dissipate I^2 x Rds, 900x0.27 = 243W.
If the load itself consumes 600W (30A 20v), then one 480 fet is not good enough.
However, spec sheet shows Id(max) is just 20A at 20C, and just 13A at 100C. Id is limited by case temp, but as you see, 20A is the limit. It simply cannot stay at 20C, so figure maybe 15A Id max per 480 fet.
With a 30A load you need two fet's min, but I would use three, maybe four.

With three fet's you get 10A per fet, so now each fet has power dissipation of 10^2 x 0.27 = 27W
There are three fet's, each at 27W, total of 81W of wasted heat.

If you run four fets, 30/4=7.5A per fet, so 7.5^2 x 0.27 = 15.2W per fet, x4 = ~61W.

To heatsink 60W to maintain case temp in ok-zone (75C or less), much easier to do.

Also remember to adjust power supply voltage up a smidge to account for Vdrop across the fet SD junction. If the load is rated 600W @ 20v and the supply is 20v, you will not be delivering 600W to the load, it's gonna be slightly less.

 

Wait a sec. Load watts is not fet dissipation watts, not even close.

fet amps x Rds(on) will give watts dissipation (when turned on fully).

IRFP480 Rds(on) is 0.27, and Pd(max) is 280W
So you not gonna get a single 480 to 600W of dissipation.

If the load is 30A, one 480 fet will dissipate I^2 x Rds, 900x0.27 = 243W.
If the load itself consumes 600W (30A 20v), then one 480 fet is not good enough.
However, spec sheet shows Id(max) is just 20A at 20C, and just 13A at 100C. Id is limited by case temp, but as you see, 20A is the limit. It simply cannot stay at 20C, so figure maybe 15A Id max per 480 fet.
With a 30A load you need two fet's min, but I would use three, maybe four.

With three fet's you get 10A per fet, so now each fet has power dissipation of 10^2 x 0.27 = 27W
There are three fet's, each at 27W, total of 81W of wasted heat.

If you run four fets, 30/4=7.5A per fet, so 7.5^2 x 0.27 = 15.2W per fet, x4 = ~61W.

To heatsink 60W to maintain case temp in ok-zone (75C or less), much easier to do.

Also remember to adjust power supply voltage up a smidge to account for Vdrop across the fet SD junction. If the load is rated 600W @ 20v and the supply is 20v, you will not be delivering 600W to the load, it's gonna be slightly less.

Dear @DC_Kid,

Thanks for your valuable reply, i think you have the right, the power consumed by a MOSFET is (A x RDSon), but my case is to use 4 MOSFETS type IRFP460 to drain 30V, 20A not 20V 30A , but in this case is the same wattage to be consumed (600W), and you mentionned that the mosfets will drain less than 600W because we have the Voltage drop on the shunt resistors also, this needs to be considered also, means, Vds = Vin - Vshunt.

So my request is to identify which is the best HEATSINK in the chinese market like ALIEXPRESS or BANGGOOD that i can use to simulate my project in the real life, preferably is to mount 4 Mosfet in one HEATSINK, and the other 4 Mosfets in other HEATSINK, The project contains 8 MOSFETS for now, any ideas ?

 

the 460 has basically same specs

So you have fet SD amps of 20.
20/4 = 5A per fet
25 x 0.27 = 6.75W per fet
x4 if all on same heatsink = 27W of wasted power

To handle 27W on one heatsink with just ambient air circulation, any of these sinks should work. They are a "smidge" bigger than required. Use what fits your mounting and space.

https://www.aliexpress.us/item/3256803831899238.html
or
https://www.aliexpress.us/item/3256802717303712.html

reference: https://celsiainc.com/resources/calculators/heat-sink-size-calculator/

side note: if the FET's need to switch on and off, then there needs to be some analysis of the non-linear areas of the switching. If the on/off is not clean and fast then the FET will get hotter.

 

the 460 has basically same specs

So you have fet SD amps of 20.
20/4 = 5A per fet
25 x 0.27 = 6.75W per fet
x4 if all on same heatsink = 27W of wasted power

To handle 27W on one heatsink with just ambient air circulation, any of these sinks should work. They are a "smidge" bigger than required. Use what fits your mounting and space.

https://www.aliexpress.us/item/3256803831899238.html
or
https://www.aliexpress.us/item/3256802717303712.html

reference: https://celsiainc.com/resources/calculators/heat-sink-size-calculator/

side note: if the FET's need to switch on and off, then there needs to be some analysis of the non-linear areas of the switching. If the on/off is not clean and fast then the FET will get hotter.

Mr. @DC_Kid


Here we have 30A to drain from the 8 Mosfet, means the power consumed is 30A² * 0.27 Rdson = 243 Watt total consumption
for one Mosfet it will be 30A/8mosfets = 3.75A
so consumption per mosfet is 3.75² * 0.27 = 3.8W per Mosfet


What i choosed for this case is to use this Heatsink, two of them, each for 4 Mosfets:
https://www.aliexpress.com/item/100...!sea!PT!934921063!&curPageLogUid=kGtvQyIt6B8T

 

fet amps x Rds(on) will give watts dissipation (when turned on fully).

True - when turned on fully as in a switch application, delivering power to a load, but here they ARE THE LOAD and so they are operating in their ohmic region - ie NOT turned on fully and your argument doesn't work.

In the TS' proposed arrangement, with 8 MOSFETs and assuming the sense resistor is still 0.22ohm, running at 75W per MOSFET (30v, 2.5A) the sense resistor dissipates I^2R = 2.5 * 2.5 * 0.22 = 1.375W & loses IR = 0.22 * 2.5 = 0.55V therefore the MOSFET dissipates (75 - 1.4)W = 73.6W (or 2.5 * (30-0.55) =73.6W)

But for heat-sink calculations we just use the 75W as its easier....

 

What i choosed for this case is to use this Heatsink, two of them, each for 4 Mosfets:
https://www.aliexpress.com/item/1005003352858467.html?spm=a2g0o.productlist.main.1.48967CA17CA1hT&algo_pvid=cb1f37b4-79bd-4f2d-80b8-6796b91e3b95&algo_exp_id=cb1f37b4-79bd-4f2d-80b8-6796b91e3b95-0&pdp_npi=4@dis!EUR!2.23!1.56!!!2.30!!@210318cf16969556735388850e0f14!12000034040485815!sea!PT!934921063!&curPageLogUid=kGtvQyIt6B8T

That's a bad sink design unless you plan to cap an end with a fan.
Convection is normal rule "heat rises", so orientation of fins and proximity to other fins, matters.

True - when turned on fully as in a switch application, delivering power to a load, but here they ARE THE LOAD and so they are operating in their ohmic region - ie NOT turned on fully and your argument doesn't work.

In the TS' proposed arrangement, with 8 MOSFETs and assuming the sense resistor is still 0.22ohm, running at 75W per MOSFET (30v, 2.5A) the sense resistor dissipates I^2R = 2.5 * 2.5 * 0.22 = 1.375W & loses IR = 0.22 * 2.5 = 0.55V therefore the MOSFET dissipates (75 - 1.4)W = 73.6W (or 2.5 * (30-0.55) =73.6W)

But for heat-sink calculations we just use the 75W as its easier....

Ok, so the ckt proposed is 75W dissipation per fet? 75 x 8 = 600W
Ok, I run sink calculator for 600W.

 

That's a bad sink design unless you plan to cap an end with a fan.
Convection is normal rule "heat rises", so orientation of fins and proximity to other fins, matters.

As a general purpose heat-sink the design is fine, used vertically with a fan blowing upwards. The problem is that even the 50mm x 50mm is probably too small, but I'll wait to see your calculations...

 

As a general purpose heat-sink the design is fine, used vertically with a fan blowing upwards. The problem is that even the 50mm x 50mm is probably too small, but I'll wait to see your calculations...

That 50-50-100 U slot, even with fans, is still shy of some needed sink.

600W is a lot of heat. That alone will raise ambient in a small room that has no AC.
Based on 100C(max) ambient and 150C(max) Tcase.

Without fans, need two of these:
300W air cool
300x200 x 50mm fin (4 fet per sink)
https://www.aliexpress.us/item/2251832835434662.html

 

That 50-50-100 U slot, even with fans, is still shy of some needed sink.

600W is a lot of heat. That alone will raise ambient in a small room that has no AC.
Based on 100C(max) ambient and 150C(max) Tcase.

Without fans, need two of these:
300W air cool
300x200 x 50mm fin (4 fet per sink)
https://www.aliexpress.us/item/2251832835434662.html

Interesting. I'd like to know how you got those results - my feeling is no more than 2 MOSFETs @ 75W each, per sink, case temp 90C, junction temp 125C ambient 40C

My own e-load runs at 1600W (30v @ 50A) and uses 16 * IXTX90N25L2 extended-SOA IXYS MOSFETs designed for linear use on much larger heatsinks with forced air cooling - sounds like a 747 taking off!

 

From the link I gave earlier

 

Hello Guys,

I appreciate your replies and your help, but for now i will go with that U shape heatsink per 4 IRFP460 Mosfets each, and see the results in real life, with temperature sensor, and then i will see if it can handle the amount of energie that comes out from the Mosfets or not, and of course i will let you know with some images of the test, i am just waiting for the sippement of the components to be here.

Another time thank you for your support, and just a simple note for @DC_Kid, the ambiant temperature will be at maximum 30ºC, because i will place the heatsink outside the box, not inside, this could help to gain more airflow passing inside the sink with the help of the FAN.

Best Regards,
Ahmed.

 

From the link I gave earlier

Ah, I missed it, you've put Tcase = 150C but thats wrong, it should be 90C (Tj-max = 150C)

 

Ah, I missed it, you've put Tcase = 150C but thats wrong, it should be 90C (Tj-max = 150C)

Well, that makes things even better.

 

Hello Guys,

I appreciate your replies and your help, but for now i will go with that U shape heatsink per 4 IRFP460 Mosfets each, and see the results in real life, with temperature sensor, and then i will see if it can handle the amount of energie that comes out from the Mosfets or not, and of course i will let you know with some images of the test, i am just waiting for the sippement of the components to be here.

Another time thank you for your support, and just a simple note for @DC_Kid, the ambiant temperature will be at maximum 30ºC, because i will place the heatsink outside the box, not inside, this could help to gain more airflow passing inside the sink with the help of the FAN.

Best Regards,
Ahmed.

Caution. You must account for heat flow vectoring (how heat moves through the material). That U shape sink is meant to have the heat source mounted to the surface from which the fins rise from (base of fins). So, just be sure to mount two FET's per side, and for 300W of heat it needs fans (two fans, you can use lower cfm for less noise, and two helps mitigate a single fan fail to save your stuff from burning up). As for fans, I would mount a fan on each end and have them blow in cool air. Mounting location of the FET's is also VERY important (location and thermal coupling). I would mount them 25mm from the end and 25mm from the side. I would also make sure the mounting surface is very clean and flat, use some 600 emery to clean off and smooth the mounting surface, clean with some rubbing alcohol, use proper heat paste.

Also to note: if that U shape is not anodized, then the natural aluminum oxidizing layer impacts heat transfer to the air. Having fans should make that a non-issue, I was just noting it for you.

Aslo to note: you mention 30C max ambient, but 600W in say a small 12x12x8 room, even with HVAC cycling, can bring ambient above 30C.

I (we AAC forum folks) would like to see your final setup.

*** NOTE ***
That U shape 100x50x50 is a small sink. It looks big in Ali pics, but it's small.

Even with big flow fans, one side is basically just 100x50x20, and for 150W (2 fets on one side) it's not good enough.

 

There's another factor in play here. Nearly all heatsink specs, and that rough estimate, assume the heat source is fully and evenly dstributed across the heatsink base, not a near 'point source' eg 20 x16mm (PLUS247 package, smaller for TO-220). This can seriously impact heatsink efficiency, making it effectively 50% smaller or worse in many cases...

 

There's another factor in play here. Nearly all heatsink specs, and that rough estimate, assume the heat source is fully and evenly dstributed across the heatsink base, not a near 'point source' eg 20 x16mm (PLUS247 package, smaller for TO-220). This can seriously impact heatsink efficiency, making it effectively 50% smaller or worse in many cases...

On that page that has the sink calculator, it has a back reference link to the math of sinks. I think the calculator accounts for some of the points you mention.

 

Using this calculator suggests that 150W from 2 MOSFETs mounted centrally on that 300 x 200 x 50 heat-sink will require an airflow of 700LFM to keep the heat-sink base at 75C which ties up with my own experiments.

 

For the 300x200x50 sink in post #28, looks ok. Two FET's per sink fours sinks in total. Yep, 600W is a lot of heat.
Need a turbine to cool? 5m/s through a 2300mm x 1050mm duct is near 26kCFM !

For the U shape OP wanted to use, no good.
That U shape sink does not have fins that are 50mm tall. The fins are approx 20mm tall, maybe less.

That 100 x 50 x 50, 1/2 of it for two fet's (150W). Heat source is meant to be mounted under fin base, for the basic math calculators.
Note the #'s I used, and then look at that U sink.

Even 75W shows temps too high with hi-flow fan.
100mm is barely 4" long. 100x50x50 with fins that are 20mm or less, is a tiny sink.

 

Any updates? I am curious.

 

Any updates? I am curious.

Dear @DC_Kid,

Actually i received the Heatsink already and i mount 4 IRFP460 on it, well for 30V, 2A without FAN, the heatsink start to heat until you can not touch it, now i am searching for a suitable FAN for the application in my company, and i will let you know with all news.

Best Regards,
Ahmed.