How much watts can these 2N3055 dissipate on this heatsink?

[Electric flower]

I am building a linear power supply 25V 10A 
But i'm concerned that this heatsink isn't big enough, but my friend says it's good enough, but in my analogy if i have 25V at input (collectors) of transistors and 5V at output while pulling 10A through it that would make Pdis.=(25-5)V × 10A = 200W, or 50 per transistor  , but he says that transistor (semiconductor) doesn't act like a resistor...

I'm hoping for some help because i'm really confused...

[KJDS]

your maths is right, so you need to lose 200W.

At a very rough guess that's a 1 degrees C per W heatsink,

Here is something similar
http://uk.farnell.com/abl-heatsinks/345ab1000b/heat-sink-1-c-w/dp/150016

Which is fine so long as you don't mind only using it when in an antarctic storm, otherwise it will get so hot that not only will your transistors melt, but the rest of the circuit will go with it.

[JacquesBBB]

It seems to me that without a fan, this will never work.

I have a Selectronic SL-1731SB linear power supply  ( 2 x 30 V - 3 A) which is regulated by 4 2N3055
as  in your project.

https://www.eevblog.com/forum/projects/changing-pots-to-multiturn-in-a-30v3a-psu/msg596332/#msg596332

The outside heatsinks seems to be much larger than the ones you are planning, and  I can tell you that this morning
I was using it at full power (3A), and I burned myself on the heatsinks when I touch them to see how
hot they were.

It seems to me that already for 5A, a fan is usually added.

[VK5RC]

Even with a crack hot (cool) heat sink, with a 2n3055 you are lucky to dissipate more than 20-30 W each, the junction to case co-efficient is a killer (as well as the heatsink itself) and then you have to derate the device a it gets hotter. In a PSU for some ham gear I had to use 12 of the buggers to dissipate 300W and even then some were getting pretty hot.

[smjcuk]

They do indeed get hot although I haven't managed to get one to go into thermal runaway yet. Package design is ancient (1950s) and inefficient so if this is a new project, I'd pick something else. The bum end of my Tek 453 has a couple of equivalent RCA parts pushing 100W and they get veeeeery hot. Well they did until the power supply totally blew out.

[PSR B1257]

but he says that transistor (semiconductor) doesn't act like a resistor...

So what?
Power is power, whether it is dissipated in a semiconductor or a resistor.

[Zero999]

If you use a centre tapped transformer and switch to the centre tap when the output voltage is low, then you can save a lot of power.

[Electric flower]

If you use a centre tapped transformer and switch to the centre tap when the output voltage is low, then you can save a lot of power.

I am already doing it that way, but didn't want to complicate my post... 

It's gonna be max. 25W per transistor, but i want to know max. capacity for dissipation

[T3sl4co1l]

Mmmm, would guess 20 or 30W for that heatsink, offhand.  Varies widely with how much temperature you want to allow it to reach, and orientation and airflow.

You *might* be able to reach 200W if you use thermally conductive epoxy to attach water cooling tubes to the bottom side.  (You still need to cool the water.)

In power levels of this range, it gets very attractive to learn how to design and build switching supplies.  It's a little bit more complicated, and opens up your knowledge base to a whole lot of modern electronics.

Tim

[SL4P]

I think the OP meant to post...
"How many powers can these 2N3055 dissipate..."   

[rob77]

I thin the OP meant to post...
"How many powers can these 2N3055 dissipate..."   

good catch but i think everyone got the idea what's the question
but you're right: "many watts" or "much power".

... or "manymuch powerwatts" 

[SL4P]

Yeah - you're right.
I was just spending a bored Sunday afternoon being pedantic.

[SL4P]

Another point is that in a PSU - you're possibly running a fairly constant load for periods - so the earlier suggestions of maybe 30W continuous on this particular heat sink (don't count the individual transistors yet).  Your ~250W supply target is going to need a LOT more cooling to run near capacity.

EIGHT times the size - assuming a perfect / linear increase in dissipation - NOT.
A heat-sink is rated by it's ability to dissipate power - one transistor or ten doesn't matter.

Naturally vented, or water/fan-cooled adds another efficiency - but make sure the fan doesn't stop - unless under temperature control.  These latter techniques just move the heat somewhere else... if the enclosure or room heats up, the heat sinking capabilities derate.
(Heat sinks are usually specc'ed in open air, or theoreticaly infinite dissipation environments - as the actual mileage varies for each application).

Audio amplifiers for example - generally run at maybe 10-20% of their rated output - even when really loud... with very short peaks (bass etc) that push the power limits, but a modest heat sink is able to handle and recover from these power 'surges'.  (Don't get me started on RMS vs Peak power in consumer amplifiers!)

[rob77]

btw.. i would rather go for a tracking switch mode pre-regulator - that solution would keep the dissipation more-or-less constant for the series pass element in the analog part of the supply. e.g. a switching regulator tracking the output e.g. 3-4V above the desired output voltage - that would give you 30-40W dissipation on the series pass element @ 10A.

[Zero999]

If you use a centre tapped transformer and switch to the centre tap when the output voltage is low, then you can save a lot of power.

I am already doing it that way, but didn't want to complicate my post... 

It's gonna be max. 25W per transistor, but i want to know max. capacity for dissipation

You need more taps on the transformer or two transformers so one could be switched off at the primary, at low voltages.

[Richard Head]

Can't you just lower the DC input voltage from 25V to say 8-10V? That would make a huge difference to the dissipation. You would have to change the transformer though.

[macboy]

Can't you just lower the DC input voltage from 25V to say 8-10V? That would make a huge difference to the dissipation. You would have to change the transformer though.

I assumed he was building an adjustable supply. He did mention a 25 V/10A supply then goes on to talk about the 5 V case.

A good solution to this power dissipation problem is a multiple tap transformer. Most pure linear adjustable bench power supplies use this as a means of crude but effective pre-regulation. You need comparators to sample the output voltage and switch in the appropriate tap(s) using relays.

[Zero999]

Can't you just lower the DC input voltage from 25V to say 8-10V? That would make a huge difference to the dissipation. You would have to change the transformer though.

I assumed he was building an adjustable supply. He did mention a 25 V/10A supply then goes on to talk about the 5 V case.

A good solution to this power dissipation problem is a multiple tap transformer. Most pure linear adjustable bench power supplies use this as a means of crude but effective pre-regulation. You need comparators to sample the output voltage and switch in the appropriate tap(s) using relays.

He's said he's already doing that:

If you use a centre tapped transformer and switch to the centre tap when the output voltage is low, then you can save a lot of power.

I am already doing it that way, but didn't want to complicate my post... 

It's gonna be max. 25W per transistor, but i want to know max. capacity for dissipation

Presumably this power supply is designed for higher voltages and the centre tap gives a maximum of 25V.

[planet12]

There was a post in the forums here with a method for determining the thermal resistance of unknown heatsinks; as I have quite a collection of them, I saved the link:

https://www.eevblog.com/forum/beginners/thermal-resistance-of-unknown-heatsink/msg293065/#msg293065