heatsink size for 2n3055g transistor

[little_carlos]

Hi
Im going to use a 2n3055g npn transistor for my power supply, it is going to handle up to 32V maximum and up to 3 amps,
is there a way to know how big should the heatsink be for this application? i have a 5.5cm x 6 cm heatsink, is it safe for this application?

[tszaboo]

First: Does it have to handle ~100W of dissipation, or less? If yes, you have to look for a heatsink with maximum 1K/W thermal resistance. The heatsink you described is probably twice of that with natural convection.
But we need more data to help you.

[wraper]

Even if you will find heatsink so huge to dissipate 100W, single 2n3055 won't be able to handle 3A with 30V voltage drop on it.

[little_carlos]

this is the heatsink that i want to use

it is 5.5 cm x 6cm x .5cm

[liquibyte]

You're going to need to parallel at least 2 2N3055's to handle that current at lower voltages if you're building a 0-30V variable supply and you're going to need significant heat sinking on top of that as well.  I just built a dual 0-30V 0-3A supply using 4 2N3055's and used this heat sink to mount them.  I'm currently working on the displays and temperature circuitry because these will be actively cooled using 2 80mm fans mounted to that heat sink.

Edited with a pic.

[Holograph]

this is the heatsink that i want to use
it is 5.5 cm x 6cm x .5cm

That heatsink is insufficient for this purpose.

[VK5RC]

I think you need to do some calculations,  the highest power dissipation is usually at high current,  low voltage out,  calculate power dissipation required,  you then look at the efficiency of heatsink,  heatsink to transistor junction and finally collector to case (all usually in degrees C/W) ,  this will tell you what temp the transistor will be at. Then look at the de-rating curve (on data sheet) . It is pretty hard to dissipate much more than 20 to 30W per 2n3055, even with a pretty good heatsink with a fan.

[liquibyte]

This is a long read but filled with very good info on heat sinking power devices and the calculations needed to get you where you need to be.

http://sound.westhost.com/heatsinks.htm

While people here can give you answers to your questions, if you do the work yourself, you'll be able answer your own questions and never have doubts about the answers.

[grumpydoc]

It is pretty hard to dissipate much more than 20 to 30W per 2n3055, even with a pretty good heatsink with a fan.

Yes - in fact this surprises people who see the 115W rating and assume that's what the transistor can handle forgetting that the 115W is at a case temperature of 25^oC.

The difficulty comes from the junction-to-case thermal resistance which, for the 2N3055, is 1.52^oC/W. The max junction temperature is 200^oC - 115*1.52 is 174.8 so if the device is dissipating that much power the case can be at no more than 200-174.8 or 25.2^oC

If you add in the typical resistance of a mica washer of 1^oC/W and maybe an 0.5^oC/W heatsink (i.e fairly meaty) then there's going to be around 3^oC/W junction-to-ambiant resistance. With an ambient temp of 25^oC that translates to 60W even with a substantial heatsink.

Personally I don't think running the junction at the maximum temp is a good idea so you need to allow some margin - 20-30W is perhaps on the low side but it's definitely a challenge to get more than 40-50W per 2N3055 and keep the junction below 150-170^oC.

MJ15015's are better but about 3x the cost.

This is a long read but filled with very good info....

http://sound.westhost.com/heatsinks.htm

+1

[rsjsouza]

If you have a transformer with taps that help reduce the voltage dropout on the regulator across the voltage range of your power supply, I think you can use an existing design as a reference. As an example for the size of the heatsink and the number of 2N3055s, check this post I did a while ago. It also has some links to app notes that show some heatsink calculations.

[jlmoon]

Hello,

Another concept would be to use (-) side regulation so you can mount those transistors with the collector at the ground potential..
you will get a much better thermal efficiency by this method and this allows you to dissipate more power into your load instead of in your active components. 
There is no reason or law that dictates you can't regulate your supply on the (-) / ground rail of your supply. 
Just a little FYI

Personally, I would loose the linear concept and go switch mode, much better all the way around..
Hope this helps!

JLM

[grumpydoc]

Another concept would be to use (-) side regulation so you can mount those transistors with the collector at the ground potential..
you will get a much better thermal efficiency by this method and this allows you to dissipate more power into your load instead of in your active components. 
There is no reason or law that dictates you can't regulate your supply on the (-) / ground rail of your supply. 
Just a little FYI

That might be useful in some limited circumstances but it's not actually all that great an idea.

If you have a metal case you will want that connected to mains ground for safety reasons. Yes, you could use a PNP transistor and regulate the -VE rail but now your -VE output is a few V above ground (and your +VE rail some V above that).

This isn't a problem is the circuit that you are powering has no connections to other equipment but if it does you will find that connecting the "0V" connection to ground via the 0V/ground connection on some other equipment will short your pass transistor and probably let the magic smoke out of your project.

You will also find that you cannot stack two power supplies designed like this to give +VE and -VE rails or you will let the magic smoke out of one (or probably both) power supplies.

Commercial PSUs have floating outputs and a separate ground connection for a reason.

[richard.cs]

If you have a metal case you will want that connected to mains ground for safety reasons. Yes, you could use a PNP transistor and regulate the -VE rail but now your -VE output is a few V above ground (and your +VE rail some V above that).

This isn't a problem is the circuit that you are powering has no connections to other equipment but if it does you will find that connecting the "0V" connection to ground via the 0V/ground connection on some other equipment will short your pass transistor and probably let the magic smoke out of your project.

Yes you might want to tie the output to mains earth but there's no reason you have to connect to the most negative point in the supply. Let the internal power rails flap and ground the output. I believe the original suggestion was to use NPN transistors common-emitter in the negative side - this allows you to drop fewer volts across them than is otherwise possible without the use of a higher voltage base supply. One design that uses this is here: http://ludens.cl/Electron/Ps20/Ps20.html - as the voltage drop above the pass element increases the internal negative rail gets pushed down below mains earth but that doesn't affect the operation of the circuit.

You will also find that you cannot stack two power supplies designed like this to give +VE and -VE rails or you will let the magic smoke out of one (or probably both) power supplies.

Commercial PSUs have floating outputs and a separate ground connection for a reason.

Certainly you cannot stack two power supplies which have an internal mains-earth to output connection, but with good design* it isn't necessary to make that connection. You are of course left with the heatsinks floating at different voltages but with suitable mechanical design there's no reason they should ever touch.

*Split bobbin transformer, double-insulation of the mains part. Essentially making sure that there are no circumstances where the mains can ever get connected to the output. And earthing the metal case does not require you to earth the output.

[kg4arn]

Since you guys are talking about the pros and cons of the high vs low side regulation:  take a look at this thread  https://www.eevblog.com/forum/beginners/pnp-vs-npn-as-pass-device-in-psu/msg406433/#msg406433  where I don't think the OP's original question was fully answered (or it was and I just don't understand it).

So what are the tradeoffs between low vs high side regulation?  Is it just arbitrary? 

I would add the restriction that the outputs are isolated unless physically connected to earth and that either the + or the minus output might be user connected to physical earth.

[grumpydoc]

Yes you might want to tie the output to mains earth but there's no reason you have to connect to the most negative point in the supply.

No, indeed. But the suggestion that jlmoon made was to tie the collector of the pass transistor to ground so that you don't need to isolate the TO3 can from the heatsink.

That isn't such a good idea because it becomes impossible to tie either rail to mains ground.

I think quite a few of the linear bench PSUs have a basic design similar to the one you reference.

[jlmoon]

Thank you Richard for seeing through my process,  Grumpy,, I stand corrected .. in my statement I used the term "Ground potential"  simply forgetting (-) and (ground potential) are not the same. 
This negative side regulation is done all the time without any trouble what so-ever.. abet confusing for some tho.. .. works really good when you need as much smoke as you can get from a linear design. 


JLM

[jlmoon]

"Since you guys are talking about the pros and cons of the high vs low side regulation:  take a look at this thread  https://www.eevblog.com/forum/beginners/pnp-vs-npn-as-pass-device-in-psu/msg406433/#msg406433  where I don't think the OP's original question was fully answered (or it was and I just don't understand it).

So what are the tradeoffs between low vs high side regulation?  Is it just arbitrary?

I would add the restriction that the outputs are isolated unless physically connected to earth and that either the + or the minus output might be user connected to physical earth."

wish I knew how to embed these 'quotes'.. That thread explains it very well!  I was just thinking off the linear line graph just a little!.. 

[rdl]

wish I knew how to embed these 'quotes'.. That thread explains it very well!  I was just thinking off the linear line graph just a little!..

You copy the text you want to quote, then paste it into your reply. In your reply, click and drag to select and highlight the quoted text and then click the "Insert Quote" button (looks like a cartoon balloon, next to the eBay button) and it will automatically add the quote tags.

You can also just add the tags manually. It's quote and /quote inside square brackets (no spaces), [likethis] and [/likethis]

[jlmoon]

 

You can also just add the tags manually.

ah.. just helps to ask..

Still a noob usr here..

Thank you