2N3055 Heat Output

[Dazmond]

Hi Folks,

I've read through a few other posts on this issue, but just to clarify is it normal for these to run burny fingers hot even at idle?

The bias settings are exactly as stated in the service documentation, I even went ahead and replaced the trimmer pots so they're nice and accurate.

I've been through every single component in the power amp and the preamp, other than a few leaky caps and a few resistors that needed replacing everything is as it should be.

All measurements match up to those in the service manual.

The transistors are all new as 1 of the originals was faulty and the other 3 were heavily corroded (the unit had been stored in a loft). The replacements came from RS so they should be decent quality.

Using my temperature probe on my meter I'm getting 56c at case when idle, according to the data sheet they're fine between -65c - 200c, but this just seems a bit excessive to me?

It's clear from the design they've taken a number of steps to help eliminate self oscillation so I don't think this is the issue. There isn't any DC creeping through to the speaker connections either.

Other than the above the amp appears to function fine and dandy.

The amp is from a Hacker Tribune, was it just an added design bonus that you could fry eggs on the back of them?

Many thanks,

Darren.

[Boater Sam]

Always did run hot, efficiency of semiconductors is much better with modern devices, these are old school lumps.
Good heatsinks and heatsink compound required.
In my youth ( Hah! ) at Ferranti's we built hugh amps with 3 phase mains supplies, mega outputs, for application labs work. We had no speakers that would take the full load so we used electric fire elements for tests. We did really boil kettles on the TO-3 can tops!

[Dazmond]

Blummin eck and I thought these were an extreme, that's serious stuff!

Thank you for putting my mind at ease - they have some compound on there (I even used the fancy stuff), would the difference between silicone rubber and mica make any difference to the heat transfer for the insulators? I'm currently using silicone rubber.

Many thanks,

Darren.

[Boater Sam]

I suspect that silicone is inferior to mica but not sure.
We used isolated heatsinks and no insulators under the cans. There were lots of them!
A computer fan adding air flow over the heatsinks would help a lot.

[Dazmond]

That being the case I might give mica a go, if nothing else just to say I've done everything I can, maybe I should attach a pair of oven gloves to a chain on the side of the unit for moving purposes

Many thanks,

Darren.

[MrBungle]

Power transistors can run crazy hot without dying. Rule I use is the sizzling finger rule, as described in The Art of Electronics. Lick your finger and touch the transistor. If it hisses it’s too hot!

(Does not apply to small signal transistors)

Really you need to just make sure the transistor is within SOA. Temperature only tells you how hot it is - there are other failure modes.

Edit: safety note! The collector and therefore case might be connected to Vcc so don’t do this if there is an electrocution risk!

[camtechman]

This thread has just triggered a memory when I worked at SGS in the late 1980's.

Not that I was in the technical division but the grape vine spread across the company and even got coverage in the monthly company magazine.

The gist, as I remember it was the drive to produce high end, high spec and "Beefy" 2N3055's for the audio & power sectors. The sales increased rapidly as the marketing & sales department went out to the relevant companies.

I wonder if this improved device ever solved running problems ?

[jamesperrett]

I'm probably stating the obvious here - but you do have some heatsink compound between the transistor, washer and heatsink don't you?

Is the amplifier running in class A?

[Refugee]

I was about to ask if it is class A too.

[Dazmond]

Yes I added a small sliver of compound either side of the insulator, you're supposed to use these silicone ones without compound but I've read that people prefer to use a little which is what I've done, but I think I may try good old mica.

It's class A with a 15ohm load (which it currently has) and class A-B with an 8ohm load.

Thanks,

Darren.

[camtechman]

I just remembered that, because of the magazine article, non tech staff were encouraged to go to the development department to experience the new device (well it was something to do in your lunch break !).

We were ushered in and told not to touch anything, the look on the faces of the techy geek guys in meeting other 'normal' humans was amazing !

Anyway, we saw banks of these 2N3055's on panels and encouraged to hold the palms of our hands no closer than 2 feet away from the panels (& NOT to touch them) and feel the radiated heat.

The manager even told us that he warmed his pack of KP salted peanuts on them "As they taste better" !!

Thrilling Eh ?

[Dazmond]

So by the looks of some of the comments here the high heat is pretty normal?

It's crazy they get this hot!

Thanks,

Darren.

[cmjones01]

Originally Posted by Dazmond

It's class A with a 15ohm load (which it currently has) and class A-B with an 8ohm load.

That's why it's getting hot, then. To run in class A with a 15 ohm load, the output transistors will have to dissipate the full power that the amplifier would be able to deliver to a 15 ohm load, even at idle. That's probably quite a lot of watts. The laws of thermodynamics mean that the transistors will get hot.

The only time to worry is if the heatsink the transistors are mounted on is significantly colder than the transistors, after the whole assembly has had 10-15 minutes for the heat to spread out. If the transistors are very much hotter than the heatsink, then it's time to look at the insulating washers and heatsink compound.

I've always found the silicone type of washer perfectly OK. What you're describing, a case temperature of 56C in a 15-ohm class A amplifier, sounds absolutely normal to me, and nothing to worry about.

Chris

[Dazmond]

Thank you for that Chris, I've not yet let it run that long as I didn't want to cook them, but that being the case I will let it sit on an 8ohm load for 15 minutes and compare the difference between the transistors and the heatsinks.

In your opinion would the silicone washers perform better with or without a little compound?

Thanks,

Darren.

[ukcol]

Originally Posted by MrBungle

Power transistors can run crazy hot without dying. Rule I use is the sizzling finger rule, as described in The Art of Electronics. Lick your finger and touch the transistor. If it hisses it’s too hot!

(Does not apply to small signal transistors)

The above may be valid for silicon power transistors like the 2N3055 but not for germanium ones found in earlier equipment.

['LIVEWIRE?']

A general rule is that all Class A amplifiers draw a steady current from their power supplies which, as Chris notes in post #13, is that needed to run the amp. at full power into it's rated load, so they will run hot. How hot depends on the number & type of output transistors use, the size & type(s) of heatsink used, etc., but I trecall repairing one quite a few years ago which you could literally boil water on (don't recall the make or model, though)

[Herald1360]

Silicone washers are used "dry" since in themselves they are capable of filling the uneven air gaps between the device and the heatsink which is the function of the compound.


Whether they're better than mica depends on their spec and the mica thickness. Generally 0.003" mica + pigeon s**t type compound will be a bit better and a bit messier! Insulated heatsinks, compound and no washers is best possible scenario.


Devices' specs will tell you the thermal resistance between the case and junctions, so a case temp measurement can be worked back to junction temperature. 2N3055s will stand junction temps up to 200C, back off to about 150C for reliability. That generally equates to case temps far too hot for human sensitivities (about 55C is seriously ouch hot).


Sooo, if the spit doesn't sizzle it's fine, if it does it's time to measure things and do the sums!


Useful reference here:


https://www.fairchildsemi.com/applic...AN/AN-4166.pdf

[kalee20]

Originally Posted by cmjones01

That's why it's getting hot, then. To run in class A with a 15 ohm load, the output transistors will have to dissipate the full power that the amplifier would be able to deliver... even at idle

In fact it's DOUBLE, plus a little bit!

A 10W Class A amplifier is at most 50% efficient, thus 10W into the load, 10W lost in the output devices, running flat out.

That's 20W from the power supply.

When no signal is there, all the 20W is dissipated by the devices, so they have to be capable of twice the output power.

In practice, the efficiency can't be 50%, it is slightly less, thus you have to be prepared for a bit more power draw (and hence loss).

[Dazmond]

Thank you all for your replies, I will put some mica's in there and report back with my findings!

efficiency of semiconductors is much better with modern devices

They all dissipate the same power with X volts across and Y amps through. True that new devices and new designs can have lower dissipation, even with new devices class A is class A. One great thing about old transistors (such as the original 2N3055) is that the die is enormous giving a large SOA (safe operating area). As said, the finger sizzle test is a good one, a small phutt is fine.