Class A Amplifier Heat Sink Requirements!?

[Robbie010]

First off, apologies for the long winded post but I’m really struggling here....

I’m about to build a 40W RMS Class A power amplifier split in to two small(ish) enclosures as monoblocks, but I am struggling to work out what heat sink setup I can get away with. I am trying to balance the design and look of the amplifiers with the requirements of the circuit design which means I don’t want to just bang in any old massive heat sink which may be overkill.

The amplifier design calls for 4 TO3 power output transistor per channel in a complimentary darlington set-up and as such will run fairly hot. My plan was to mount each transistor on its own 10cm x 10cm multi-finned heat sink to the outside of the enclosure, two on the left and two on the right and then cap all 4 transistors with a mini heat sink to maximize dissipation.

See pictures for the type of heat sink & cap I was planning to use.

Now, I am completely bemused by all the calculations / equations in trying to work out whether my planned heat sink setup will be up to scratch. One thing that I am unsure of is whether to account for the side walls on the enclosure as well as the main heat sink and the cap / mini heat sink in any calculations???

The side walls of the enclosures will have the heat sinks mounted directly to them and are 4mm thick aluminium, measuring roughly 30cm x 12cm, amounting to approximately 720 Sq cm (360 Sq cm per face) of surface area, which could potentially have a major effect on the dissipation of heat and subsequently the calculations / equations.

Can anyone comment on whether I should account for the side walls of the enclosure in any calculations??

Also, is there any easy way to work out whether my planned heat sink set-up will be sufficient?

Thanks in advance

Noob.

[Jwillis]

I wouldn't rely on the enclosure for added heat sinking. Just calculate for the heat sink only. You can get pre drilled finned TO-3 heatsinks for 4 transistors and they work very well at peak power of a 2N3055.You can never have to much cooling.

[GigaJoe]

Long cables to power output transistors are kinda bad idea. you may get oscillation. I would think about active heatsink like computer CPU, large fan usually noiseless, but an active airflow significantly decrease heat-sink size.

[TERRA Operative]

Here's a good resource on heatinks, written in an easy to understand way:

http://sound.whsites.net/heatsinks.htm


Also, as GigaJoe says, fans are good if you can use them, even a minimal amount of airflow can have a big impact on heat transfer.

[Calvin]

Hi,

Nelson Pass is rumoured to have said: " Just make ´em big and heavy". 

Anyway, some heatsink manufacturers like Fischer Elektronik offer guides on their websites for heatsink dimensioning.

regards
Calvin

[aheid]

The calculations are in theory rather straight forwards, I mean it's just a heat (current) source driving some heat resistance (resistors), leading to a temperature differential (voltage drop).

What I found most tricky was finding the heat resistance for various components and setups. So in your case, if the heat sinks are not exposed to the outside, you'd need to estimate the heat resistance of the enclosure to get an estimate of the steady-state temperature inside the enclosure.

Though if you find the thermal transfer coefficient for your wall material, you could consider each wall a separate heat resistance and put them all in parallel, with the amplifier board as a single heat source.

I'm no expert though but this is my understanding.

[Gyro]

40W RMS Class A in a small enclosure is a bit of a tall order. You may have to drop your ambitions a bit to fit your dissipation capabilities - It's going to be dissipating an awful lot more than 40W.

The calculation is very simple. Class A amps dissipate the most heat at idle (which also includes playing music at normal volumes). What is your total supply voltage +ve to -ve rails? What is the quiescent current setting for the output transistors? The maximum power dissipation is one multiplied by the other.

Then you can start getting onto heatsink thermal resistance requirements, maximum temperature rise etc.

[Robbie010]

40W RMS Class A in a small enclosure is a bit of a tall order. You may have to drop your ambitions a bit to fit your dissipation capabilities - It's going to be dissipating an awful lot more than 40W.

The calculation is very simple. Class A amps dissipate the most heat at idle (which also includes playing music at normal volumes). What is your total supply voltage +ve to -ve rails? What is the quiescent current setting for the output transistors? The maximum power dissipation is one multiplied by the other.

Then you can start getting onto heatsink thermal resistance requirements, maximum temperature rise etc.

Thanks all.

This is my first DIY build, the amp is actually a Nelson Pass A40 design.

The design calls for a transformer with a secondary voltage of 44V, which I believe makes the rail voltage 40V. I believe the quiescent current is 1.6A from what I’ve read.

Does this make the max power dissipation 40 x 1.6 = 64??

The enclosures are 220 x 120 x 300, so not tiny. Also, each enclosure will only contain one channel and the transistors are to be mounted on the outside of the case for maximum ambient dissipation.

[soldar]

The design calls for a transformer with a secondary voltage of 44V, which I believe makes the rail voltage 40V.

That can't be right. How did you come to that?

  I believe the quiescent current is 1.6A from what I’ve read.

That can't be right. How did you come to that?

  Does this make the max power dissipation 40 x 1.6 = 64??

I don't think so. Do you understand what "quiescent current" is?

I think you are being a bit too ambitious for your first build. If I were you I would build it with plenty heat sink and later you can decide. You can always reduce.

Some back-of-the-envelope calculations.

A 44 Vac secondary rectified would give roughly about 60 volts with no load and, say, 45 V under full load. Of course, it all depends on transformer, load, etc. but let us assume generously we have 44 volts rail to rail and, say 40 volts peak to peak output.

Assuming a sine wave signal output 40 Vpp then we have 40/2/1.4 = 14 Vrms which gives about 25 W on a 8 ohm load and 50 W on a 4 ohm load.

That is a sine wave signal. Real audio signals have a much lower power factor.

For 40 W into 8 Ohm I would say you need a bit more voltage rail to rail, say about 60 V under full load.

Have a look at the attached schematic which I have posted before in another thread.

[Benta]

One of my first projects many years ago was a Linsley-Hood 2 x 15 W class A amplifier. The heatsinks were enormous (finned, 10 x 25 x 6 cm), but still got so hot you could hardly touch them.

For class A, your quiescent current is a bit more than the peak signal output current. At 60 V supply voltage, you'll have around 55 Vpp output swing maximum = 27.5 V peak. For an 8 ohm load, this gives 3.44 A peak, ~3,5 A quiescent current.
This results in an output stage power dissipation of 210 W. And this is per channel! To keep the transistor case temperature at around 80 C at 20 C ambient, the heatsink system (per channel) including mica washers and all the other stuff, has to exhibit a thermal resistance of 0.3 K/W.

I suggest you visit the heatsink companies' catalogs to get an impression of the size needed for this figure.

[Gyro]

^ What they said.

I don't think it's possible for the voltage between the +Ve and -Ve rails to be as low as 40V  . 40W RMS is a pretty high output goal for a class A amp and couldn't be achieved by the figures you mentioned. Even modest Class A amps, like the Linsley-Hood, dump an awful lot of heat, as Benta mentioned.

Maybe you could post the schematic?

[soldar]

This is my first DIY build, the amp is actually a Nelson Pass A40 design.

I believe that is not a Class A but a complimentary pair class B amplifier.

[Benta]

This is my first DIY build, the amp is actually a Nelson Pass A40 design.

I believe that is not a Class A but a complimentary pair class B amplifier.

The Nelson Pass A40 is indeed a class A design, and from pictures it has enormous "cooling towers" with fans:

[Gyro]

It's a shame 2SK170s and 2SJ74s are unobtainable these days, you could get away without heatsinks at all  ...

The Nelson Pass 'Beast with Two Thousand Three Hundred and Fifty Two JFETS': http://www.firstwatt.com/pdf/art_beast.pdf

[Benta]

It's a shame 2SK170s and 2SJ74s are unobtainable these days, you could get away without heatsinks at all  ...

The Nelson Pass 'Beast with Two Thousand Three Hundred and Fifty Two JFETS': http://www.firstwatt.com/pdf/art_beast.pdf

That's one of the craziest designs I've seen. But the measurement results are impressive.

2SK170 and 2SJ74 are indeed available today from Linear Systems as LSK170 and LSJ74.

[soldar]

The Nelson Pass A40 is indeed a class A design, and from pictures it has enormous "cooling towers" with fans:

I stand corrected. It seems a bit crazy though unless you want it to double as a space heater.

[Robbie010]

Pass A40 schematics attached.

I have seen A40 builds with massive heat sinks and other with very similar setups to the one I was planning.

[Benta]

The Nelson Pass A40 is indeed a class A design, and from pictures it has enormous "cooling towers" with fans:

I stand corrected. It seems a bit crazy though unless you want it to double as a space heater.

No, you don't stand corrected
I've now read the Nelson Pass article on the A40, and I conclude that it's a high quiescent current class AB amplifier. He presents some waffle about "push-pull class A", which is nonsense in this context.

A pure 40 W class A will dissipate at least 210 W per channel, 250 W is more likely.

[soldar]

Yes, I took a quick look and saw the topology with complementary pair and assumed it was just another of the many of this type. I can't see much benefit to raising the quiescent current like that but it can be done in any of that topology, including the one I posted further up. Maybe someone will say it reduces distortion somewhat to which I would ask if it is noticeable and if it is worth it. Especially because if you want to increase quality the first thing I would do is get a regulated power supply.

[Refrigerator]

A pure 40 W class A will dissipate at least 210 W per channel, 250 W is more likely.

Surely it produces "warmer" music

[soldar]

Surely it produces "warmer" music

Best post / poster combination

[Audioguru]

This 40W amplifier produces 100W of heat all the time and should be banned for wasting so much electricity and air-conditioning.
Class-AB amplifiers perform fine without the waste.

[David Hess]

This is my first DIY build, the amp is actually a Nelson Pass A40 design.

I believe that is not a Class A but a complimentary pair class B amplifier.

I wondered what he was talking about with class-a and complementary.  That is class-ab where 40 watts in a small enclosure is very feasible.

[soldar]

I had another quick look at the article and I find it unconvincing in several ways.

It uses a center tap 44 + 44 Vac transformer, rectified, and later says it is using 32 + 32V. That does not add up. 60 ~64 Vdc rail to rail sounds about right so I would say the transformer spec is mistaken.

The high quiescent current means the need for four power transistors on a huge heat-sink. This is unnecessary and the article says you can decrease the quiescent current and the amp will perform well. So why not do it? That removes the need for four power transistors and only two will be needed and much less heat sink.

If I looked at it slowly I am sure I would find more problems and inconsistencies. Not to mention a few typos. (If the amplifier can be "fumed on" instead or "turned on".)

That project does not inspire confidence in me.

These complementary pair amplifiers are classics and there are a million variations out there but they are basically the same idea. I just can't see any benefits to unnecessarily increasing the quiescent current unless it is heating the room. 

Look at the similar design I posted upthread; it says to adjust quiescent current to 40 mA which at 60 volts is 2.4 W quiescent power.  That makes much more sense.

This 40W amplifier produces 100W of heat all the time and should be banned for wasting so much electricity and air-conditioning.
Class-AB amplifiers perform fine without the waste.

What he said. Bad design.

[Gary.M]

The Nelson Pass A40 is indeed a class A design, and from pictures it has enormous "cooling towers" with fans:

I stand corrected. It seems a bit crazy though unless you want it to double as a space heater.

No, you don't stand corrected
I've now read the Nelson Pass article on the A40, and I conclude that it's a high quiescent current class AB amplifier. He presents some waffle about "push-pull class A", which is nonsense in this context.

A pure 40 W class A will dissipate at least 210 W per channel, 250 W is more likely.

Nothing Nelson Pass has to say is waffle. Class A operation is solely determined by the quiescent current vs load demand, not the topology. In this amp the peak output current at which the amplifier comes out of Class A will be approx 2 x the idle (bias) current. This is where one side of the output stage cuts off. You could bias it into Class B, Class AB, or Class A on a continuum.

Sent from my MI 8 using Tapatalk