Re: Heatsink fin area for given temp rise.


"Patrick Turner" <info@xxxxxxxxxxxxxxxxxx> wrote in message
news:47A478B2.BFE8A725@xxxxxxxxxxxxxxxxxxxxx


Trevor Wilson wrote:

"Patrick Turner" <info@xxxxxxxxxxxxxxxxxx> wrote in message
news:47A328DE.A55BA0F0@xxxxxxxxxxxxxxxxxxxxx
As work progresses with the pair of 845 SET power amps,
some serious consideration to getting rid of heat
in 2 cathode resistors and several others in the circuit, resulting in
a
total od 55 watts
of dissipation.

At first I tried normal white ceramic wire wound R
all operating at 1/4 or less of their wattage rating and mounted on Al
angles
but when the prototype was trialed the Al angle small sized heatsinks
must have gone up to nearly 100C...not so good.
On a day here where my shed was 36C inside, the whole amp with R
underneath the chassis
got fairly warm; too warm to my liking.

So I made a heatsink 185mm high and 210mm wide with a 3mm plate and 2mm
plate with R sandwitched between them
and silicone adhered and with silicone paste to each plate.
On the back plate facing rearwards, I fitted 12 20mm x 32 mm x 1.6mm
angles running vertical,
and finally I could touch the hottest part of the sink for 5 seconds
without scorching myself,
and my thermometer indicated 47C max for an evening of 23C ambient.
The heatsink parts were all well machine screwed together with high
temp
silicone.
The beauty of making one's own heatsink is that you get the size you
want,
and the shape wanted amd you are not addapting a less than perfect
asian
made HS.

The useful info I should share with you is the final area per watt
required for dissipation of devices.

I ended up needing 40square cm per watt to get no more than 24C
heatsink
T rise.

The total area was measured of the Al. This is surface area of front
back of metal plates holding resistors
between them, and of both sides od fins.

Fins are vertical, and have natural unobstructed air flow not dependand
of fans.

**I trust they are anodised black, as this will improve cooling
efficiency
somewhat.

Blacking the heatsink makes them only 10% more efective; ie, instaed of
24C t rise.
maybe you get 22C T rise, not a huge difference.

HS are blacked mainly for appearances sake.....

**Nope. They're black to ensure highest possible efficiency. Manufacturers
could choose any colour they like, but most choose black, because it works
best. Your figure of 10% is close to the mark. At 100oC that makes a 10oC
difference.



The resistors are spread out evenly over the available 210mm x 170mm
plate area,
and values carefully worked out so that the wattage per unit area was
about equal.

I tried a range of fans though, and powered them so they were nearly
silent.
The 12Vdc types are the best, but gees they are fragile!
even when running slow they can still be heard. So although a couple of
slow running 80mm fans
did cool the heatsink so I got only 15C rise, I decided I couldn't use
any type of fans.

**The best fans (by far) are 240VAC types. They are tough, quiet and easy
to
control.


Sure they are tough...

But you CANNOT run then full bore because the noise is INTOLERABLE.

**That would depend on the fan. I have an ancient Boxer fan (made in the US)
which is amazingly silent at full RPM. Having said that, the noise is
proportional to air flow VELOCITY.

OK with a an RF transmitter or industrial or PA amp, but NOT
something in someone's lounge with audiophiles with perfect hearing
able to hear a pin drop or low level fan whine or transformer hums.

**That would depend on the fan, how it is mounted and how it is used. I've
measured a fan in a piece of audio equipment (at 4:00AM) which was not as
noisy as the transformer in that equipment. The transformer was a potted,
toroidal type and was VERY quiet. Of course, the fan was large, 240VAC type
and mounted on compliant rubber.


I tried using a few spare 120mm dia fans here with 240, and they are
effective
when fed only 95Vrms instead of full blast 240, so you have to series a
pair
and use dropping resistance They are plainly audible even at low speed,
and
trying to get them slower and quieter is self defeating and an exposed
fin passive HS is then
just as good if the fin area is large enough.

**Here are some specs I gleaned from a single manufacturer (Papst) for two
fans. One, 80mm and one 199mm. Both are 230VAC types.

Size Air flow (L/min) dBa@1m
80 13.7 24
119 22 25

These fans are rated by Papst as "super low noise". You'll note that for a
miniscule increase in SPL (1dB), with the 119mm fan you get nearly double
the air flow of the 80mm fan. Decrease the speed of the fan to match the
volume of air from the 80mm fan and the noise will be dramatically lower. My
rough measurements in the past indicate an approximate 10dB fall in noise
levels. An excellent trade-off. BTW: Looking at 80mm 12/24 Volt DC fans from
Papst, indicate noise levels in the range of 23 - 32dB. An insignificant
decrease in noise, for far less controllability.


My experience with ME amps and their fan control and heat control
leads me to think I'd always rather have a passive HS like Quad 405 etc.

**All very well, except the you forge several things:

* The Quad is a Class B amplifier.
* The ME850 is a very heavily biased Class A/B amplifier, with around 16
Watts of Class A power.
* The ME is SPECIFICALLY designed to operate at a constant temperature of
between 50 - 60oC.


The ME850 has a nice fan system, surrounded by foam.
In the last one I repaired the foam is on its last legs and will have to
be replaced in
a a year perhaps. Foam rots and falls to bits, and the foam in the ME
isn't
an exception to this.

**Indeed. Which is why I ALWAYS replace the support foam, if I consider it
has less than a few year's life left. I see no point in gouging customers
unnecessarily. This way, they only pay for one labour charge, rather than
two. In normal operation, the support foam last between 10 and 15 years. Not
an unreasonably short period of time, IMO.


Modern DC fans are not so easy to control quietly. The motor is
usually a brushless, AC type. The difficulty arises with the controller.

Just alterning the input Voltage can cause 'cogging' which translates to
noise. Best to use an AC type, with a suiable controller, or even a
resistor.

The fan noise is caused mainly by the speed of blades whipping tiny eddy
air currents.
Its basically PROPELLER NOISE that we have to face, not "cogging".

**You have cogging problems with DC fans.


I tried resistors to drop the AC or the DC voltage.

And, like ALL fans, the bigger the quieter.

I found a pair of 80mm 1vDC fans were much quieter than ONE 120mm fan.

**That would depend on the choice of fans. Two good fans will be quieter
than one crappy fan.


The dc fans waste far less resistor feed current, and ran at 6V OK
for reducing T rise by 50%. But that's all.

This is for a class A amp so once up to running temp the heat is STABLE.
Its not a class AB amp where Pd of the devices and HS swing wildly due
to signal levels.

**I understand that.




A large, slow revving
fan, will be quieter and move as much, or more air, than a smaller type.
In
general. I have a bunch of 175mm fans, which I operate at 50% of their
rated
Voltage and they are inaudible. Mind you: They are old, mad ein Japan and
very high quality.

I have a few old fans; none are quiet, all are audible, and a passive HS
is the best solution for me. I have no room left on the chassis to
accomodate a
suitably formed metalwork plenum to flow the air without turbulence.

**Fair enough. I wasn't commenting on your final solution. I was commenting
on what I feel to be your misconception about certain fans and their noise
levels. You say it yourself, in your reference to turbulence. Lower velocity
air creates less turbulence. Lower velocity air, with higher volume can pay
big dividends.




So where one was building a 25 watt class A SS amp, one might have to
dissipate 55Watts,
so you'd need a total Al fin and plate surface area 55 x 40 = 2,200
sq.cm to get a T rise of about 24C.

**Which is where using a large, slow revving fan, makes perfect sense.

Sure, but all ther fans I have tried are too noisy for anyone I know.

Its better to use the extra one kG of Al in a custom made amp.

**And again. I was not commenting on your final solution. That is entirely
up to you. I was merely correctly your misconception about fans. Big, slow
revving fans are quieter than fast revving small ones (all things being
equal).


High bias Class A or AB amps will not have varying heat once up to
working temp. Some examples I have seen are plain awful like Sugden who
have horizontal HS fins, YUK!! when will these doltheads learn??

**They probably won't. Musical Fidelity have made similar stupid mistakes.




This is what I have for my resistor pack on the mono 845.

The placement of mosfets and any hot parts must be so they are spread
out as much as possible or else
you get devices being hotter than you'd want if they are clustered
together
and a heatsink running too hot near the devices and too cold further
away from the devices,
so the devices are stressed and the heatsink would be inefficient.

**Which is why fan cooling makes perfect sense.

We will have to agree to disagree Trevor.

I just spent a day fucking around with fans and decided they are a
miserabally Royal Pain In The Arse.

**A whole day! Gee! I've spent weeks mucking about with fans. Big, slow
revving fans are quieter than small fast revving ones (all things being
equal).


In hindsight, it'd have been better to mount the 845 and other tubes in
a glass chimney tube,
perhaps made from 4 rectangles of glass, like a box without top or
bottom, then have incomming
air to the tubes flowing past the HS under the amp. Total flow would
still cool the tubes, but it makes the
whole design look unconventional and is horribly expensive to make.....

I am making the eqiuivalent of top performance class A SET equal to
anything
made by KR Audio who make whole amplifiers not just vacuum tubes,
and I am doing it this time for less than 1/2 their price, and it
includes the same R&D thy'd have used
for a batch.

I figure I should be getting twice the KR price because they are
specials but
nobody ever wants to pay thay price.

I could do fabulously MUCH BETTER amplifiers if only the bastards out
there
would pay better.





I don't like hot running resistors in any amps I build, and if you run
a
10W wire wound
resistor which is say 50mm long, with 10mm side size, its T will be
high, maybe 100C++,
and it simply won't last very long, because the welds fatigue or ther
ceramic cracks
or the wire finally oxidizes, hence 2.5W is the max I ever use with any
10W resistance.
Strung netween two tags on a board it still gets hot, and where you
have
lots of them,
all that heat has to go somewhere........

**Most power resistors are rated for 200oC for long periods. Having said
that, I agree with you. I rarely allow resistors to reach 40% of their
maximum rating. 25% is much more sensible. Glue the resistor to the
chassis
and things usually last a long time.

Standard fare is to flue rectangular ceramic bodied R to a chassis and
have them set up
to run at 25% of their power rating. This is OK if you don't have too
many R
and Pd is only a few watts

I have 55 watts to get rid of.....

KR Audio faced the same bother in their amps using their tubes so they
cheated by using
as much CRAPPERAMA CIRCUTRY ahead of the tubes, ie, the whole input
stage and as much as possible
is SS and on generic PCBs. They secumbed to crumminess.

They still have a heatsink on the chassis.





The vitreous enamleled ww R and the AL clad R ARE GOOD, and can be
allowed to run hot,
say above 50C but I don't like it at all.
And they are expensive. The AL clads need a heatsink.

**They are, indeed, very good. They're more compact too. Still, you can
bolt
them to the chassis, which is prefereable to glue. And, they can cope
with
200oC no problem.

Yes, but 55 watts of hot resistors still have to be placed somewhere
on the cramped chassis of mine; if under the chassis its like having a
55 watt light bulb
there, or a pair of enclosed tubes running at 28 watts each.
Things just get hot.

The HS keeps the R cool, and I don't tempt fate trying to have hot
running R.




Doing it the way I have and limiting heatsink T to 24C rise
will mean that elements in the R would probably be at around 30C above
ambient
which is quite acceptable for WW R.

**The real risk with high power resistors, is to the surrounding
components.
The resistors, themselves, are pretty tough.

The vertical casket full of R I have made is above the back plate of the
amp
and away from anything that can ever become too hot.

I tried it out today with shed C = 28C, and got HS at 52C. Actual R
might be
62C.

Reliability in general for R drops 5% for each 10C rise in Temp.

So at 100C, the R are 50% unsafe, and at 200C, VERY likely to fail.

Solid state device reliability falls 10% for each 10C rise.

**Nope. A common misconception. Capacitors, maybe, but not semis. Operating
semis at a constant elevated temperature can pay big dividends in
reliability. This research was done decades ago by RCA engineers. For that
reason, RCA chose to use steel rather than aluminim for their TO3 devices.
It is the thermal cycling which kills the devices, not the absolute
temperatures (assuming the devices are kept below (say) 100oC.


So with 100C rise, BJTs etc are 100% likely to fail soon.

**100oC is too hot for semis.


And these things do fail. My shed if full of amps that
have failed due to heat resulting from failed bits and peices,
and very often it goes back to the designer who reliaes ona Hail Mary or
Allah to
prevent failure.
Dry joints are very likely when the leads, tracks, boards get hot; heat
causes expansion differentials
on PCBs and some things are not mounted to allow for the microscopic
movements that occur.

**Sure. And, strangely enough, I rarely find dry joints on hand soldered
PCBs and American machine soldered equipment. I don't know why.


Making things rugged, easily serviceable, simple as possible,
but with excellent sound, and good measurements costs money.

**Yup. And selecting fans takes more than 1 day.

Trevor Wilson


.

Quantcast