[Elecraft] KPA-500: Mitigating Thermal Expansion Generated Noise

[Clay Autery]

OK....  I don't have a KPA-500 to take apart and play with, but I just 
spent about an hour reviewing the images in the assembly manual, et al.

Yep.... It is likely the thermal expansion and contraction of the 
relatively thin aluminum Z-bracket.  The Z-bracket appears to be acting 
as a heat sink/spreader/connector between the 4 each rectifiers and the 
heat sink.  There are a NUMBER of potential issues in this setup that 
may be causing the noise via thermal cycling:

1) In high demand conditions, the rectifiers are dumping a sufficient 
amount of heat into the Z-bracket to cause the aluminum to increase in 
temperature enough to expand just the amount required to "tink" during 
thermal cycling.... because...
2) The Z-bracket is relatively thin...
3) The Z-bracket is constrained from expansion top to bottom by its 
connection to the top and bottom of the chassis.
4) The rectifiers are secured to a "tab" that sticks out of the end of 
the Z-bracket that is NOT constrained from expansion in either the 
top/bottom or side/side directions.  It reminds me of the little metal 
tabs in old thermostats designed to change position with temperature in 
order to switch conditions on the thermostat...  also reminds me of the 
temperature activated auto-resetting circuit breakers....

The heat imparted to that tab by the rectifiers causes the aluminum tab 
to heat up and expand....  that movement is transmitted to the rest of 
the Z-bracket....  the Z-bracket is constrained by its connections to 
the top/bottom chassis plates, AND in a thin row to the heatsink...  The 
unequal distortion of the thin metal produces a sound under thermal 
cycling...

There are a NUMBER of factors that may have prevented this.... Thicker 
material in the Z-bracket would have provided more rigidity that MAY 
have allowed it to resist the distortion events long enough to smooth 
the thermal cycling....  More importantly, there would have been more 
thermal mass to spread the heat dump from the rectifiers.  But we can't 
do anything about that part....  but perhaps some mitigation (see below).

The rectifier Z-arms could have been extended to include the rectifier 
tab so that it wasn't a small/thin aluminum tab floating in space taking 
heat hits from 4 rectifiers....

The Z-bracket has a relatively small thermal connection to the 
heatsink...  Had the contact patch between the Z-bracket and heat-sink 
been larger, the heat load MAY have been able to disperse into the 
heatsink fast enough to reduce the thermal distortion sufficiently to 
eliminate the sound under high demand through the rectifiers....  
maybe.  (IAW...  attached to a FIN face, rather than the base strip)

Other factors I won't address at present.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Mitigation:

OK... there are a number of things that might be tried to mitigate and 
perhaps eliminate the sound...  The bottom line (I think) is that we 
need to provide a better thermal pathway for the heat from the 
rectifiers THROUGH the Z-bracket to SOMEWHERE ELSE at a rate that will 
prevent the Z-bracket from distorting to the "tink point".  <smile>

1) Connections to the chassis:  The chassis has the largest relative 
contact patch with the Z-bracket, and has thermal mass that opens to the 
exterior for heat rejection.  IF we can improve the thermal conductivity 
of the interface between the Z-bracket arms and the chassis panels, we 
can INCREASE the rate and magnitude of the heat dump into the chassis 
panels.  The idea is already to provide continuity between the Z-bracket 
arms and the top/bottom panels since the paint is masked to provide a 
metal to metal contact.  It also attached to the rear panel at two 
points as well...  Here are some things that can be done to increase the 
thermal conductivity between the Z-bracket and the chassis panels.

a) Reduce the interface gaps at top, bottom, and rear attachment 
points.  How smooth is the aluminum plate on the arms and rear mounting 
bosses?  How flat is it?  One MIGHT consider lapping those areas FLAT 
and SMOOTH with a true surface/fixture and progressively finer wet/dry 
sandpapers to a grit size of 1500 or so...  This WOULD require 
disassembly of the amp such that you could work on the Z-bracket by 
itself.  And lapping the areas flat requires that you be able to do so 
squarely.... likely requiring a fixture.  I would do this.  But I've 
been lapping processor cases and heatsinks flat for many years.

b) Further improve the interfaces both thermally and electrically by 
making sure the corresponding areas on the chassis panels are smooth and 
"bright"  (no paint, tape residue, overspray, or any other 
contaminant....  I would wet sand these areas with a small block and 
progressively fine grits of sandpaper or polishing compound...  Must be 
careful to keep the run-out minimized along the long axis of the Z-arm 
contact patches, et al....  I don't see whether the paint is masked off 
on the rear panel at all.

c) Finally, these areas should be cleaned WELL with 91%+ alcohol or 
other suitably low-residue solvent to get them super clean.... think, 
mounting a heat-sink.  THEN, on reassembly, you want to use a QUALITY 
thermal compound..... Arctic Silver 5 would be almost ideal. IDEALLY, 
you would use an electrically conductive thermal compound, BUT because 
in this case, we are just "gap filling" to increase the contact patch 
integrity, it will still be electrically connected at the SAME points it 
would have without the compound.  AND the compound will keep the bright 
aluminum form oxidizing over time.

2) Connection to the heatsink:  The thin strip/patch where the heatsink 
mates with the Z-bracket is IMPORTANT.  This is the only path for heat 
to effectively travel from the rectifiers to the heatsink where it can 
be rejected via airflow from the fan to the outside world.  Using the 
techniques above similarly.

a) Lap the z-bracket face that contacts the heat-sink flat and smooth to 
1500 grit minimum.

b) Lap the EDGE of the heat-sink flat and smooth to 1500 grit minimum.  
This removes the "black" anodizing layer and provides a full and bright 
aluminum to aluminum thermal interface between the heatsink and the 
Z-bracket which SHOULD similarly increase the efficiency of heat removal 
from the rectifiers/Z-bracket. NOTE:  The heatsink is part of the 
PRE-assembled PA module.... If you do this, you MUST be careful to NOT 
contaminate or damage the PA modul board, traces, components et al with 
lapping wastes.  DO SO AT YOUR OWN RISK.

c) Properly clean both surfaces and apply the same thermal compound at 
this interface joint.  Use ENOUGH to provide full coverage, but not so 
much that you have excessive squeeze out.... clean up the excess around 
the edges with q-tips and paper towels... Do not use alcohol as you 
don't want it to wash out the thermal paste at the edges or undercut the 
interface connection.

3) Sound/vibration absorption/conversion:  We can't just go hog wild and 
stuff the box full of dampers and sound absorption material.  We don't 
want to reduce the interior free volume any more than absolutely 
necessary, and for SURE do not want to jeopardize the air flow over the 
heatsink and out via the fan opening.

I suggest using a layer of Sorbothane 1-side PSA sheeting of the 
appropriate durometer and of sufficient thickness to absorb any 
vibration/sound from the Z-bracket that might be made during a high 
demand thermal cycle...
The sheeting should ONLY be applied on the torroid side of the Z-bracket 
so as not to interrupt/block that last airflow channel between the 
Z-bracket and the first heatsink fin.  The more area that is covered the 
better.... with the trade-off being that you are consuming interior free 
volume.  The area on the free-floating rectifier tab is of particular 
importance....  the area between the rectifier board and the tab should 
be fully populated to the end, around the stand-offs and to within about 
and 1/8' (or closer if you are good) of the mounting patches of the 
rectifiers making SURE NOT to interfere with the rectifier packages 
making full contact with the bracket.  You probably can't do a 
continuous section on the other side of the rectifiers and past the 
torroid.... and probably should not try...  the torroid appears to be 
very close to the bracket at the tangent point of the torroid.... and 
perhaps a 1/4" channel left for aiflow would be better....  We don't 
want to cause any flow short circuits that would produce stagnant air 
that will heat up and work against our purposes.

You could also put a 1/8" thick sheet strip on the non-mating sides of 
the Z-bracket ears/arms....

AND one MIGHT consider a 1/8" sheet on the interior top panel (and 
bottom though the area is probably not sufficient for the trouble/small 
additional gains).

As an aside and not related really....  There is the POTENTIAL that the 
40-wire flat ribbon cable could be split and bound into a "rounded" 
cable, but I would NOT do that without first thoroughly reviewing the 
schematics and examined the signal arrangements....  there may be 
signals that need to be separate by intervening grounds on the flat 
cable....  We addressed this in the old days when I built 8 foot long 
PATA cables with 10x CAT-5 runs by giving EACH signal it's own twisted 
pair partner ground.  The only benefit to this would be to improve 
airflow in/around where the ribbon cable sits.

4) Other thoughts:  My original suggestion to put a Sorbothane damper 
across the heatsink fins in some fashion to break the resonance of the 
sound transmitted to them.... while perhaps offering some mitigation, is 
likely not needed if the above is done and may well reduce efficient 
operation of the heatsink and airflow.  IF I did this, I think I would 
likely try some small diameter disks just thick enough to offer an 
interference fit, and insert them between fins on the front, top corners 
of the fins.

==========================================================================

Without a unit to examine, this is my best, first SWAG at this...  I 
could be completely off base.  But I think this is a good start.  There 
are a number of additional things that could be done.

I have to say though.... I am overall very, VERY impressed with the 
package as illustrated.  Getting what they got into THAT package and 
making it do what it does is a testament to fine engineering!

If anyone has a KPA-500-K they want assembled, I would LOVE to have the 
opportunity to put one together.... my way.  <big smile>

73,

______________________
Clay Autery, KY5G
(318) 518-1389