IBM heat sink for 600 Watt Processor...

[amdx]

I found this IBM heat sink for a 23 die processor interesting.

> https://youtu.be/xQ3oJlt4GrI?t=389

It\'s Dave, so for those of you that don\'t like to listen to Dave, I
start at the important part.

                                   Mikek


--
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

 

[Phil Hobbs]

On 2020-10-08 08:50, amdx wrote:

I found this IBM heat sink for a 23 die processor interesting.

https://youtu.be/xQ3oJlt4GrI?t=389

It\'s Dave, so for those of you that don\'t like to listen to Dave, I
start at the important part.

                                   Mikek

Yeah, that\'s a TCM (Thermal Conduction Module). Back in the late \'70s
and \'80s a lot of IBM\'s technological lead came from packaging, which
allowed them to pack a lot of ECL into a small space, which lets more of
the processor be synchronous. ECL\'s main advantage was its huge
transconductance, which lets it drive a lot of capacitance fast.

I remember going to a talk by a mainframe guy who plotted cycle time vs.
power dissipation for ECL (IBM ATX-4 process iirc). It rose steeply at
very low power due to low transconductance, and more shallowly at high
power due to having to space the chips out further for cooling purposes.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Robert Baer]

amdx wrote:

I found this IBM heat sink for a 23 die processor interesting.

https://youtu.be/xQ3oJlt4GrI?t=389

It\'s Dave, so for those of you that don\'t like to listen to Dave, I
start at the important part.

                                   Mikek


Eye aint kno enginea, but the design looks poor.

Having a gap between the chip and the copper slug heatsink is not too
swift.
Worse, the oil surrounding them is a poor conductor.

Better would be to use a diamond as the \"mounting pad\" for the chip;
much higher thermal conductivity with advantage of electrical isolation
which might be useful.

Use the metallic \"solder\" trick to interconnect to NOT aluminum
heatsink; copper at worst, silver is better.
Top-of-the-line result deserves not cheap design.
Think of Rolls Royce.

Thanks

 

[John S]

On 10/8/2020 10:44 PM, Robert Baer wrote:


>   Eye aint kno enginea

On several counts. Obviously.

 

[Cydrome Leader]

Robert Baer <robertbaer@localnet.com> wrote:

amdx wrote:
I found this IBM heat sink for a 23 die processor interesting.

https://youtu.be/xQ3oJlt4GrI?t=389

It\'s Dave, so for those of you that don\'t like to listen to Dave, I
start at the important part.

???????????????????????????????????????????????????????????????????? Mikek


Eye aint kno enginea, but the design looks poor.

Having a gap between the chip and the copper slug heatsink is not too
swift.
Worse, the oil surrounding them is a poor conductor.

Better would be to use a diamond as the \"mounting pad\" for the chip;
much higher thermal conductivity with advantage of electrical isolation
which might be useful.

Use the metallic \"solder\" trick to interconnect to NOT aluminum
heatsink; copper at worst, silver is better.
Top-of-the-line result deserves not cheap design.
Think of Rolls Royce.

nice try.

IBM is the only one in the game for making modules this complex, at
production levels for decades. They know what they\'re doing.