one point of consideration when purchasing a camera


Richard Crisp
 

The designers of the cooling systems in these cameras seem to fall into two categories: those that understand heat transfer and those that do not.

 

The ones that understand heat transfer orient the fans so they blow across the fins of the heatsink without leading to stagnation of the airflow in the heatsink

 

The ones that don’t arrange the fan so it blows straight down the heatsink, stagnating airflow in the process…

 

The latter case is sort of what arises if you put a sheet of metal directly behind the radiator of your car and then wonder why your engine overheats (assumes an internal combustion engine).

 

My operating assumption is that if the people designing a camera can’t get this right (a junior high school kid that’s reasonably mechanically skilled can figure that one out by inspection), then what else that might be a bit more subtle have they failed to get right?

 

I chatted up the designer of one of the two Chinese made cameras I see discussed here from time to time. He was at NEAF 2018.

 

I brought up the air / stagnation issue and used the radiator analogy…/ it went right over his head (blank stare)…maybe he has no experience with cars that overheat?

 

So if they miss the obvious simple stuff, what else are they missing?


Kent Kirkley
 

Richard

Good one!!!

Kent Kirkley



-----Original Message-----
From: Richard Crisp <rdcrisp@...>
To: main@ap-ug.groups.io
Sent: Mon, Jul 5, 2021 10:21 am
Subject: [ap-ug] one point of consideration when purchasing a camera

The designers of the cooling systems in these cameras seem to fall into two categories: those that understand heat transfer and those that do not.
 
The ones that understand heat transfer orient the fans so they blow across the fins of the heatsink without leading to stagnation of the airflow in the heatsink
 
The ones that don’t arrange the fan so it blows straight down the heatsink, stagnating airflow in the process…
 
The latter case is sort of what arises if you put a sheet of metal directly behind the radiator of your car and then wonder why your engine overheats (assumes an internal combustion engine).
 
My operating assumption is that if the people designing a camera can’t get this right (a junior high school kid that’s reasonably mechanically skilled can figure that one out by inspection), then what else that might be a bit more subtle have they failed to get right?
 
I chatted up the designer of one of the two Chinese made cameras I see discussed here from time to time. He was at NEAF 2018.
 
I brought up the air / stagnation issue and used the radiator analogy…/ it went right over his head (blank stare)…maybe he has no experience with cars that overheat?
 
So if they miss the obvious simple stuff, what else are they missing?


ROBERT WYNNE
 

What you faced was astonishment by a engineer/representative who believes the home design group could not possibly have introduced a design/product that was so obviously deficient at this late sage in its product life. I have run into this situation every now and then. When I used to make vendor calls I had the opportunity to see other vendor's designs on the manufacturing floor and concurrent production/design problems which for me were obvious. It always struck me that my more local vendors accepted this as friendly if  not helpful advice whereas others from more distant locations regarded my advice as an insult if not an affront even though much to their disadvantage. -Best, Robert

On 07/05/2021 9:43 AM Kent Kirkley via groups.io <kgkirkley@...> wrote:
 
 
Richard

Good one!!!
 
Kent Kirkley
 
 
 
-----Original Message-----
From: Richard Crisp <rdcrisp@...>
To: main@ap-ug.groups.io
Sent: Mon, Jul 5, 2021 10:21 am
Subject: [ap-ug] one point of consideration when purchasing a camera

The designers of the cooling systems in these cameras seem to fall into two categories: those that understand heat transfer and those that do not.
 
The ones that understand heat transfer orient the fans so they blow across the fins of the heatsink without leading to stagnation of the airflow in the heatsink
 
The ones that don’t arrange the fan so it blows straight down the heatsink, stagnating airflow in the process…
 
The latter case is sort of what arises if you put a sheet of metal directly behind the radiator of your car and then wonder why your engine overheats (assumes an internal combustion engine).
 
My operating assumption is that if the people designing a camera can’t get this right (a junior high school kid that’s reasonably mechanically skilled can figure that one out by inspection), then what else that might be a bit more subtle have they failed to get right?
 
I chatted up the designer of one of the two Chinese made cameras I see discussed here from time to time. He was at NEAF 2018.
 
I brought up the air / stagnation issue and used the radiator analogy…/ it went right over his head (blank stare)…maybe he has no experience with cars that overheat?
 
So if they miss the obvious simple stuff, what else are they missing?


Richard Crisp
 

AKA NIH (not invented here). I was in the technology development/licensing business for years. The business model was “we know how to do it better than you do and we have it all patented so even if you try to do what we have done already in the future, you will still need a license, so pay up”. I got to see about as many forms of NIH as existed in the world and on three continents!

 

That was not the point of my posting, however. The point was there’s a clearly preferred way to do the thermal job in a camera if you want cooling to work well, and I see the majority of designs doing it wrong: They are leaving 5-7C on the table.

 

If they make such a bonehead mistake on the simple stuff that can be seen by inspection if you happen to know even the most basic things about heat transfer, how well do you think they designed the signal processing, interrupt handling etc and how clean do you think the source code is for the drivers and the overall mechanical integrity?

 

Anyway: I have no dog in this hunt but I do have a huge amount of experience in both the hobby and in the engineering of the hardware so I am just passing on some observations you may or may not find helpful.

 

 

From: main@ap-ug.groups.io <main@ap-ug.groups.io> On Behalf Of ROBERT WYNNE
Sent: Monday, July 5, 2021 12:15 PM
To: main@ap-ug.groups.io; Kent Kirkley via groups.io <kgkirkley@...>
Subject: Re: [ap-ug] one point of consideration when purchasing a camera

 

What you faced was astonishment by a engineer/representative who believes the home design group could not possibly have introduced a design/product that was so obviously deficient at this late sage in its product life. I have run into this situation every now and then. When I used to make vendor calls I had the opportunity to see other vendor's designs on the manufacturing floor and concurrent production/design problems which for me were obvious. It always struck me that my more local vendors accepted this as friendly if  not helpful advice whereas others from more distant locations regarded my advice as an insult if not an affront even though much to their disadvantage. -Best, Robert

On 07/05/2021 9:43 AM Kent Kirkley via groups.io <kgkirkley@...> wrote:

 

 

Richard

Good one!!!

 

Kent Kirkley

 

 

 

-----Original Message-----
From: Richard Crisp <rdcrisp@...>
To: main@ap-ug.groups.io
Sent: Mon, Jul 5, 2021 10:21 am
Subject: [ap-ug] one point of consideration when purchasing a camera

The designers of the cooling systems in these cameras seem to fall into two categories: those that understand heat transfer and those that do not.

 

The ones that understand heat transfer orient the fans so they blow across the fins of the heatsink without leading to stagnation of the airflow in the heatsink

 

The ones that don’t arrange the fan so it blows straight down the heatsink, stagnating airflow in the process…

 

The latter case is sort of what arises if you put a sheet of metal directly behind the radiator of your car and then wonder why your engine overheats (assumes an internal combustion engine).

 

My operating assumption is that if the people designing a camera can’t get this right (a junior high school kid that’s reasonably mechanically skilled can figure that one out by inspection), then what else that might be a bit more subtle have they failed to get right?

 

I chatted up the designer of one of the two Chinese made cameras I see discussed here from time to time. He was at NEAF 2018.

 

I brought up the air / stagnation issue and used the radiator analogy…/ it went right over his head (blank stare)…maybe he has no experience with cars that overheat?

 

So if they miss the obvious simple stuff, what else are they missing?


ROBERT WYNNE
 

I find all you say a helpful insight. As a holder of some basic IP I have found the rats are always nibbling away at the corners. One patent actually had a preferred embodiment removed from the wrapper without my knowledge. It took forever to correct and impose royalties.
 
But I agree the sophistication of a new product belies the intrinsic knowledge of the inventors and its engineering team. If you are going to cool something as advanced as a high end camera its design should represent state of the art design across the board. Otherwise it's a bit like Ferrari using a low compression engine in one of their F1's. -Best, Robert

On 07/05/2021 1:16 PM Richard Crisp <rdcrisp@...> wrote:
 
 

AKA NIH (not invented here). I was in the technology development/licensing business for years. The business model was “we know how to do it better than you do and we have it all patented so even if you try to do what we have done already in the future, you will still need a license, so pay up”. I got to see about as many forms of NIH as existed in the world and on three continents!

 

That was not the point of my posting, however. The point was there’s a clearly preferred way to do the thermal job in a camera if you want cooling to work well, and I see the majority of designs doing it wrong: They are leaving 5-7C on the table.

 

If they make such a bonehead mistake on the simple stuff that can be seen by inspection if you happen to know even the most basic things about heat transfer, how well do you think they designed the signal processing, interrupt handling etc and how clean do you think the source code is for the drivers and the overall mechanical integrity?

 

Anyway: I have no dog in this hunt but I do have a huge amount of experience in both the hobby and in the engineering of the hardware so I am just passing on some observations you may or may not find helpful.

 

 

From: main@ap-ug.groups.io <main@ap-ug.groups.io> On Behalf Of ROBERT WYNNE
Sent: Monday, July 5, 2021 12:15 PM
To: main@ap-ug.groups.io; Kent Kirkley via groups.io <kgkirkley@...>
Subject: Re: [ap-ug] one point of consideration when purchasing a camera

 

What you faced was astonishment by a engineer/representative who believes the home design group could not possibly have introduced a design/product that was so obviously deficient at this late sage in its product life. I have run into this situation every now and then. When I used to make vendor calls I had the opportunity to see other vendor's designs on the manufacturing floor and concurrent production/design problems which for me were obvious. It always struck me that my more local vendors accepted this as friendly if  not helpful advice whereas others from more distant locations regarded my advice as an insult if not an affront even though much to their disadvantage. -Best, Robert

On 07/05/2021 9:43 AM Kent Kirkley via groups.io <kgkirkley@...> wrote:

 

 

Richard

Good one!!!

 

Kent Kirkley

 

 

 

-----Original Message-----
From: Richard Crisp <rdcrisp@...>
To: main@ap-ug.groups.io
Sent: Mon, Jul 5, 2021 10:21 am
Subject: [ap-ug] one point of consideration when purchasing a camera

The designers of the cooling systems in these cameras seem to fall into two categories: those that understand heat transfer and those that do not.

 

The ones that understand heat transfer orient the fans so they blow across the fins of the heatsink without leading to stagnation of the airflow in the heatsink

 

The ones that don’t arrange the fan so it blows straight down the heatsink, stagnating airflow in the process…

 

The latter case is sort of what arises if you put a sheet of metal directly behind the radiator of your car and then wonder why your engine overheats (assumes an internal combustion engine).

 

My operating assumption is that if the people designing a camera can’t get this right (a junior high school kid that’s reasonably mechanically skilled can figure that one out by inspection), then what else that might be a bit more subtle have they failed to get right?

 

I chatted up the designer of one of the two Chinese made cameras I see discussed here from time to time. He was at NEAF 2018.

 

I brought up the air / stagnation issue and used the radiator analogy…/ it went right over his head (blank stare)…maybe he has no experience with cars that overheat?

 

So if they miss the obvious simple stuff, what else are they missing?

 


Ray Gralak
 

The ones that don't arrange the fan so it blows straight down the heatsink, stagnating airflow in the process.
Hey wait... I've seen many CPU heat sinks that do the same thing. I guess maybe there is a reason for that! Care to guess? :-)

-Ray


Richard Crisp
 

Intel’s poor thermal engineering team started the mess in about 1989 with the 486.

When they did that with the 486 everyone else (cpu chip makers) were buying expensive pin grid arrays with integrated tungsten copper heat spreaders and affixing expensive heat sinks onto the assembly

At Moto and MIPS where
I was developing CPUs at the time we were grumbling we had to spend $20-$30 on a such a fancy ceramic package (customers like DEC & IBM NON-ESG demanded a packaging vs fan solution), but Intel could get away with requiring an integrated fan heatsink purely because of their “take it leave it” market power back in the day. Amd & Cyrix still weren’t a credible threat to Wintel.

Of course they only had a 10watt to 20watt problem in those days: we had an bipolar ECL MIPS processor (remember BiT in oregon?) with a 60watt problem

But you are right they started the trend in the cpu world. Component level solution to a system level problem.

FLI’s maxcams (~2001) used a cpu cooler too. Terrible product with abominable performance.





“Corrected” by my iPhone

On Jul 5, 2021, at 4:09 PM, Ray Gralak <iogroups@siriusimaging.com> wrote:



The ones that don't arrange the fan so it blows straight down the heatsink, stagnating airflow in the process.
Hey wait... I've seen many CPU heat sinks that do the same thing. I guess maybe there is a reason for that! Care to guess? :-)

-Ray







ROBERT WYNNE
 

Still think CPU's should be cooled by operating in a circulating liquid nitrogen bath if you can afford it. Just like Cray did it in the old days. I find it somewhat of a marketing ploy that end users could up the speed of their system by simply lowering the operating temperature by 30 degrees in place of buying the latest and greatest smallest etch CPU only to run it at room temperature. A while back on this board A-P reported a fellow in Australia who obtained a near 1,000 fold increase in processing power by simply immersing the CPU in liquid N. But then I had the luxury of a 25 liter liquid nitrogen Dewar. Now we have bolt on coolers that represent a super cooled CPU environment by only 20-30 degrees for a 20-50% improvement in performance for several hundred dollars. As far as CPU's and other components operating at below spec. temps, that subject has already been beaten to death and disproven. Components guaranteed to operate below specified temperature are simply chips taken off the shelf from one batch and actually run and tested at below spec. temps and then sold as specially fabricated to operate at lower temperatures. There's nothing special about their fabrication. Funny the current trend is to market boot up software that overclocks or ramps up performance by increasing CPU voltage so that one can "supercharge" a system. Sure if you want to prematurely burn up your computer. Todays best performing quantum supercomputers run at -460 K. in liquid Helium. -Best, Robert

On 07/05/2021 4:27 PM Richard Crisp <rdcrisp@sbcglobal.net> wrote:


Intel’s poor thermal engineering team started the mess in about 1989 with the 486.

When they did that with the 486 everyone else (cpu chip makers) were buying expensive pin grid arrays with integrated tungsten copper heat spreaders and affixing expensive heat sinks onto the assembly

At Moto and MIPS where
I was developing CPUs at the time we were grumbling we had to spend $20-$30 on a such a fancy ceramic package (customers like DEC & IBM NON-ESG demanded a packaging vs fan solution), but Intel could get away with requiring an integrated fan heatsink purely because of their “take it leave it” market power back in the day. Amd & Cyrix still weren’t a credible threat to Wintel.

Of course they only had a 10watt to 20watt problem in those days: we had an bipolar ECL MIPS processor (remember BiT in oregon?) with a 60watt problem

But you are right they started the trend in the cpu world. Component level solution to a system level problem.

FLI’s maxcams (~2001) used a cpu cooler too. Terrible product with abominable performance.





“Corrected” by my iPhone

On Jul 5, 2021, at 4:09 PM, Ray Gralak <iogroups@siriusimaging.com> wrote:



The ones that don't arrange the fan so it blows straight down the heatsink, stagnating airflow in the process.
Hey wait... I've seen many CPU heat sinks that do the same thing. I guess maybe there is a reason for that! Care to guess? :-)

-Ray









Ray Gralak
 

So, you think it was just Intel? :-)

Check out this comparison of CPU coolers. Most are pointing down towards the CPU:

https://www.anandtech.com/show/10500/stock-cooler-roundup-intel-amd-vs-evo-212

-Ray

-----Original Message-----
From: main@ap-ug.groups.io [mailto:main@ap-ug.groups.io] On Behalf Of Richard Crisp
Sent: Monday, July 5, 2021 4:27 PM
To: main@ap-ug.groups.io
Subject: Re: [ap-ug] one point of consideration when purchasing a camera

Intel’s poor thermal engineering team started the mess in about 1989 with the 486.

When they did that with the 486 everyone else (cpu chip makers) were buying expensive pin grid arrays with
integrated tungsten copper heat spreaders and affixing expensive heat sinks onto the assembly

At Moto and MIPS where
I was developing CPUs at the time we were grumbling we had to spend $20-$30 on a such a fancy ceramic package
(customers like DEC & IBM NON-ESG demanded a packaging vs fan solution), but Intel could get away with
requiring an integrated fan heatsink purely because of their “take it leave it” market power back in the day. Amd &
Cyrix still weren’t a credible threat to Wintel.

Of course they only had a 10watt to 20watt problem in those days: we had an bipolar ECL MIPS processor
(remember BiT in oregon?) with a 60watt problem

But you are right they started the trend in the cpu world. Component level solution to a system level problem.

FLI’s maxcams (~2001) used a cpu cooler too. Terrible product with abominable performance.





“Corrected” by my iPhone

On Jul 5, 2021, at 4:09 PM, Ray Gralak <iogroups@siriusimaging.com> wrote:



The ones that don't arrange the fan so it blows straight down the heatsink, stagnating airflow in the process.
Hey wait... I've seen many CPU heat sinks that do the same thing. I guess maybe there is a reason for that! Care
to guess? :-)

-Ray









Richard Crisp
 

Let’s not get into the weeds

The fan with heatsink on the 486 was adopted by Intel 486 platforms in 1989. Intel sold some CPUs with heatsink attached but most were cheap third party.

I’m talking about Astro cameras and poorly engineered cooling systems that are broken by inspection and that are developed 32 years later when a hell of
A lot more is known.

Modern computer cooling systems typically use heat pipes and the heat exchangers have radiators with fins and air that blows through the fins without stagnation

Those are used in blade servers and notebooks when the TDP is too high to use conductive/convective cooling (~5w)

Nobody in the computer business uses fans/heatsinks with stagnated airflow today

The astro cam makers that do use stagnated airflow are shooting themselves in the foot by leaving 5-7C of cooling on the table in my professional judgement a d in my engineering experience. It costs nothing incremental to do it correctly.

When someone makes a boneheaded move like that after 25 - 30 years of collective experience saying “it’s a boneheaded move” they aren’t doing as good of an engineering job as can be done: not practicing best known methods as we say in the business.

That would make me suspicious they fell short in other ways

That’s my suspicion,
Your mileage may vary

I’m finished with this thread
“Corrected” by my iPhone

On Jul 5, 2021, at 6:23 PM, Ray Gralak <iogroups@siriusimaging.com> wrote:

So, you think it was just Intel? :-)

Check out this comparison of CPU coolers. Most are pointing down towards the CPU:

https://www.anandtech.com/show/10500/stock-cooler-roundup-intel-amd-vs-evo-212

-Ray

-----Original Message-----
From: main@ap-ug.groups.io [mailto:main@ap-ug.groups.io] On Behalf Of Richard Crisp
Sent: Monday, July 5, 2021 4:27 PM
To: main@ap-ug.groups.io
Subject: Re: [ap-ug] one point of consideration when purchasing a camera

Intel’s poor thermal engineering team started the mess in about 1989 with the 486.

When they did that with the 486 everyone else (cpu chip makers) were buying expensive pin grid arrays with
integrated tungsten copper heat spreaders and affixing expensive heat sinks onto the assembly

At Moto and MIPS where
I was developing CPUs at the time we were grumbling we had to spend $20-$30 on a such a fancy ceramic package
(customers like DEC & IBM NON-ESG demanded a packaging vs fan solution), but Intel could get away with
requiring an integrated fan heatsink purely because of their “take it leave it” market power back in the day. Amd &
Cyrix still weren’t a credible threat to Wintel.

Of course they only had a 10watt to 20watt problem in those days: we had an bipolar ECL MIPS processor
(remember BiT in oregon?) with a 60watt problem

But you are right they started the trend in the cpu world. Component level solution to a system level problem.

FLI’s maxcams (~2001) used a cpu cooler too. Terrible product with abominable performance.





“Corrected” by my iPhone

On Jul 5, 2021, at 4:09 PM, Ray Gralak <iogroups@siriusimaging.com> wrote:


The ones that don't arrange the fan so it blows straight down the heatsink, stagnating airflow in the process.
Hey wait... I've seen many CPU heat sinks that do the same thing. I guess maybe there is a reason for that! Care
to guess? :-)

-Ray














Ray Gralak
 

Did you look at the results of the tests at the link I provided? Sure, the one CPU cooler with the fan crossflowing air across the fins was near the top of the chart, but several of the coolers with fan flow down onto the fins/cpu were very close.

So, back to the camera, my point is that concluding performance simply by physically inspecting the position of the camera's fan doesn't necessarily translate into a much poorer camera performance. Excessive cooling is not really required for CMOS, as it is for some CCDs. The design decision for whatever camera you are talking about may have been made to balance size, weight, andcooling performance.

BTW, I am a big "fan" (hehe) of large finned CPU coolers with multiple fans. However, I can't see three pounds of aluminum heat pipes and fins plus fans would be a great solution on any astro-camera sitting on the end of one of my scopes! :-)

-Ray


Ross Elkins
 

I was helping with the LAN/WAN stuff for the Head of computation at Lawrence Livermore. This guy had a signed pic of him and almost every astronaut in his office! Everyday I would walk by this building with clear glass walls on the bottom floors that housed the largest array of Cray’s on earth at the time. Setup to impress visiting vip’s, the external cooling pipes were all visible traveling through the floors and on to eventually their heat exchangers. It kinda looked like a bunch of Robby the Robots lining the space like on a giant checkerboard.
It made you feel good to be an American which I guess is todays theme 🇺🇸!!!
Meanwhile, more to the north, up at Mare Island Naval electronics, I worked on networking the individual Crays that were installed in each ICBM submarine that came thru there. Working for 3Com got me into many neat places. FYI, 3Com was founded by Dr Metcalfe, one of the several inventors of ethernet for Xerox. Now for a story straight out of left field, there was this super laser at Livermore that had the power of a sizable towns electrical grid in its buildings basement to drive it. I did the network wiring and interfaces to transmit experimental data for the laser array. The chief of the show kept offering to burn up any handy large bug for us in the test area to demo our install! Haha, we said “right on“ and one day poof! A blast of bright and not even a tinge of ash left! I think the “demo” cost $20k’s worth of electrons!


weems@...
 

Just a point of correction. The Cray 2 logic was immersed in fluorinert. Not liquid nitrogen. Fluorinert needs to be aerated, and each Cray 2 customer got to choose the design for the decorative fountain that sprayed it, and also served as the storage reservoir during service. The prior Crays all used piped Freon cooling. 

Also, there are no quantum supercomputers. We’re still a long way from having enough qubits to do serious computing. The low temperatures aren’t for cooling heat dissipation. They are necessary to keep the particles entangled, and the effect actually slows the computation. If there was a way to build a practical quantum computer to operate at room temperature, it would be a game changer. 

Chip


ROBERT WYNNE
 

Isn't keeping particles from becoming entangled another way to describe thermal transfer? -Best, Robert

On 07/05/2021 10:16 PM weems@... wrote:
 
 

Just a point of correction. The Cray 2 logic was immersed in fluorinert. Not liquid nitrogen. Fluorinert needs to be aerated, and each Cray 2 customer got to choose the design for the decorative fountain that sprayed it, and also served as the storage reservoir during service. The prior Crays all used piped Freon cooling. 

Also, there are no quantum supercomputers. We’re still a long way from having enough qubits to do serious computing. The low temperatures aren’t for cooling heat dissipation. They are necessary to keep the particles entangled, and the effect actually slows the computation. If there was a way to build a practical quantum computer to operate at room temperature, it would be a game changer. 

Chip


ROBERT WYNNE
 

I sure was flummoxed when you stated the original Crays were not cooled by liquid nitrogen which runs counter to my recollection. Just before turning in tonight I found this: The ETA10-F and ETA10-G (7 ns clock cycle) were the highest-performing members of the ETA10 line, and used liquid nitrogen cooling to achieve rapid cycle times.  ETA10 - Wikipedia  -Best, Robert

On 07/05/2021 10:16 PM weems@... wrote:
 
 

Just a point of correction. The Cray 2 logic was immersed in fluorinert. Not liquid nitrogen. Fluorinert needs to be aerated, and each Cray 2 customer got to choose the design for the decorative fountain that sprayed it, and also served as the storage reservoir during service. The prior Crays all used piped Freon cooling. 

Also, there are no quantum supercomputers. We’re still a long way from having enough qubits to do serious computing. The low temperatures aren’t for cooling heat dissipation. They are necessary to keep the particles entangled, and the effect actually slows the computation. If there was a way to build a practical quantum computer to operate at room temperature, it would be a game changer. 

Chip


Christopher Erickson
 

Fans that blow across heat sinks instead of straight down on them create uneven heat stripping comparing the side closest to the fan to the side farthest.

There are no perfect solutions. Liquid circulation cooling systems are significantly better than air cooling.

Liquid nitrogen based evaporative cooing systems have the best performance and capabilities but are rather overkill for amateur-grade equipment.

Personally, I feel that a fan pointed straight down into a heat sink gives the most balanced cooling, although it does have a dead zone in the middle. The thermal conductivity of the heat sink itself allows sufficient flow of center heat towards the edges of the heat sink, where the heat is then stripped away.

I worry a lot more about dust and lint collection in heat sink fins, making sure that all cooling fans contain ball bearings instead of sleeve bushings, and the effects of high altitude on various air cooling designs.

In my experience, if a cooling fan does not mention ball bearings on its sticker, it doesn't have them and should be replaced.

I also make sure that mechanical hard drives that depend on air bearings aren't used at altitudes over about 8,000'. Which includes just about all modern hard drives less than 25 years old. I am a big advocate of SSD drives.

Everyone should add all fans and heat sinks to their observatory annual inspection lists.

I hope this helps.

-Christopher Erickson
Observatory engineer
Waikoloa, HI 96738
www.summitkinetics.com
   


On Mon, Jul 5, 2021, 4:09 PM Ray Gralak <iogroups@...> wrote:
> The ones that don't arrange the fan so it blows straight down the heatsink, stagnating airflow in the process.

Hey wait... I've seen many CPU heat sinks that do the same thing. I guess maybe there is a reason for that! Care to guess? :-)

-Ray








weems@...
 

This is my research area. I knew many of the people involved with various high performance computing system projects over the last 40 years, and I could go on at painful length about the history. ETA was the last gasp of the CDC Star 100 effort, and it was no relation to Cray, other than both having teams who drew from their experience with the CDC 6600. Although they delivered some machines, they were generally seen as failures by those who tried to use them. Seymour Cray understood that a usable supercomputer has to be practical to build and maintain, provide a carefully balanced set of compute resources, and that it needs an effective software development stack that includes performance analysis and tuning tools. ETA was focused on maximizing clock speed, and neglected most of the rest.

The goal of a quantum computer is to keep particles in an entangled state until you can apply the observation condition that causes them to collapse into a classical state that gives the answer. The more you entangle, however, the more sensitive they become to outside influences that can trigger their collapse into a pattern other than the answer. Cooling the entangled qubits reduces their sensitivity to that noise. But it also slows their response to the intentionally applied conditioning. 

Most modern high-performance systems that use liquid cooling just use chilled water. There is a hobbyist community that plays with extreme overclocking of standard chips, where they use liquid nitrogen drips onto the heat sink. 

When I did a couple of observing runs on the 0.9m WIYN at Kitt Peak, we used a camera in a liquid nitrogen dewar, that we had to refill twice a day. It was important to know where to stand while doing that because the "full" indicator was when the liquid nitrogen would shoot out of the dewar in a stream for about 6 to 10 feet.

Chip


Christopher Erickson
 

I have used and maintained both liquid-cooled (glycerin or propylene glycol) and liquid nitrogen instrument cooling systems. Both can be a PITA. Witnessed a liquid-cooled instrument cooling system failure that drenched an 8 meter mirror and several racks of electronics in a huge, sticky mess. And liquid nitrogen systems deserve a lot of respect. You definitely want to be careful and alert when swapping bottles and filling chambers with that stuff. Especially at 14,000', where the brain doesn't work really well in the first place.

-Christopher Erickson
Observatory engineer
Waikoloa, HI 96738
www.summitkinetics.com
   

On Tue, Jul 6, 2021, 8:44 AM <weems@...> wrote:
This is my research area. I knew many of the people involved with various high performance computing system projects over the last 40 years, and I could go on at painful length about the history. ETA was the last gasp of the CDC Star 100 effort, and it was no relation to Cray, other than both having teams who drew from their experience with the CDC 6600. Although they delivered some machines, they were generally seen as failures by those who tried to use them. Seymour Cray understood that a usable supercomputer has to be practical to build and maintain, provide a carefully balanced set of compute resources, and that it needs an effective software development stack that includes performance analysis and tuning tools. ETA was focused on maximizing clock speed, and neglected most of the rest.

The goal of a quantum computer is to keep particles in an entangled state until you can apply the observation condition that causes them to collapse into a classical state that gives the answer. The more you entangle, however, the more sensitive they become to outside influences that can trigger their collapse into a pattern other than the answer. Cooling the entangled qubits reduces their sensitivity to that noise. But it also slows their response to the intentionally applied conditioning. 

Most modern high-performance systems that use liquid cooling just use chilled water. There is a hobbyist community that plays with extreme overclocking of standard chips, where they use liquid nitrogen drips onto the heat sink. 

When I did a couple of observing runs on the 0.9m WIYN at Kitt Peak, we used a camera in a liquid nitrogen dewar, that we had to refill twice a day. It was important to know where to stand while doing that because the "full" indicator was when the liquid nitrogen would shoot out of the dewar in a stream for about 6 to 10 feet.

Chip


ROBERT WYNNE
 

ETA was acquired by Cray and for a period of a couple years and were essentially the same company. There is a difference in coolant between the early Cray's and ETA's but with the incorporation of the two company's at the early stage of both at the time of consolidation/ incorporation, attribution of a coolant system is at best a gray area. Did the original Cray computer use liquid nitrogen as a coolant - no. Did the Cray corporation ever use liquid nitrogen as a coolant for its super computers - yes. But here I speak of the corporate history of the two companies early supercomputers and not the earliest individual designs of each team.
 
The SEM in my lab had a FET for an added XRS unit. The FET had to be kept totally immersed in liquid nitrogen otherwise the FET would become unstable and possibly permanently unusable. 
 
The Dewar's I'm familiar all have 310 stainless steel valves and regulators to restrict flow and would not be capable of sending a stream 6 -10' across the lab. I'm not certain what type of Dewar you refer? We had to decant liquid nitrogen from the main Dewar in order to safely fill the XRS tank. -Best, Robert

On 07/06/2021 8:44 AM weems@... wrote:
 
 
This is my research area. I knew many of the people involved with various high performance computing system projects over the last 40 years, and I could go on at painful length about the history. ETA was the last gasp of the CDC Star 100 effort, and it was no relation to Cray, other than both having teams who drew from their experience with the CDC 6600. Although they delivered some machines, they were generally seen as failures by those who tried to use them. Seymour Cray understood that a usable supercomputer has to be practical to build and maintain, provide a carefully balanced set of compute resources, and that it needs an effective software development stack that includes performance analysis and tuning tools. ETA was focused on maximizing clock speed, and neglected most of the rest.

The goal of a quantum computer is to keep particles in an entangled state until you can apply the observation condition that causes them to collapse into a classical state that gives the answer. The more you entangle, however, the more sensitive they become to outside influences that can trigger their collapse into a pattern other than the answer. Cooling the entangled qubits reduces their sensitivity to that noise. But it also slows their response to the intentionally applied conditioning. 

Most modern high-performance systems that use liquid cooling just use chilled water. There is a hobbyist community that plays with extreme overclocking of standard chips, where they use liquid nitrogen drips onto the heat sink. 

When I did a couple of observing runs on the 0.9m WIYN at Kitt Peak, we used a camera in a liquid nitrogen dewar, that we had to refill twice a day. It was important to know where to stand while doing that because the "full" indicator was when the liquid nitrogen would shoot out of the dewar in a stream for about 6 to 10 feet.

Chip


weems@...
 

No, ETA was a spin-off from Control Data, and when they failed, their assets were merged back into CDC. They were never owned by Cray. Their architecture was a pipelined vector design, which wasn’t at all compatible with Cray’s register design. 

I don’t know what kind of dewar the camera on the 0.9m used. There was a nitrogen supply in the basement that came up through a heavy hose.  You would screw it onto the fill port, then go turn a valve, and stand back until the LN2 shot out. It looked to me like custom hardware cooked up in a machine shop. The sensor was a very early backside illuminated design with a 2K square array, and my understanding was that it was a lab-built one-off unit that had been cutting edge some years earlier, and was by then having a second life for educational programs and some background projects. 

An interesting anecdote from my second run there was that when we were getting trained, the tech invited my students to climb up a ladder and look inside. The first one up turned back to the tech and asked, “Why is there a garbage bag on the mirror?”  Nobody knew how long it had been down there. The fix was to tape a couple of broom handles together with a wad of duct tape on the end and poke at it until it stuck, then pull it out. This was a great introduction to professional level astronomy research. 

Chip