On Sat, 29 Jan 2022 14:38:15 -0800 (PST), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

>l&#4294967295;rdag den 29. januar 2022 kl. 23.01.28 UTC+1 skrev jla...@highlandsniptechnology.com:
>> On Sun, 16 Jan 2022 14:11:22 -0800 (PST), Lasse Langwadt Christensen 
>> <lang...@fonz.dk> wrote: 
>> 
>> >s&#4294967295;ndag den 16. januar 2022 kl. 22.39.02 UTC+1 skrev jla...@highlandsniptechnology.com: 
>> >> On Sun, 16 Jan 2022 20:50:15 -0000 (UTC), Cydrome Leader 
>> >> <pres...@MUNGEpanix.com> wrote: 
>> >> 
>> >> >jla...@highlandsniptechnology.com wrote: 
>> >> >> On Fri, 14 Jan 2022 01:34:35 -0000 (UTC), Cydrome Leader 
>> >> >> <pres...@MUNGEpanix.com> wrote: 
>> >> >> 
>> >> >>>jla...@highlandsniptechnology.com wrote: 
>> >> >>>> Has anyone run bldc fans in series? I want to run three 12v fans from 
>> >> >>>> a 48v supply, with a resistor or something in series 
>> >> >>>> 
>> >> >>>> I guess I could be cautious and put a cap and a 12v TVS across each 
>> >> >>>> fan. Hmmm, 3 fans and 4 TVSs might work. 
>> >> >>>> 
>> >> >>>> Worst case, I can make a 48 to 12 switcher. 
>> >> >>>> 
>> >> >>>> Here's the idea: 
>> >> >>>> 
>> >> >>>> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
>> >> >>> 
>> >> >>>Don't do it. They will not share voltage and the losers will burn out on 
>> >> >>>the first start. Been there, done that. Find 48 volt telecom fans or 24 
>> >> >>>volts ones and use individual series resistors. Fans are mechanical so 
>> >> >>>don't increase your failure rate by using any sort of chain. 
>> >> >> 
>> >> >> The 35mm fans I want to use seem to only come in 5v and 12v. And the 
>> >> >> supply is 48. 
>> >> >> 
>> >> >> I suppose I can make a 48-to-12 switcher on this board. Still, a 
>> >> >> series string with shunt zeners ought to work. 
>> >> > 
>> >> >If you can go for a larger diameter fan, even mounted at an angle. They 
>> >> >run slower, make less noise and can last longer. This also opens up the 
>> >> >ability to get telecom fans. Fancy options allow for analog or PWM speed 
>> >> >control as well. 
>> >> > 
>> >> >If reliability is key, run two fans in series (air flow wise). There is no 
>> >> >major performance difference otherwise. You can even get counterrotating 
>> >> >double thick fan modules as used in servers. The major brand ones like 
>> >> >Nidec are actually extremely reliable at high temps and speed, even with 
>> >> >ball bearings. They really figured figured these things out. 
>> >> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
>> >> 
>> >> I considered all sorts of ways to use one big fan, horizontal or 
>> >> angled, and couldn't make it work. It would need some sort of ducting, 
>> >> and the next board is 1.6" away so there's no way to get the air into 
>> >> and out of a big fan. We do want to shoot the hot air out the rear of 
>> >> the box, not stir it around inside, another constraint. 
>> >> 
>> >> We're building a mockup for thermal testing. I have no analytical or 
>> >> simulation tools for a thing like this, and my instincts for air flow 
>> >> are all mediocre guesses. 
>> > 
>> >https://www.amazon.com/Cooler-Cooling-LGA1156-LGA1155-Utral-Thin/dp/B07PFH5B69
>> I ordered one just for fun. I'd have to stick my four giant mosfets on 
>> the bottom somehow.
>
>flip the whole pcb uupside down and place them around a square cutout in the middle of the pcb?
>
>mount them tap up and squeeze them between the pcb and heatsink with the X bracket

Spehro has been helping us with the thermals of our dummy load module.
He found this heat sink

https://www.dropbox.com/s/0kszc4nltr1q8d3/P944_HS_2.jpg?raw=1

which I think we can push 250 LFPM or so of air flow through without a
local fan. That might get us down to maybe 0.2 to 0.25 K/W... on a
PCB!

Well, the pcb is mounted on the heat sink more than vice versa.

If we put two giant fans on the front of the rack box, and make their
speed adaptive on heat sink temps, they will deafen only a minority of
our customers.

We can bolt that bottom heat sink fin to the bottom of the rackmount
enclosure. That will improve cooling some, and keep things from
ripping apart when some UPS guy throws it off the back of a truck.



-- 

I yam what I yam - Popeye

l&oslash;rdag den 29. januar 2022 kl. 23.01.28 UTC+1 skrev jla...@highlandsniptechnology.com:
> On Sun, 16 Jan 2022 14:11:22 -0800 (PST), Lasse Langwadt Christensen 
> <lang...@fonz.dk> wrote: 
> 
> >s&oslash;ndag den 16. januar 2022 kl. 22.39.02 UTC+1 skrev jla...@highlandsniptechnology.com: 
> >> On Sun, 16 Jan 2022 20:50:15 -0000 (UTC), Cydrome Leader 
> >> <pres...@MUNGEpanix.com> wrote: 
> >> 
> >> >jla...@highlandsniptechnology.com wrote: 
> >> >> On Fri, 14 Jan 2022 01:34:35 -0000 (UTC), Cydrome Leader 
> >> >> <pres...@MUNGEpanix.com> wrote: 
> >> >> 
> >> >>>jla...@highlandsniptechnology.com wrote: 
> >> >>>> Has anyone run bldc fans in series? I want to run three 12v fans from 
> >> >>>> a 48v supply, with a resistor or something in series 
> >> >>>> 
> >> >>>> I guess I could be cautious and put a cap and a 12v TVS across each 
> >> >>>> fan. Hmmm, 3 fans and 4 TVSs might work. 
> >> >>>> 
> >> >>>> Worst case, I can make a 48 to 12 switcher. 
> >> >>>> 
> >> >>>> Here's the idea: 
> >> >>>> 
> >> >>>> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
> >> >>> 
> >> >>>Don't do it. They will not share voltage and the losers will burn out on 
> >> >>>the first start. Been there, done that. Find 48 volt telecom fans or 24 
> >> >>>volts ones and use individual series resistors. Fans are mechanical so 
> >> >>>don't increase your failure rate by using any sort of chain. 
> >> >> 
> >> >> The 35mm fans I want to use seem to only come in 5v and 12v. And the 
> >> >> supply is 48. 
> >> >> 
> >> >> I suppose I can make a 48-to-12 switcher on this board. Still, a 
> >> >> series string with shunt zeners ought to work. 
> >> > 
> >> >If you can go for a larger diameter fan, even mounted at an angle. They 
> >> >run slower, make less noise and can last longer. This also opens up the 
> >> >ability to get telecom fans. Fancy options allow for analog or PWM speed 
> >> >control as well. 
> >> > 
> >> >If reliability is key, run two fans in series (air flow wise). There is no 
> >> >major performance difference otherwise. You can even get counterrotating 
> >> >double thick fan modules as used in servers. The major brand ones like 
> >> >Nidec are actually extremely reliable at high temps and speed, even with 
> >> >ball bearings. They really figured figured these things out. 
> >> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
> >> 
> >> I considered all sorts of ways to use one big fan, horizontal or 
> >> angled, and couldn't make it work. It would need some sort of ducting, 
> >> and the next board is 1.6" away so there's no way to get the air into 
> >> and out of a big fan. We do want to shoot the hot air out the rear of 
> >> the box, not stir it around inside, another constraint. 
> >> 
> >> We're building a mockup for thermal testing. I have no analytical or 
> >> simulation tools for a thing like this, and my instincts for air flow 
> >> are all mediocre guesses. 
> > 
> >https://www.amazon.com/Cooler-Cooling-LGA1156-LGA1155-Utral-Thin/dp/B07PFH5B69
> I ordered one just for fun. I'd have to stick my four giant mosfets on 
> the bottom somehow.

flip the whole pcb uupside down and place them around a square cutout in the middle of the pcb?

mount them tap up and squeeze them between the pcb and heatsink with the X bracket

On Sun, 16 Jan 2022 14:11:22 -0800 (PST), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

>s&#4294967295;ndag den 16. januar 2022 kl. 22.39.02 UTC+1 skrev jla...@highlandsniptechnology.com:
>> On Sun, 16 Jan 2022 20:50:15 -0000 (UTC), Cydrome Leader
>> <pres...@MUNGEpanix.com> wrote: 
>> 
>> >jla...@highlandsniptechnology.com wrote: 
>> >> On Fri, 14 Jan 2022 01:34:35 -0000 (UTC), Cydrome Leader 
>> >> <pres...@MUNGEpanix.com> wrote: 
>> >> 
>> >>>jla...@highlandsniptechnology.com wrote: 
>> >>>> Has anyone run bldc fans in series? I want to run three 12v fans from 
>> >>>> a 48v supply, with a resistor or something in series 
>> >>>> 
>> >>>> I guess I could be cautious and put a cap and a 12v TVS across each 
>> >>>> fan. Hmmm, 3 fans and 4 TVSs might work. 
>> >>>> 
>> >>>> Worst case, I can make a 48 to 12 switcher. 
>> >>>> 
>> >>>> Here's the idea: 
>> >>>> 
>> >>>> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
>> >>> 
>> >>>Don't do it. They will not share voltage and the losers will burn out on 
>> >>>the first start. Been there, done that. Find 48 volt telecom fans or 24 
>> >>>volts ones and use individual series resistors. Fans are mechanical so 
>> >>>don't increase your failure rate by using any sort of chain. 
>> >> 
>> >> The 35mm fans I want to use seem to only come in 5v and 12v. And the 
>> >> supply is 48. 
>> >> 
>> >> I suppose I can make a 48-to-12 switcher on this board. Still, a 
>> >> series string with shunt zeners ought to work. 
>> > 
>> >If you can go for a larger diameter fan, even mounted at an angle. They 
>> >run slower, make less noise and can last longer. This also opens up the 
>> >ability to get telecom fans. Fancy options allow for analog or PWM speed 
>> >control as well. 
>> > 
>> >If reliability is key, run two fans in series (air flow wise). There is no 
>> >major performance difference otherwise. You can even get counterrotating 
>> >double thick fan modules as used in servers. The major brand ones like 
>> >Nidec are actually extremely reliable at high temps and speed, even with 
>> >ball bearings. They really figured figured these things out.
>> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
>> 
>> I considered all sorts of ways to use one big fan, horizontal or 
>> angled, and couldn't make it work. It would need some sort of ducting, 
>> and the next board is 1.6" away so there's no way to get the air into 
>> and out of a big fan. We do want to shoot the hot air out the rear of 
>> the box, not stir it around inside, another constraint. 
>> 
>> We're building a mockup for thermal testing. I have no analytical or 
>> simulation tools for a thing like this, and my instincts for air flow 
>> are all mediocre guesses.
>
>https://www.amazon.com/Cooler-Cooling-LGA1156-LGA1155-Utral-Thin/dp/B07PFH5B69

I ordered one just for fun. I'd have to stick my four giant mosfets on
the bottom somehow.



-- 

I yam what I yam - Popeye

On 1/13/22 06:47, John Robertson wrote:
> On 2022/01/12 7:48 p.m., jlarkin@highlandsniptechnology.com wrote:
>> Has anyone run bldc fans in series? I want to run three 12v fans from
>> a 48v supply, with a resistor or something in series
> 
> Uneven airflow loading could burn out a fan...

So make sure that at least one fan is running to git rid of the smoke.


-- 
Reinhardt

jlarkin@highlandsniptechnology.com wrote:
> On Sun, 16 Jan 2022 20:42:19 -0000 (UTC), Cydrome Leader
> <presence@MUNGEpanix.com> wrote:
> 
>>whit3rd <whit3rd@gmail.com> wrote:
>>> On Thursday, January 13, 2022 at 5:34:42 PM UTC-8, Cydrome Leader wrote:
>>>> jla...@highlandsniptechnology.com wrote: 
>>>> > Has anyone run bldc fans in series? I want to run three 12v fans from 
>>>> > a 48v supply, with a resistor or something in series 
>>> 
>>>> Don't do it. They will not share voltage and the losers will burn out on 
>>>> the first start.
>>> 
>>> Yeah, it's the equivalent of a house with a floating neutral; well known for wild
>>> voltage fluctuations.   If the fans are identical, after the first fails, will the others
>>> fail the same way, and make a short?  Or will one or more fail open and leave
>>> everything unventilated pending combustion?   
>>> 
>>> I want to see that video.
>>> Meanwhile, enjoy this one
>>> <https://youtu.be/YSOpl4gG95w>
>>
>>That guy is pretty bold with the stuff he does indoors.
>>
>>For brushless fans, they just give up smoke and go open, and there's 
>>likely a fusible resistor in the larger ones. Never seen on fail short and 
>>consume more current than expected.
> 
> It wouldn't be hard to make a rotating magnetic field that would spin
> a metal disk or sphere to, say, a million RPM, enough to tear anything
> apart.

The supreme leader would like a word with you.

jlarkin@highlandsniptechnology.com wrote:
> On Tue, 18 Jan 2022 12:21:02 -0800 (PST), three_jeeps
> <jjhudak@gmail.com> wrote:
> 
>> On Wednesday, January 12, 2022 at 10:49:34 PM UTC-5, jla...@highlandsniptechnology.com wrote:
>>> Has anyone run bldc fans in series? I want to run three 12v fans from
>>> a 48v supply, with a resistor or something in series
>>>
>>> I guess I could be cautious and put a cap and a 12v TVS across each
>>> fan. Hmmm, 3 fans and 4 TVSs might work.
>>>
>>> Worst case, I can make a 48 to 12 switcher.
>>>
>>> Here's the idea:
>>>
>>> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1
>>>
>>>
>>>
>>>
>>> -- 
>>>
>>> I yam what I yam - Popeye
>> Are you pushing air over the exchanger fins or pulling air through them? the diagram suggest pushing.
>> I remember reading a paper some time ago that one needs to consider the spacing of the fins...it is a tradeoff: amount fin spacing/surface area, air flow, and channel cross sectional area.
>> Heat transfer optimally occurs with turbulent flow,  R =>3500.  Also shrouding of the fan and heat exchanger assembly can usually increase the heat transfer rate.
>> If you want to model this, COMSOL Multiphysics with the heat transfer module is a nice tool and would probably give you some useful results. You can get rather detailed and account for all 3 forms of heat transfer (conductive, convective, and radiative ).
>>
>> Then again, you could experiment with some RTDs, fan orientation, channels/shrouds, etc. Would five you some sight as to necessary fan flows, and not make the device sound like a jet engine when running.
>> J
>>
> 
> https://www.dropbox.com/s/o1ikkiolefx9csv/3_fans.jpg?raw=1
> 
> The idea is to stuff air into the heat sink. Some sort of channeling
> might help. If the heat sink has high fin density, there will be a lot
> of flow restriction - we *want* a lot of flow restriction - so the air
> will want to sneak around the heat sink and not go through the length
> of the fins.
> 
> Think we can dump 100 watts? Maybe we should dump the little fans...
> they would be a nuisance to install and wire up. With a really big
> heat sink, the overall box air flow might work as well.
> 
> We need to try it.
> 
> "One experiment is worth a thousand expert opinions."
> 
> I think Werner Von Braun said that.

Well, he was a very flexible fellow. ;)

https://www.youtube.com/watch?v=TjDEsGZLbio

Cheers

Phil Hobbs

On Tue, 18 Jan 2022 12:21:02 -0800 (PST), three_jeeps
<jjhudak@gmail.com> wrote:

>On Wednesday, January 12, 2022 at 10:49:34 PM UTC-5, jla...@highlandsniptechnology.com wrote:
>> Has anyone run bldc fans in series? I want to run three 12v fans from 
>> a 48v supply, with a resistor or something in series 
>> 
>> I guess I could be cautious and put a cap and a 12v TVS across each 
>> fan. Hmmm, 3 fans and 4 TVSs might work. 
>> 
>> Worst case, I can make a 48 to 12 switcher. 
>> 
>> Here's the idea: 
>> 
>> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
>> 
>> 
>> 
>> 
>> -- 
>> 
>> I yam what I yam - Popeye
>Are you pushing air over the exchanger fins or pulling air through them? the diagram suggest pushing.
>I remember reading a paper some time ago that one needs to consider the spacing of the fins...it is a tradeoff: amount fin spacing/surface area, air flow, and channel cross sectional area.     
>Heat transfer optimally occurs with turbulent flow,  R =>3500.  Also shrouding of the fan and heat exchanger assembly can usually increase the heat transfer rate.  
>If you want to model this, COMSOL Multiphysics with the heat transfer module is a nice tool and would probably give you some useful results. You can get rather detailed and account for all 3 forms of heat transfer (conductive, convective, and radiative ).  
>
>Then again, you could experiment with some RTDs, fan orientation, channels/shrouds, etc. Would five you some sight as to necessary fan flows, and not make the device sound like a jet engine when running.
>J
>

https://www.dropbox.com/s/o1ikkiolefx9csv/3_fans.jpg?raw=1

The idea is to stuff air into the heat sink. Some sort of channeling
might help. If the heat sink has high fin density, there will be a lot
of flow restriction - we *want* a lot of flow restriction - so the air
will want to sneak around the heat sink and not go through the length
of the fins. 

Think we can dump 100 watts? Maybe we should dump the little fans...
they would be a nuisance to install and wire up. With a really big
heat sink, the overall box air flow might work as well.

We need to try it.

"One experiment is worth a thousand expert opinions."

I think Werner Von Braun said that.

 



-- 

I yam what I yam - Popeye

On Tuesday, January 18, 2022 at 11:50:56 AM UTC-8, three_jeeps wrote:
> On Wednesday, January 12, 2022 at 10:49:34 PM UTC-5, jla...@highlandsniptechnology.com wrote: 
> > Has anyone run bldc fans in series? I want to run three 12v fans from 
> > a 48v supply, with a resistor or something in series 
> > 
> > I guess I could be cautious and put a cap and a 12v TVS across each 
> > fan. Hmmm, 3 fans and 4 TVSs might work. 
> > 
> > Worst case, I can make a 48 to 12 switcher. 

It might be simpler to use a 120V fan in a big box, with a thermostat fan control.   
Regulated voltages are unnecessary, dependence on the 48V source is unnecessary,
fitting  into the thin box is unnecessary, and a thermostat allows complete independence
of any logic signals.  When the fan needs lube, the guy in the field doesn't need to  open
a computer at a static-safe station and recalibrate a precision instrument.

On Wednesday, January 12, 2022 at 10:49:34 PM UTC-5, jla...@highlandsniptechnology.com wrote:
> Has anyone run bldc fans in series? I want to run three 12v fans from 
> a 48v supply, with a resistor or something in series 
> 
> I guess I could be cautious and put a cap and a 12v TVS across each 
> fan. Hmmm, 3 fans and 4 TVSs might work. 
> 
> Worst case, I can make a 48 to 12 switcher. 
> 
> Here's the idea: 
> 
> https://www.dropbox.com/s/5hx2hzrq4edd3ye/P944_Load_Module_1.jpg?raw=1 
> 
> 
> 
> 
> -- 
> 
> I yam what I yam - Popeye
Are you pushing air over the exchanger fins or pulling air through them? the diagram suggest pushing.
I remember reading a paper some time ago that one needs to consider the spacing of the fins...it is a tradeoff: amount fin spacing/surface area, air flow, and channel cross sectional area.     
Heat transfer optimally occurs with turbulent flow,  R =>3500.  Also shrouding of the fan and heat exchanger assembly can usually increase the heat transfer rate.  
If you want to model this, COMSOL Multiphysics with the heat transfer module is a nice tool and would probably give you some useful results. You can get rather detailed and account for all 3 forms of heat transfer (conductive, convective, and radiative ).  

Then again, you could experiment with some RTDs, fan orientation, channels/shrouds, etc. Would five you some sight as to necessary fan flows, and not make the device sound like a jet engine when running.
J