VOGONS


Absolute best, Most reliable fans

Topic actions

First post, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

I am in need of some military grade fans for a project, this project will experience the arizona heat and very hot equipment (as much as 100c) these need to be able to run 12 hours a day every day for decades without failure or any maintenance whatsoever they will be buried deep in the equipment and can not fail because their failure won’t be detected and will result in equipment failure, weeks of down time and many hours and 700$ lost each time one fails.

Since I know you all are curious,

These fans will be upgrading the fans in solar charge controllers the fans that are in them have been repeatedly failing every couple years and it has been costing a fortune and much frustration to fix them each time it happens. I need to get ahead of this. Need better fans.

I will follow up with pics and sizes of current fans which need replacing.

Note that a higher cfm unit is highly desirable as it will allow more solar harvesting on hot days and keep the units from throttling. Unfortunately the throttling doesn’t protect from total fan failure.

I know this is the crowd to ask because if anyone know’s fans, it’s you all.

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 1 of 28, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

First fan type is 40mm x 10mm

Attachments

  • image.jpg
    Filename
    image.jpg
    File size
    1.4 MiB
    Views
    950 views
    File license
    Public domain

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 2 of 28, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

Second fan type is 40mm x 40mm 20mm thick

Attachments

  • IMG_3296.jpeg
    Filename
    IMG_3296.jpeg
    File size
    1.19 MiB
    Views
    637 views
    File license
    Public domain
Last edited by Sphere478 on 2024-05-13, 01:29. Edited 4 times in total.

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 4 of 28, by MikeSG

User metadata
Rank Member
Rank
Member

There's some 12V Vapo Bearing fans in 40x20mm that push 10.8cfm. They run at 8000RPM but a reasonable 27.5dBA.

https://au.element14.com/w/c/cooling-th ... 18702_bv|1

12v Sanyo Denki's (ball bearing) push 2-3x more air, at 50-75dBa noise.

https://au.element14.com/w/c/cooling-th ... 18702_bv|1

Reply 6 of 28, by kaputnik

User metadata
Rank Oldbie
Rank
Oldbie

Interesting thread. I'm in the maritime transport business. In addition to ambient temperatures in the engine room at least locally reaching 50-60°C during the summer, we have to deal with vibrations. Also had a hard time finding small axial fans made for use in extreme environments, everything seems to be 80x80mm or larger. The main problem is seizing bearings. Would guess either the lubricant dries up/breaks down due to the heat, or bearing tolerances simply being optimized for lower temperatures.

paradigital wrote on 2024-05-12, 06:59:

Most industrial kit I’ve come across will use Delta branded fans. They seem to last forever.

Yes, they're relatively good, but still fail now and then. EBM Papst is common aswell, but same goes for them.

Reply 7 of 28, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

Giving these a try.

They have ball bearing and I think are a little more air flow.

Attachments

  • IMG_3294.jpeg
    Filename
    IMG_3294.jpeg
    File size
    41.82 KiB
    Views
    751 views
    File license
    Public domain

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 8 of 28, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

This is a interesting link

https://www.cuidevices.com/blog/fan-bearing-t … %20temperatures.

———————————————————————————

https://au.element14.com/sanyo-denki-sanace-f … 4dba/dp/2679917

This one is interesting but idk if the 12v source can handle it.

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 9 of 28, by dionb

User metadata
Rank l33t++
Rank
l33t++

their failure won’t be detected and will result in equipment failure

Given the severe impact when they fail and proven track record of failure, might it not be worthwhile to invest in a means to detect that failure before loss of the cooled hardware occurs, and/or thermal fuses for the hardware so if the temperature gets too high, it cuts out avoiding the USD 700 loss.

Such fuses are simple and common - pretty much every kitchen appliance with a resistive heater has one. They cost a few cents at most. You still get the downtime after a fan dies, but it will reduce costs significantly.

Reply 10 of 28, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
dionb wrote on 2024-05-12, 18:47:

their failure won’t be detected and will result in equipment failure

Given the severe impact when they fail and proven track record of failure, might it not be worthwhile to invest in a means to detect that failure before loss of the cooled hardware occurs, and/or thermal fuses for the hardware so if the temperature gets too high, it cuts out avoiding the USD 700 loss.

Such fuses are simple and common - pretty much every kitchen appliance with a resistive heater has one. They cost a few cents at most. You still get the downtime after a fan dies, but it will reduce costs significantly.

Might not be a bad idea, the location I could install such a device is in a cool part of the device though therefore won’t be accurate a k couple controlled relay may be a better option.

However, I am actually working with one of the designers of the equipment about possible software based protections, and new version of the unit which may include a tach feature. Or better mosfet over current protections.

Software based approach is also possible, but what we want to control to help avoid failure is partially hardware based, it’s a difficult problem. A, K couple triggered SSR would probably do the trick but it’s a bandaid for something that may be solvable more directly

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 11 of 28, by kingcake

User metadata
Rank Oldbie
Rank
Oldbie

Delta. Make sure they are real Delta from a reputable source like Mouser or Digikey.

That being said, placing the "don't fail burden" on the fan is a major system design flaw. Mechanical things will always fail over time. The device should be able to detect the failure.

Reply 12 of 28, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
kingcake wrote on 2024-05-13, 00:43:

Delta. Make sure they are real Delta from a reputable source like Mouser or Digikey.

That being said, placing the "don't fail burden" on the fan is a major system design flaw. Mechanical things will always fail over time. The device should be able to detect the failure.

Agree

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 13 of 28, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

You need redundancy and RPM monitoring. For fans, can't go wrong with Sanyo-Denki and Delta as mentioned. Panaflo was also top tier, but I'm not sure if they've held up now they're made by NMB.

All hail the Great Capacitor Brand Finder

Reply 14 of 28, by Minutemanqvs

User metadata
Rank Member
Rank
Member

From what I see in servers that I'm around since 20 years, running usually up to 6-7 years you often encounter Nidec and Delta fans.
I have not seen a single Nidec fan fail, ever: https://www.nidec.com/en/product/search/categ … B101/M111/S100/

Searching a Nexgen Nx586 with FPU, PM me if you have one. I have some Athlon MP systems and cookies.

Reply 15 of 28, by demiurge

User metadata
Rank Member
Rank
Member

In the nuclear industry we have lots of experience with things that have to work 99.99999% of the time. When that cannot be achieved, we build in redundancy. If I was the engineer I would have to build in a parallel supply fan that could be automagically put online if the other's tachometer read below spec. You might want to consider looking at the MTTF specifications to determine your engineering outcomes.
My personal experience with Delta fans are that they move air like mad and are loud because of it.

Reply 16 of 28, by lti

User metadata
Rank Member
Rank
Member

I would agree with everyone else here. Better quality ball bearing fans (Papst, Nidec, Sanyo Denki, NMB, or Delta) will be better than some sleeve bearing Aliexpress special, but they will still fail eventually. You need fan monitoring.

gdjacobs wrote on 2024-05-13, 04:32:

Panaflo was also top tier, but I'm not sure if they've held up now they're made by NMB.

NMB is a major bearing manufacturer, so you'd hope they're good.

demiurge wrote on 2024-05-13, 23:07:

My personal experience with Delta fans are that they move air like mad and are loud because of it.

Delta is famous for those, but they make low-speed fans as well. They're used in a lot of desktop computers and some of Intel's stock coolers (which aren't known for low noise, but they aren't as loud as a server).

Reply 17 of 28, by darry

User metadata
Rank l33t++
Rank
l33t++

will experience the arizona heat and very hot equipment (as much as 100c) these need to be able to run 12 hours a day every day for decades without failure or any maintenance

While this is not my field of expertise, I believe there are some potentially important variables to consider here

a) Is the ambient temperature expected to reach 100C (boiling point of water) during "normal" operation ?

b) What is the normal operational temperature range of the equipment that needs to be cooled ?

c) is there a serious expectation for the system to operate completely maintenance free for 20+ years (decades) ? Or is it the best possible outcome hoped for with rare maintenance being possible?

d) Will the system be exposed to unfiltered outside air containing dust and other contaminants ?

If ambient temperatures can reach 100C in a dusty environment, where there will be intense airflow and zero cleaning/dusting and/or dust filter replacement, I get the feeling/hunch that even 5 years of reliable uninterrupted operation might be problematic.

Fan redundancy and overspeccing can only take you so far.

Again, I am not an expert in any of this, but this potentially feels like either massive passive cooling using heat pipes or something using a heat pump (compressor) or possibly even multiple load-sharing or redundant ones might be of interest if redesigning this. I suspect the latter might not be practically feasible, but the former, complemented by multiple redundant fans that only need to run for during the hottest times, might be more reliable as it would help limit dust accumulation while also reducing fan-hours.

Reply 18 of 28, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

The internals of the unit have been getting hot enough to brown varnish, brown plastic, makie it brittle, and make the previous fans very hot. During operation in summer the units are too hot to touch on exterior.

It is possible working and higher cfm fans may help with this but it seems to be a normal condition.

Yes ideally these units would be untouched for decades. That was the hope and expectation installing them. Fan maintenance requires significant disassembly and will damage one time use parts getting to them.

They are in dusty air (dirt, not hair or skin, not like a computer environment, think desert air) there is no filter. Dust build uo doesn’t appear to be an issue though. Only light dust on parts that doesn’t seem to keep building up. It may be getting into fans though

Designing a back plane cooling system for them to mount to is possible, I have thought about heat pipe, but where to get or make something like that is a challenge.

Active phase change seems not necessary a heat pipe setup with large passive (18”x18”) with 20-30 6” fins would be ideal but how to make it…

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 19 of 28, by Sudos

User metadata
Rank Newbie
Rank
Newbie

I'm gonna say definitely Nidec would be a brand to look at. either new or new old stock, there are some decent ones in their current and previous model lineups that may benefit you. I wouldn't be surprised if there's an UltraFlo in a 40mm size that will do what you're looking for... but if there's more space for a 20mm thick fan where the first fan is at, I'd do that too. the thicker fans tend to last longer, I've found.

However I think that something like that being untouched "for decades" is a long shot. I'm going to assume there's electrolytic caps and other stuff in play here that will not last more than maybe 5 years. Unless they're of a good quality brand and such, they will give out before the fans will, from the heat. You should definitely look into other ways to keep these cool aside from fans. peltier cooling isn't an option due to diminishing returns, and phase change with heatpipes can definitely do what you're looking for but will require extra engineering and also coatings for the heatpipes to exist in the elements.

What you're basically trying to do here is akin to military-spec hardware in an enclosed space that needs cooled, and it doesn't seem very well designed if it's only making use of 40mm fans. 60mm fans that ramp up and down their speed with change in temperature via a thermistor are doable but I don't think Nidec has anything like that at the 40mm level. smallest I've seen are TA225DC Beta Vs at 60mm. I use one in the PSU of my 386 without an issue, and it stays relatively quiet as it's not just pushing a TON of air through regularly and is instead regulated to a slower speed with the thermistor. shorting the thermistor causes the fan to run at full tilt where it pushes immense amounts of airflow.

8ips6a-2.png
qxkaxq-2.png