Vertical vs. horizontal - what bearings work when?

[Thomc]

I have heard/read that fans with sleeve bearings are quieter than fans with ball bearings - although ball bearings may last longer. I have also read that in general, fans with sleeve bearings work better/last longer when the fan is mounted so that the fan shaft is horizontal - like a case fan mounted on the back of a case, fer instance. Ball bearings, on the other hand, apparently don't much care whether they are mounted vertically or horizontally.

I am no mechanical engineer, but this kinda' makes sense to me when I think about it. It does raise a couple of questions though...

First, does anyone here think they have had first hand experience that would tend to prove - or disprove - the tenet that fans with sleeve bearings don't work as well/last as long when they are mounted with the shaft in a vertical position?

I'm not looking for something that would stand up in court or would be considered statistically significant; general impressions from much experience would be great.

Second question is, how does the Panaflo HydraWave bearing fit in here? Should fans equipped with HydraWave bearings be mounted with the shaft horizontal - or do they work just as well with the shaft in a vertical position?

[silvervarg]

I have done some investigations in this subject, but have not drawn any big conclusions or done scientific long term tests, but I will give a try to answer some of your questions.

First lets make it clear that all good fans have 2 bearings.
What we call ball bearing fans use ball bearings in both bearings.
What we call sleave bearing or hydro wave bearing uses ball bearing in one of the bearings.

Ball bearing fans tend to last longer, but they often start to make irritating noises long before they are "worn out". Once they start to make much noise we want to replace them anyway, so from my point of view they are worn out.
So it becomes very hard to just tell how long a fan is really working. It depends how you define if the fan is working.

If you mount a fan in a vertical position it matters if the fan is bowing air upwards or downwards. Blowing downwards should let gravity pull the fan apart, which is bad. Once it is rotating at a high speed the air beeing pushed down makes a counter-force, that will reduce the wear.
Blowing upwards makes little stress, but the air pushed will give a force in the same direction as gravity, so the wear increases with airflow.

Very fast fan speeds will also increase wear.
A voltage far out from the intended voltage might increase wear, depending on the design.
Sleeve bearings can be made of different materials. Typically plastic or ceramic materials are used. Plastic will generally wear faster than ceramic.

The weight and balance of the moving part of a fan make a lot of difference for how fast a fan will wear out.

Dust can kill a fan several magnitudes faster than it would in an environment with little dust.

Starting and stopping a fan will cause additional wear. It is possible the big changes in speed has the same effect but to a much lesser degree.

As you can see the factors that makes a difference quickly grovs so large that it becomes hard to isolate a single factor without a very structured scientific test.
What I think is a concensus here is:
Sleeve bearing fans in general make less noise than ball bearing when fans are new.
Sleeve bearing fans typically work fine for at least 2 years, and they are rather inexpensive to replace.
_________________
Passive XP2500+: Convection cooled, ~XP1500+, DP-102 cooler, Abit NF7-M, CNPS6000Cu on NB, Antec Overture, Samsung P80 120GB

[jafb2000]

Actually Hydrowave is one of 3 fluid-type bearings:
o Pivot bearing
o Hydrowave bearing
o Fluid Dynamic bearing - development of the hydrowave capsule

None use ball-bearings.
o Pivot bearing gives most of the benefit of Hydrowave at a lower cost,
and in SOME fan sizes can be a little quieter (eg, 60mm re FBZ model).
o Hydrowave give extremely long life
---- actual Panasonic field data found they outlast sintered sleeve by 50%
---- they also outlasted most ball-bearing, matching NMB but quieter
---- field data showed 114,000hrs at 70oC over 1990-1996, revised 2002
o Fluid Dynamic Bearing are a development of the hydrowave capsule
---- these from Panasonic are silly expensive & precision
---- the hydrowave multiple-cam using boundary layer effects is boosted
---- a more complex groove etching in the capsule enhances the effect
---- life is enhanced and particularly dynamic stability - which is why you
find developments of this bearing (by NMB) in the world's hard drives


SLEEVE - noisier with life, but are cheap
o cheap is a reality - can be 2-8x cheaper than a good ball-bearing fan
o life is shortish - 25,000hrs at 25oC on the cheaper fan is common, with
a MTTF of 36% failing quite low, and L10 relatively low too (10% fail)
---- life is particularly short at extreme temps re low or high
---- thus generic sleeve are a bad choice for PSUs
---- hence cheapy PSU fans often use 1 ball & 1 sleeve (& call them 2-ball)
o noise grows - from day 1 by 12-15dB(A) rise over 30,000-40,000hrs

PAPST SLEEVE - Sintec - quiter over life, longer life, pricey
o pricey, but good life - 40,000-80,000hrs is a reality not a gimmick
---- high temp data is ok since the oil is good
---- works by sintered bronze bearing = porous = holds more oil
---- oil oxidation rate is the dominant factor in bearing life
---- at 50% oil loss, the bearing will degrade relatively rapidly & noisily
o noise grows - 3dB(A) rise over 20,000-32,000hrs
---- that's pretty good, and basically meant Papst won until 1990
---- when in 1990 Panasonic moved hydrowave from videos to fans
---- hence papst moved 80mm production to hungary from germany

A sleeve bearing fan like Sintec is a 1-main-fan bearing & 1 thrust washer.
A lot of the harmonics/ticking you hear is the thrust washer, but it is not
a life threatening - or dB(A) threatening issue, consider it "characteristic".

BALL-BEARING
o pricey - can be extremely pricey, or relatively crude & cheap
---- Adda typically a bit gritty, Papst a bit industrial
---- NMB from another planet, but really horribly expensive
---- you also have 2 ball bearings in a fan remember
o life - can be good 40-50,000hrs at 25oC for cheapies
---- or can be 200,000-350,000hrs at 25oC for NMB
---- Papst will do 120,000hrs at 25oC, 80,000hrs at 40oC typically
o noise - 5-12dB(A) rise over 30,000-40,000hrs
---- cheapies are 12dB(A), but NMB is actually at the 4-5dB(A) level
---- more importantly, the NMB have a super-light rotor
---- which means they tend to have less harmonics as they age

NMB Ball are a special ball-bearing:
o Pre-load is carefully managed to avoid thrust-shift ticking (Papst Sintec)
o Bearings are assymmetic, opposed, so complimentary to fan use

Nidec, Adda (Hypro) & others have variations of sleeve.
They can get down to 2dB(A) rise - but over a short 20,000hrs and
it will begin to climb more rapidly after that. It's all product-price-mix

Fluid dynamic bearing derivatives - Panasonic in fans, NMB in HDs
o makeup - no ball-bearing
---- fluid dynamic bearing via non-cylindrical shaft
---- thrust bearing is limited to that required for ruggedness
------ you actually have a separate capsule & need thrust washer/bearing
------ otherwise there's no way to keep the rotor on the shaft obviously
o noise
---- Panasonic Pivot - 0dB(A) rise over 20,000-30,000hrs
---- Panasonic Hydrowave - 0dB(A) rise over 30,000-40,000hrs
---- Panasonic Hydrodynamic - 0dB(A) over life >1M hrs (at huge price)
o life
---- Panasonic Hydrowave - actual field data is 114,000hrs at 70oC (L10, 10%)
------- that is data from the Mk1 for 1990-1996, revised in 2002
------- if comparison is by MTTF the figure is around 350,000hrs (36.8%)
---- the Hydrodynamic would be >1M hrs, like in Seagate Savvio
------- as yet it's not been implemented in fans, it's too expensive

Considering most fans run at <50oC, I don't think the Hydrodynamic
bearing has legs due to enormously expensive precision costs. It's a
hydrowave capsule, same huge heavy lump as FBA, but very expensive.
Could be NMB can resolve that cost issue - like they did for the Hard
Drive version - eg, 10-15mm bearing-motor-combo in the world's HDs.
That could be a focus of the merger to create M3, and tech papers do
seem to indicate they want to migrate to that level at some point.

That is the key differentiator of the Panaflo over all other fans.
Next will be the Panaflo Pivot bearing - which is actually a bit quieter,
for example the 60mm FBZ is 19dB(A) for nearly the same cfm as the
24dB(A) FBA 60mm fan - but doesn't stay so quiet for as long a life.

Re vertical/horizontal mounting:
o Sleeve bearings prefer vertical-fan mounting & low temps
---- there's nothing to stop oil getting out
---- oil is oxidised based on temperature re intake v PSU fan use
o Hybrid sleeve bearings, hypro, Nidec's, few others tolerate horizontal
---- not sealed, metal contact normal so tolerances varies & oil loss rises
---- design basically limits the oil loss, but still tolerance will vary
o Ball-bearing fine in any angle (*)
---- exception is for high loads - where you want flat or vertical
---- some early high-rpm non-fluid-bearing HDs disliked 45-degrees
o Fluid Dynamic Bearings
---- any angle
---- marginally you may find some change on Panaflo since you have a
tiny extremely low load thrust bearing noise (swish) coming in. You can
prove this by holding a fan horizontal intake-up or intake-down, and so
vary the load on the thrust bearing bringing it in or out of the noise.
---- the most rugged fluid dynamic bearing isn't yet in use in a fan
------ it's in use in hard drive motors down to 15-10mm in size (!)
------ and it is very very expensive, Panasonics the most expensive
------ but Panasonics clever beyond-hydrowave capsule etching will allow
it to be used in hard-drives into super-high rpm as disk size falls, recall a
15,300rpm 3.5" disk drive is actually using 2.5" platters for this reason

The Hydrowave bearing falters at very high-rpm - you end up with not
a boundary layer separation reflected high pressure wave from the cams
in the fluid dynamic capsule, but cavitation and so a LOW pressure wave.
Hence hydrowave aren't used in super high rpm fans - Sanyo own that
technology and use it commonly in the 40x28mm fans. You will see more
of it from Delta in their 56dB(A) screamers for 1U rack use in 2004/5, at
23cfm at 0.6" pressure - 50x40x38mm to allow quad-opteron in 1U and
high density 2.5" SCSI 10,000rpm drives from Seagate (and later Fujitsu).

So if money is tight:
o Choose Ball or Sintec or Panaflo for your PSU exhaust fan
o Choose cheaper fans for the intake fans

Then it's up to you re overall life-tradeoff & noise-over-life-tradeoff.
There is quite a large decision matrix re quantitative & qualitative
parameters in choosing a fan, eg, Dell on top-line servers will fit NMB,
on cheaper will fit Nidec, and Jameco & similar on bottom line servers.

Brand doesn't necessarily indicate bearing choice:
o Some YS-Tech use a good SKF ball-bearing
o Some YS-Tech also use a primitive transistor motor switching system
---- no complex soft-switching motor ic's to maintain torque & so airflow
---- that incidentally is how they vary the fan speeds

Time may see the "S" spec Panaflo come out in other than 60mm FBZ,
but as yet production capacity is flat out re demand. Will take time.

Sleeve bearings have thrust washers, which contact the oil supply in
the sleeve - and so as a result promote capillary migration as the fan
spins. You can minimise this with oil choice, but it's a balancing act.
A hydrowave capsule for comparison is completely sealed, and the
thrust system is purely a hardened steel tiny floating shim behind a
small bronze bushing on the shaft (below the label, below a cap).
So there is no connection or co-dependency on the two, it's a trick
to get beyond the 40G rating of ball-bearing and up to 60G actually

Typed quick, hope it makes sense & proves of interest
_________________
www.stores.ebay.co.uk/panaflofan for fans, books & other items

[alglove]

Typed quick?! My jaw dropped when I read this post! The detail is awesome!

[Thomc]

jafb2000 - wow! A million thanks for all the information - I hope you can type faster than I can! Like alglove, I was blown away.

Your author field indicates that you are a vendor - can you tell me what and for whom you vend? Surely you didn't pick up all this info working in PC components wholesale/retail. It sounds more like you are a mechanical engineer involved in industrial component housing design - or possibly in designing bearings yourself?

One of my original questions had to do with whether sleeve bearing fans worked better when the fan shaft was mounted horizontal. In your post, you stated that sleeve bearing fans worked better with a "vertical-fan mounting". Am I correct in assuming that you mean by this that the fan blade is vertical - and shaft running through the bearing is horizontal?

I am going to have to re-read and re-read your posting many times before I feel like I have gotten my head stretched all the way around it. In the mean time, do you know of any web sites that have good line-drawing of cut-aways of these fan motors? I would really like to understand the location of the thrust washers - and how the different bearing types handle the axial thrust. Even with the shaft mounted horizontally, a fan bearing is still obviously subjected thrust because of the pull of the fan blade.

You sound very knowledgable about these fan bearing issues - things like Papst moving their production from Germany to Hungary. Can you give any insight on whether Panaflo fans really do have lower quality now that they are no longer built in Japan? If both Papst and Panaflo have dropped their quality due to moving into cheaper manufacturing, who is still building high quality product? Near the end of your post, you give a nice little synopsis of your recommendation of what to do if money is tight. Suppose money isn't tight - what fans would you recommend if the main concern is quiet?

[silvervarg]

jafb2000 is a vendor that sells Panaflo, NMB, Adda and Delta fans. At the moment he doesn't have any Adda it seems.
He is known to supply quite a few countries in europe with Panaflo fans.
I have never seen he sell any 120mm fans.
For specific vendor information I think you should go to the SPCR vendor section, so we keep clear from the grey-zone of advertising in the normal forum categories.
_________________
Passive XP2500+: Convection cooled, ~XP1500+, DP-102 cooler, Abit NF7-M, CNPS6000Cu on NB, Antec Overture, Samsung P80 120GB

[jafb2000]

o Vendor - fans (Panaflo, NMB, Papst, some Adda/Delta)

o Sleeve - yes, fan blade vertical

o Line drawings - Papst & NMB have pictures
---- Panasonic Panaflo have some cut-away drawings
---- however a fluid bearing isn't much to see since they rely on fluid-dynamics
---- the discrete components are somewhat difficult to make out on the scanned images

Typo on the Panaflo fluid bearing - it IS fluid-bearing on axial-thrust ALSO.

Panasonic Panaflo bearing design visualisation:
o Working from the intake-side to exhaust-over-struts label-side
---- rotor -- super-thin tunnel vortex design
---- capsule - big heavy fluid-dynamic-Radial-&-Axial-Thrust-bearing capsule
---- thin-shim - non-circular shim with non-circular hole to limit noise
---- thrust bushing - end-cap not related to the fan's internal thrust bearing
o Capsule is huge & heavy
---- large capsule - pushes the motor further out (obvious on 120mm Panaflo's huge hub)
---- heavy capsule - a major weight component (80mm are heavy, 120mm 2-3x the weight of NMB fans)
o Capsule is the fan bearing
---- main cams - provide Radial fluid-dynamic pressurisation & pumping of oil
---- cam profiled washer - provides Axial-Thrust fluid-dynamic pressurisation & pumping of oil

The outside-capsule thrust bushing:
o Axial positioning of rotor/bearing w.r.t. the stationary motor
o Axial shock mounting for higher G rating than ball & offloads non-running load from fan

That outside-capsule thrust bushing doesn't sit against the capsule but has a shim in the
way which is non-circular and non-circular hole in it to stop rotation noise/ticking. Some
of the fans may tick just perceptibly, but I filter those fans out myself. The Papst tick
is louder because it's a) ball-bearing play or b) thrust-washer axial/radial components on
the Sintec fans, thus inside the main bearing itself directly. They are too hard to filter,
since Papst vary say 6-9 on broadband noise, Panasonic about 8.5-9.5 on narrowband noise.

If you put a lot of Papst together, the broadband noise is very clear & so are variations
within that noise as you get reinforcement, beat frequencies & focusing. If you put a lot
of Panaflo together, you STILL get narrowband noise and the tiny variations that you may
pick up at 3-4" from your ear vanish without reinforcement as noise/vibration are low.

A comparison of noise/vibration between fluid-bearing panaflo & ball-bearing is at:
http://homepage.ntlworld.com/dorothy.bradbury/panaflo.htm (Direct)

I'll try and get a c/section CAD rendering of the 120mm Panaflo fluid bearing, since that
is where you can see the radical differences between itself & other fans. The Panasonic &
NMB merger (they have co-worked for decades, and share most of the patents) is as much
about getting that motor & bearing design into general motor applications. Yes, you can
fit a quiet Panaflo fan, but what about the same bearing technology throughout equipment?
Panasonic use the bearing in their whisper HVAC systems - but lots of motors in the world.

The existing c/sections I have a panasonic scan-ins mainly showing the difference between
the cheaper fluid-dynamic Pivot & the longer-life fluid-dynamic Hydrowave (normal Panaflo).
Basically it's a big fluid-dynamic bearing capsule, which operates in both axial & radial.


120mm fans
o Have sold Papst & NMB in the past, and a few Panaflo M1A
o Barrier has long been 120mm Panaflo weight - it's 3x the weight of others re shipping
---- barrier has now gone
o New barrier is the distributors don't have any
---- no L1A 12V, few M1A 12V, mostly 24V since SGI/SUN/DEC etc super-servers use them
o L1BX & M1BX are not in production, and haven't been for a very long time remember
---- so they would be 2-wire non-tacho if ordered eventually

Bear in mind I select my fans, so whilst I may order, I may also return & not sell.
That happened with the Papst 8412NGMLV/2 some may recall, which I pulled & returned.


Note Panaflo 120mm are 38mm only - no fluid-bearing will EVER be made in 120x25mm.
If you need 120x25mm, see the Papst or NMB - NMB is a better fan, but I can say the
Papst 4312MV is perhaps the nicest 120mm altho some, yes, may have a little tick. That
is a Papst ball-bearing or Sintec tick, which is more noticeable than other fans.

40x10mm Adda
o Mkt is flooded with old discounted &/or out of production stock - so I can't compete
o Indeed distributors cancelled their own orders until that competitor's supply is gone

Temp control is on the way out, instead replaced by current control - no lag/latency etc.
Temp control of laptop fans, Panaflo or not, is a game in itself re noise/lag

I'll shut up now and this hasn't been proof read either!
--
Dorothy Bradbury
www.stores.ebay.co.uk/panaflofan for fans, books & other items
http://homepage.ntlworld.com/dorothy.bradbury/panaflo.htm (Direct)