I would go for number 3

The old trick was to wire a 12v fan to 5 and 12 =~7v.

The less dodgy version is a resistor/fan controller

ahah, never considered the 7V version!

Yes, maybe option 3 would be the best. I'd end up with a modern PSU, probably plenty of airflow (the original PSU has an 80mm fan) and reliability.

The issue would be to find a PSU with a fan which can deliver a good airflow at maximum. That figure would not be disclosed on a datasheet I think. Also, a PSU with old molex power connectors. Yes, adaptors can be bought.

Any recommendations? I need to check the original PSU but I don't think I need more than 2-300W here. I'll share the exact specs asap.

Edit: worse case scenario I could replace the fan with one with good airflow. Noise is not a concern here.

Heey! It's Tony!

I would also go with option 3, but heavily modify the PSU. Remove all the unnecessary connectors and maybe even re-use the, cooler, multi-color 20-pin from the old unit. Heck, use the old everything. There is also usually a plastic flap covering half the fan. I'd take that out, if equipped.

The most Important thing is which way the fan was originally pointed. If it was pulling, like 99% of PSUs, a 120mm fan, even on LOW will move more air than the 80mm, but doesn't "feel" like as much because the static pressure is lower. As long as air is free to be pulled through the case, the 120 will work just fine for cooling passively cooled components.

Now, for the voltage, I'd say it depends on the fan. A stock PSU fan may not even be noisy at 12V, so 12V would be perfect. Some 120s can hit 3k+ RPMs, though, so those type, I would give 5V. Check the fan's CFM rating. Plain-jane 120mms, typically around 100CFM on high. An industrial 80mm hitting 5K will still only get 60CFM, but be noisy as F. A 120mm on 5V may still be able to move 40-50CFM, if the fan is quality. If there is room, putting the fan outside the PSU can give the fan more room to breathe as well.

CFM is important to keep is mind, as this will also change the rate at which dirt is packed into the system, 🤣

Have a old Compaq that relied on the PSU for near all airflow from the PSU (Same no fan on the HS). So for me #1 or simpler #3 on a newer and yes and direct wire the psu fan to 12v, not some in between due to when the cpu is working hard the fan will never run full and could still cause overheat issues...

I routinely mod old PSUs with modern fans. You don't want to skimp on airflow, but a good modern Noctua can move more air virtually silently than a rattly old sleeve-bearing does.

It's also possible to do it the other way round, but challenge here is that modern PSUs tend to have the fan in a different location and have a much larger vent, resulting in slower, less focused airflow even if the whole volume of air moved is greater. That is a risk when combined with passive CPU heatsinks.

dionb wrote on 2025-06-12, 07:05:

You don't want to skimp on airflow, but a good modern Noctua can move more air virtually silently than a rattly old sleeve-bearing does.

I've actually played around with it a bit - noctua seems to have cult-like following in some communities (like 3d printing), but reality is - there is no magic here - low noise and high airflow are mutually exclusive.

This basically are low RPM/low airflow fans with good bearings. Nothing special about them. The secret specifically is low airflow - because as airflow goes up, even if bearings are dead quiet, airflow itself starts generating noise. Especially if it goes through stuff like typical meshes cut in sheet metal with sharp edges.

If you take a fancy noctua fan and any old generic one (but not worn out/faulty obviously) and match their RPM you are going to get very similar results - both noise and airflow. Especially with sleeve bearings, as those tend to be quieter than ball bearings and FDB ones were not used in fans back then.

The only advantage, theoretically, is longevity - FDB should last for a long, long time. Theoretically because in practice they still fail...

If it's not possible to cut a hole for an 80-120 fan somewhere on the case, what about using a blower in the PCI slot if one is free https://www.titan-cd.com/en/product/TTC-003.html? Some airflow will create it after all.

Thanks for all the hints!

On reusing the wiring: those PSUs can be busy inside, swapping wiring is not a simple task. But can be an option. I need to check exactly what I need.
On swapping the fan or just using a new PSU with fan at 100%: probably a good idea. My only concern is whether those fans are designed to be used at 100% for many hours. I might look into swapping it with a good quality, industrial Delta fan. That said, that box shouldn't be used a lot so maybe I am overthinking here

Adding fans is probably not a good idea. It's a cinema SOUND player and I know how easy it is to unintentionally inject fan noise into the outputs!

Yes, I also made a Noctua mistake before - nothing wrong with Noctua but as you say, airflow and noise are walking hand in hand. I almost blew up my NAS PSU because of that. And I resolved by wiring the very same fan to 12V directly!

Ok, let me take a look at the box, I'll see what the power required is and count the plugs. Then I'll move from there.

I think I'll make a video about it, it's quite a unique box - nothing special as a "PC" but it's not something you'd see around a lot!

Cheers!

Archer57 wrote on 2025-06-12, 07:27:

I've actually played around with it a bit - noctua seems to have cult-like following in some communities (like 3d printing), but […]
Show full quote

dionb wrote on 2025-06-12, 07:05:

You don't want to skimp on airflow, but a good modern Noctua can move more air virtually silently than a rattly old sleeve-bearing does.

I've actually played around with it a bit - noctua seems to have cult-like following in some communities (like 3d printing), but reality is - there is no magic here - low noise and high airflow are mutually exclusive.

This basically are low RPM/low airflow fans with good bearings. Nothing special about them. The secret specifically is low airflow - because as airflow goes up, even if bearings are dead quiet, airflow itself starts generating noise. Especially if it goes through stuff like typical meshes cut in sheet metal with sharp edges.

If you take a fancy noctua fan and any old generic one (but not worn out/faulty obviously) and match their RPM you are going to get very similar results - both noise and airflow. Especially with sleeve bearings, as those tend to be quieter than ball bearings and FDB ones were not used in fans back then.

The only advantage, theoretically, is longevity - FDB should last for a long, long time. Theoretically because in practice they still fail...

Disagree here. The difference isn't just the low airflow, it's also the blade design and the mounting. The edges of the blade are slightly feathered which reduces noise compared to similar slow fans without the design. Also the mounting points are made of silicon rubber rather than hard plastic, which dampens vibrations. Interestingly, I find I prefer the sound of the cheaper, simpler Noctua Redux fans to the regular expensive ones.

Slap a Sunon MagLev fan in there and a manual fan control resistor and set it for whatever airflow and noise you find acceptable.

dionb wrote on 2025-06-12, 17:46:

Disagree here. The difference isn't just the low airflow, it's also the blade design and the mounting. The edges of the blade are slightly feathered which reduces noise compared to similar slow fans without the design. Also the mounting points are made of silicon rubber rather than hard plastic, which dampens vibrations. Interestingly, I find I prefer the sound of the cheaper, simpler Noctua Redux fans to the regular expensive ones.

Everybody is going to have their own opinion and sometimes this topic can get as hot as intel vs amd, but my opinion is:

Blade shape: A lot of manufactures experimented with a lot of different stuff here, some still do. And this has been in marketing a lot. "Shark fins", "bat wings", whatever, a bunch of silliness. It is one of those things which sound plausible and theoretically can make a difference, but practically does not. At least not enough of a difference to matter or be outside margin of error in specific application. Which resulted in many manufacturers gravitating back to simpler stuff after it stopped being useful for marketing.

Mounting: It is not very useful because they only go half of the way. You still use a screw which presses against the case and is screwed into hard plastic on the fan. Zalman has been shipping their fans with mounting hardware which avoids it entirely for decades at this point and can be used on any generic fan. Sometimes does help, mostly when the fan is worn/half dead and vibrations become significant or RPM is really high. On a fan which is working properly and is not running at very high RPM makes very little to no difference.

From what i've practically seen fans from sunon, delta, etc. tend to be practically the most reliable and if you slow them down to comparable RPM - quiet. But that's just my experience...

Slap a Sunon MagLev fan in there and a manual fan control resistor and set it for whatever airflow and noise you find acceptable.

Yes, that is probably the best option, thanks.

A quick comment on fans: yes, plenty of marketing in there. Though the fixing hardware does make a difference. As long as the screw is NOT touching the chassis (or the fan frame) on BOTH ENDS, it will do a massive difference.

myne wrote on 2025-06-11, 15:51:

The old trick was to wire a 12v fan to 5 and 12 =~7v.

The less dodgy version is a resistor/fan controller

Wiring it to 12V and 5V together sounds like a pretty bad idea..

The resistor is pretty easy to do, just find the Amperage of the fan and calculate its impedance, then you can calculate the resistors required to bring the 12V rail down to about 7-9V.

As an alternative to a resistor - one of those adjustable DC-DC converter modules can be used. Seems like overkill, but they are cheap enough, convenient and do not get hot unlike a resistor...

Wiring it to 12V and 5V together sounds like a pretty bad idea..

Not together - across. Basically 12-5 = 7V. The fan is not referenced to ground so it'll just see 7V DC.

I'd like to avoid those little chinese DC-DC converters, they'd probably be the weakest link in the whole box. But again, noise is not a concern.

I've now inspected the machine, it's got an Asus PSP800-VM in it. The PSU is 250W and I need 4 Molex, 1 Floppy and 1x20pin ATX. That's it.

The size is ATX, the manufacturer is still around "Sparkle Power" which reminds me of Homer Simpson's Mr Sparkle! 😀

I might end up "upgrading" the HDD to Sata so only 3 Molex would be required. I'd imagine I could buy some good quality Sata to Molex adaptors? Not the Aliexpress one which catch fire - I speak for experience, one caught fire inside my PC and I mean there was a naked flame coming out of the connector!

The PSU is also attached at the back with a bracket but I guess that's not so important. The air opening must be at the top though - with the power plugs facing right when looking from the back.

I found this online on Digikey (= not chinesium and not with silly features I don't need): https://www.fsp-group.com/download/pro/FSP300 … B_Datasheet.pdf

It seems to have exactly what I need, a small-ish fan at the back as the original, probably spinning fast most of the time, a little more power than the original (I'll check the individual currents) and vents in the right place. Connectors are also perfect. That's great as I could only find 500W+ PSUs for modern PCs 🙂