Help, PWM fan on a non-standard HP 4-pin header?

[FelixC]

I picked up an HP 8300 Elitedesk USDT, the really small one, but the chassis fan was broken out of the box and the front one's kinda noisy (both 60x25mm), so I'd like to replace both, but I'm having a problem...

The fan headers on the motherboard are 4-pins but the plastic "lock" tab on them doesn't match a standard 4-pin PWM connector. Specifically, the lock runs the full width of the header and HP's fan connector has ridges on either side, rather than separating one pin from the other three. Here's a similar comparison picture from an old discussion I came across, but it ended inconclusively:

I used to know that a 4-pin header always meant PWM, but now I've been reading that's no longer necessarily the case? That some will be PWM, others will be regulating speed through the voltage on the 2nd pin? I've tried looking up the actual motherboard schematic from HP but no luck, all I've turned up is the part listing for the HP fan itself:

691352-001 In stock
Cooling fan assembly, size 60mm x 25mm, rated at 12VDC, 0.70A (USDT)

Now I know that cheap 3-pin fan will work just fine this header, presumably at full tilt (tested with a spare 80mm I had around, just enough to make contact on the pins), but I was hoping to get a pair of Noctua NF-A6 units to replace the broken one at the back and also the front one for the CPU because of the noise issue. That model's a 4-pin PWM, but I figured I could just force the connector in or break off the lock on the header or file off the ridge on the fan connector.

But now I'm confused, if those 4-pin headers on the motherboard aren't PWM type, would a PWM fan like the Noctua work properly, let alone scale RPMs? The Noctuas are a little pricey and I'd be very grateful for some advice before I go and make a mistake.

 

[pendragon1]

wiring looks the same, id just jam it on there.

 

[Nobu]

What you could do, if you have a multimeter, is hook up your (good OE) fan and power it up, then probe each of the pins to see what they are. If they match what a standard PWM fan expects, you can just trim the blocking tab and put your noctua on it.

 

[FelixC]

wiring looks the same, id just jam it on there.

It's not my picture, I just posted it because it shows the same connector key ridges layouts side by side, left is how the HP fan is, keys on either side, right's a regular PWM.

I guess my main question is - ignoring the ridges, which I could just file off, would plugging a 4-pin PWM fan in a non-PWM, voltage-controlled 4-pin header still work properly or at all? Both the HP fan and the Noctua are listed as 12V and HP won't have developed their own new standard, so that header's gotta be either PWM or DC-based.

Noctua's got a little article here outlining the two different solutions, but they don't actually spell out whether I can safely plug a PWM fan in a DC header. Judging by the below diagram, the first three pins should be fine, but it's that last "or +5V" bit on the fourth pin that has me worried - forgive me, my electronics lessons are long behind me, but if the motherboard "puts" that 5V on the 4th pin where the Noctua's expecting a PWM signal, would that cause a problem?

What you could do, if you have a multimeter, is hook up your (good OE) fan and power it up, then probe each of the pins to see what they are. If they match what a standard PWM fan expects, you can just trim the blocking tab and put your noctua on it.

Unfortunately, I don't have a multimetre or anyone to borrow one from. The above schematic lays out the theory, but I just don't know what to make of that 4th pin discrepancy.

 

[Nobu]

It's not my picture, I just posted it because it shows the same connector key ridges layouts side by side, left is how the HP fan is, keys on either side, right's a regular PWM.

I guess my main question is - ignoring the ridges, which I could just file off, would plugging a 4-pin PWM fan in a non-PWM, voltage-controlled 4-pin header still work properly or at all? Both the HP fan and the Noctua are listed as 12V and HP won't have developed their own new standard, so that header's gotta be either PWM or DC-based.

Noctua's got a little article here outlining the two different solutions, but they don't actually spell out whether I can safely plug a PWM fan in a DC header. Judging by the below diagram, the first three pins should be fine, but it's that last "or +5V" bit on the fourth pin that has me worried - forgive me, my electronics lessons are long behind me, but if the motherboard "puts" that 5V on the 4th pin where the Noctua's expecting a PWM signal, would that cause a problem?

View attachment 455867

Depends on what the fan expects the voltage levels for high and low on the pwm signal wire, but if 5V is considered high, it'll probably run the fan at 100% duty cycle. If it's low, then it'll run minimum speed.

The issue would be if it's too high for the pwm circuit to handle. Noctua should have that spec somewhere on their site.

 

[pendragon1]

It's not my picture, I just posted it because it shows the same connector key ridges layouts side by side, left is how the HP fan is, keys on either side, right's a regular PWM.

I guess my main question is - ignoring the ridges, which I could just file off, would plugging a 4-pin PWM fan in a non-PWM, voltage-controlled 4-pin header still work properly or at all? Both the HP fan and the Noctua are listed as 12V and HP won't have developed their own new standard, so that header's gotta be either PWM or DC-based.

Noctua's got a little article here outlining the two different solutions, but they don't actually spell out whether I can safely plug a PWM fan in a DC header. Judging by the below diagram, the first three pins should be fine, but it's that last "or +5V" bit on the fourth pin that has me worried - forgive me, my electronics lessons are long behind me, but if the motherboard "puts" that 5V on the 4th pin where the Noctua's expecting a PWM signal, would that cause a problem?

View attachment 455867

Unfortunately, I don't have a multimetre or anyone to borrow one from. The above schematic lays out the theory, but I just don't know what to make of that 4th pin discrepancy.

if the header isnt sending pwm it will run full speed.
the only consumer oem that ive seen change headers is dell and they just reversed the pwm/rpm pin, and you could just swap them to match.

 

[cdabc123]

Voltage controlled headers normally work just fine for pwm fans. The 5v pegs the pmw high and the +v regulates motor speed. They only spot you would get into trouble is sending 12v where it should not be. Probe the header to find 12v and ground and go from there

 

[pendragon1]

Voltage controlled headers normally work just fine for pwm fans. The 5v pegs the pmw high and the +v regulates motor speed. They only spot you would get into trouble is sending 12v where it should not be. Probe the header to find 12v and ground and go from there

Unfortunately, I don't have a multimetre or anyone to borrow one from.

 

[cdabc123]

https://www.harborfreight.com/elect...ters/7-function-digital-multimeter-63759.html

I think there are coupons to grab it for $1 or so

or alternatively, plug it in. The only time I ever broke anything fan related was pluigging a 8pin dual rotor fan into the wrong connector and killing the fan driver chip.

 

[FelixC]

Voltage controlled headers normally work just fine for pwm fans. The 5v pegs the pmw high and the +v regulates motor speed. They only spot you would get into trouble is sending 12v where it should not be.

Well, I just switched tracks a little bit and ordered a pair of Noctua NF-A6 FLX units instead, these are 3-pin models meant for voltage control instead of PWM. They were the last two in stock, so I hit it. My rationale is I looked up both of the fans the HP came with, the AUB0612VH for the chassis and the AVC DS06025B12U for the CPU, and couldn't find the word "PWM" anywhere. The amps differ but if I recall correctly, they'll draw what they need. I'll probably still have to file a ridge off the connector thanks to HP, but hey.

The issue would be if it's too high for the pwm circuit to handle. Noctua should have that spec somewhere on their site.

I found this white paper for Noctua's PWM, gave it a skim and it does indeed seem like 5V is the standard on the 4th pin. So I can't be sure, but I'm guessing it would've been fine. And searching some more, Noctua do have some FAQs on their site that address fans idling at above normal speed or stopping altogether (gotta increase your minimum Idle speed in the BIOS) when plugging a PWM into a DC-controlled 4-pin.

if the header isnt sending pwm it will run full speed.
the only consumer oem that ive seen change headers is dell and they just reversed the pwm/rpm pin, and you could just swap them to match.

That kinda clued me in... it would've been an option to just pull the 4th pin from the connector, use it as a 3-pin. But then I figured why not just get a 3-pin and save a couple of bucks in the process? And if they run at 100% it's no big deal, their 20dB will still be much quieter than HP's 40dB lawnmowers.

Anyway, order's in so I'll post back once I can test 'em in case others run into the same issue. Looks like shipping's gonna be about a week, though.

 

[FelixC]

Well, results are in, guess I should've gotten the PWMs after all. Got the new fans, ended up snapping the locking tabs off the motherboard headers (half by accident), and...

The 3-pin Noctuas work at full speed when plugged in. The original 4-pin CPU fan, however, does scale RPMs, which I could test with the minimum Idle speed setting in the BIOS. So I guess the process of elimination would suggest those 4-pin headers on the motherboard are PWM after all. Might be useful to someone else in the future.

On my end, it'll do. The original fan buzzes at low speeds and sounds like a hairdrier at max, but the Noctuas are nice and quiet with the ULNA plugs in. Temps are up a couple of degrees, 34C idle, 43C playing local video or DVD, and a bit more for YouTube video at 49C with the CPU ~35% use. Not ideal but tolerable for my use cases, worst case I swap the ULNAs for LNAs at the cost of a few extra dBs.

Bottom line, if I were doing it over I'd probably try the PWMs, but this is okay too.

 

[DanNeely]

It's not my picture, I just posted it because it shows the same connector key ridges layouts side by side, left is how the HP fan is, keys on either side, right's a regular PWM.

I guess my main question is - ignoring the ridges, which I could just file off, would plugging a 4-pin PWM fan in a non-PWM, voltage-controlled 4-pin header still work properly or at all? Both the HP fan and the Noctua are listed as 12V and HP won't have developed their own new standard, so that header's gotta be either PWM or DC-based.

Noctua's got a little article here outlining the two different solutions, but they don't actually spell out whether I can safely plug a PWM fan in a DC header. Judging by the below diagram, the first three pins should be fine, but it's that last "or +5V" bit on the fourth pin that has me worried - forgive me, my electronics lessons are long behind me, but if the motherboard "puts" that 5V on the 4th pin where the Noctua's expecting a PWM signal, would that cause a problem?

View attachment 455867

Unfortunately, I don't have a multimetre or anyone to borrow one from. The above schematic lays out the theory, but I just don't know what to make of that 4th pin discrepancy.

For future reference, if anyone else is really concerned: You could make pin 4 not connected by just cutting the wire on the fan.

Non-PWM fans will then just run at full speed on a PWM header. PWM fans on a voltage controlled header should still work more or less; but if may stop spinning entirely if you lower the voltage too much. (You need to monitor RPMs when adjusting their speed since friction makes them go very non-linear as the voltage drops towards the won't spin at all point.)