Zalman Fanmate-2 speed controller crashed my PC

[redwire]

Sometimes I'd hear a coil whine which kept getting lower in pitch until poof blue screen on a PC.

Tried everything, thought it was VRM instability and basically lived with it for 2 years.
Other day I hear the whine again and rip the PC apart and find a cooling fan stalled and whistling. Nudge the fan it works, no whistle.

Investigated and Zalman Fanmate-2 speed controller is a piece of garbage.
Uses a linear 7805 regulator, no capacitors anywhere.

So what happened is the 7805 would oscillate and the fan would whistle, I think sending bogus high freq. tach signal to motherboard and then it blue screens.

Turfed the Zalman and the PC is stable like a rock.

Unbelievable, after 2 years of this.
Who'd thing a fan tach can crash a PC?

Link to a guy's teardown of the Zalman Fanmate-2. Yes it's a garbage design.

Gigabyte GA-Z97X Gaming GT motherboard bluescreen

[eccerr0r]

Re: Zalman Fanmate-2 speed controller crashed my PC

I figure most of the time, to keep a computer stable, cool it as much as you can -- slowing a fan down simply doesn't make sense...



unless it causes someone's head to go unstable... Mine filters out the nearly white noise fairly well.

[redwire]

Re: Zalman Fanmate-2 speed controller crashed my PC

This was driving a HDD cooler fan, pointing right at the controller pcb.

It's slim- the sheet-metal is pretty restrictive and high/full speeds just made too much noise, all turbulence and no better airflow.

[momaka]

Re: Zalman Fanmate-2 speed controller crashed my PC

Interesting. I wonder how the fan controller can actually cause that. Perhaps it injects lots on noise on the fan tach line, thus causing the motherboard to go unstable?

Gigabyte GA-Z97X Gaming GT motherboard bluescreen

With a motherboard like that, I imagine you should be able to control the fan speed directly from the board. If you have spare 3 or 4-pin fan headers, just grab some spare female connectors and rewire your fan to be controlled by the motherboard's fan headers. These days, most motherboards offer PWM as well as Voltage-mode control.

I figure most of the time, to keep a computer stable, cool it as much as you can -- slowing a fan down simply doesn't make sense...

I disagree.

Not everyone likes loud PCs. I am using quite a loud PC myself, and although I am used to it, sometimes it can get very annoying in the summer when the temps in my room kick up close to 30C/86F.

Also, the fans wear out much faster at higher speed. Not to mention they collected dust faster, as well as the whole PC. So there is legitimate reason to not go overboard with the airflow.

This was driving a HDD cooler fan, pointing right at the controller pcb.

It's slim- the sheet-metal is pretty restrictive and high/full speeds just made too much noise, all turbulence and no better airflow.

Ah, I see Molex connectors.
Just mod the connector so the fan is running either at 5V or 7V, whichever works better with you. Then you won't even need a fan controller.

[eccerr0r]

Re: Zalman Fanmate-2 speed controller crashed my PC

I disagree.

Not everyone likes loud PCs. I am using quite a loud PC myself, and although I am used to it, sometimes it can get very annoying in the summer when the temps in my room kick up close to 30C/86F.

Also, the fans wear out much faster at higher speed. Not to mention they collected dust faster, as well as the whole PC. So there is legitimate reason to not go overboard with the airflow.

Yes, as said, some peoples heads go unstable with loud fans. However it can't be denied that the cooler the CPU is kept, the longer the CPU will last and less likely that the cpu will go unstable. Yes, longevity of the fan may suffer but that's a fan problem.

The other issue at hand is whether the tach output suffers with a lower voltage. Ideally the tach output is an open collector/open drain so the output should not be affected by the lower voltage. However the input to that BJT/MOSFET may now no longer meet threshold requirement to saturate, and the output stops working -- and the PC thinks the fan seized. Now depending on the hardware and settings, the hardware monitoring chip can then tell the PC it's time to do an emergency shutdown - it quickly dying usually is better than erratic operation causing unreliable output.

Anyway I think the threshold problem is happening with my reduced speed fan that doesn't register RPM anymore. This is most unfortunate as reduced speed fans are more likely to not spin up (due to overcoming initial inertia) and that tachometer data would be quite interesting to detect this situation.

(I was always wondering why AC plug-in fans that have multiple speeds always clicks on high first...well this inertia problem is why!)

[momaka]

Re: Zalman Fanmate-2 speed controller crashed my PC

Anyway I think the threshold problem is happening with my reduced speed fan that doesn't register RPM anymore. This is most unfortunate as reduced speed fans are more likely to not spin up (due to overcoming initial inertia) and that tachometer data would be quite interesting to detect this situation.

So just run the fan at a voltage that is not too high to make it loud and not too low so that it can't start.

I find that 7V is typically okay for most fans. Only the very low power (low RPM) ones tend to require more voltage - but those you can typically run at full speed / 12V anyways, and they won't be too loud.

However it can't be denied that the cooler the CPU is kept, the longer the CPU will last and less likely that the cpu will go unstable.

Well, you only need to keep the CPU cooled below a certain temperature if you want it to last, nothing more and nothing less. Typically, the danger threshold starts above 60°C. I personally prefer to keep my CPUs below 55°C (and if possible under 50°C, even better). Reason being is that while the thermal diode in the CPU may read one temperature, the CPU core has many layers, and some of those layers could be running at 5-10°C higher than what the thermal diode is reading (particularly, the bottom layers).

Take for example, AMD/ATI's Radeon HD4800 series of video cards - they have several thermal diodes in the cores: one for the "main" core, one for the memory handling part of the core, and one for the shader core. I suspect the memory and shader cores are in a lower layer further away from the heatsink, because they typically run 5-10C higher than the GPU's "main" temperature.

So for the above reason, I like to see my CPU and GPU temperatures below 55°C (under load, of course). But anything lower than that isn't necessary, and you will hardly see any change in lifetime with temperatures lower that that.

[eccerr0r]

Re: Zalman Fanmate-2 speed controller crashed my PC

Well, the problem isn't getting the fan started, it's getting the tach data so I know if it seizes for some other reason...
---

Actually from a [device] physics perspective, the cooler it is, the longer it will last. At 50C it will last longer than 60C. Granted it may already last quite a while at 60C, but then... you can overclock it...

The only layer gets hot is the diffusion layer as that's the most resistive layer. The metal layers won't get all that hot, fortunately the flip chips the diffusion layer is closest to the heatsink... The temperature sensor is on the same layer as the diffusions, however it definitely could be further in an XY direction on the die from the hotspots - but it won't be in the Z direction.

[stj]

Re: Zalman Fanmate-2 speed controller crashed my PC

the harddrive life is more effected than the electronics.

[eccerr0r]

Re: Zalman Fanmate-2 speed controller crashed my PC

The other "funny" aspect is that if someone implemented a true PWM controller for these tachometer fans, during the PWM dead time, the tachometer probably wouldn't work, again due to not being able to saturate the output transistor...

ahhh... screwed both ways, hence needing a 4-wire fan :-(

[redwire]

Re: Zalman Fanmate-2 speed controller crashed my PC

HDD's surely designed for some airflow, the power IC's get hot so I added the HDD cooling fan.
This old post says google data centers found no correlation with HDD lifetime and ambient temperature. But I think they all had lots of airflow, even if hotter air. Regardless, I like them cool

Intel 4-wire fan spec says "1.1 The expectation is a 4 wire PWM controlled fan when properly implemented will be significantly quieter than a similar 3 wire fan."

I'm using a 4-wire now and done with these cheeseball linear speed controllers.

[momaka]

Re: Zalman Fanmate-2 speed controller crashed my PC

HDD's surely designed for some airflow, the power IC's get hot so I added the HDD cooling fan.

Yeah, that's the only reason I like to have some airflow over my HDDs as well. Otherwise, I don't really care what temperature the internal platters run at, as long as it's not something crazy (i.e. over 50°C).

That said, for the older HDDs that don't use leaded solder, I don't bother to provide any cooling at all, as they won't have problems with Tin whisker growth and contacts becoming corroded.

The 20 GB IBM Travelstar in my laptop typically runs around 50°C in the summer. I try to turn off my laptop when I see it go beyond 53°C... but a few times I forgot and it hit 59°C. Still alright, though.

I'm using a 4-wire now and done with these cheeseball linear speed controllers.

Nothing wrong with linear controllers. It's just that not all 3-wire fans play nice with them. Some will start to click rather loudly loud when pushed under a certain speed.