Comment by jrockway | Hacker News Reader

Comment by jrockway | Hacker News Reader

Parent Full thread

Interesting to see first party support for this. It is somewhat unfortunate that the official fan lets you read the fan speed, but not set it. I put a Noctua fan on my Pi and it sucks in a surprising amount of dust; mostly an artifact of too much ventilation when the computer doesn't actually need the airflow. (I put a filter on it and it's fine now, but sometimes I wonder whether the dust is actually filtered out, or if it just reduces airflow by so much that you don't suck in any dust to begin with :)

While writing this comment, I decided to investigate whether or not speed control is even possible on a 5V fan with 3V3 logic and found this: https://noctua.at/pub/media/wysiwyg/Noctua_PWM_specification...

The answer is: I still don't know. They say that the PWM voltage is pulled up to "3.3V/5V" internally but the spec says the PWM signal has to be 5V, so I suppose the results are undefined (and they just left 3.3V in their document accidentally).

But, a first-party fan could make this work, so I'm kind of sad that they just chose some off-the-shelf fan that was cheap today in Shenzhen. If you could do better than Noctua, then there's value in making it first-party. If you can't, why bother? Everyone that wanted a fan already has one.

I may have read your post the wrong way, but it seems that they did provide a way? The specifications on the product page [0] mentions it got speed control with PWM, I figured up to the Intel specs, to be able to use an off-the-shelf fan.

So if anything you should be able to connect the PWM pin of your Noctua fan to one of the GPIO pins, and use their software to specify the PIN number of your PWM output. What seems more complex is to get the input from the fan about the actual speed, which would be useful to detect that the fan is blocked and is not doing the job you ask it to do.

[0]: https://www.raspberrypi.org/products/raspberry-pi-4-case-fan...

No, I must have glossed over that. I saw 3 pins, and assumed that it was power/ground/speed output like other 3-pin fans.

IMO I would interpret "External pull-up is not necessary as the signal is pulled up to 3,3V/5V inside the fan." as indicative of expecting to being driven by an open drain output. There also is no spec for "Minimum voltage for logic high". You should be able to test this by hooking the fan up to power, measuring the voltage on its disconnected PWM line, and finding a steady voltage (5v?).

I don't know enough about the RPi to know if any PWMs can be configured as open drain and 5v tolerant. If not you'd need to add a single discrete transistor.

I used a small Noctua fan on my Pi project (actually a relatively large box) that I'm controlling via PWM. Didn't need any logic level conversion.

>I put a Noctua fan on my Pi and it sucks in a surprising amount of dust; mostly an artifact of too much ventilation when the computer doesn't actually need the airflow.

I'm not sure what "too much ventilation when the computer doesn't actually need it" mean.

Either there's dust in the air/surfaces near the intake or not.

It needs active cooling for about 10 minutes out of every month. If the fan runs for the other 43190 minutes in the month, then that's dust being sucked into the computer for no real reason, and thus just makes a mess without providing any value in return.

Which cost more, the Pi or the fan?

The 3D printer to print the case with a fan hole.

This is why my pi's case is wooden.

Reply on news.ycombinator.com