Apache Labs Community Forum

Hello Gentlemen,

I am exploring options to have some sort of meaningful fan operation for the Anan G2 that improves noise without compromising operation. It looks like the stock fan in use is a 5VDC and is constantly connected to 5V which spins it at maximum RPM all the time, which is annoyingly noisy, in fact much louder than my legal limit LDMOS amp

Since I noticed J53 on the Saturn board is a 4-pin connector and out of curiosity I measured the pinout to find:

P1: GND
P2: 5V
P3: 3.3V
P4: 0.01V

I also measured with an oscilloscope but could not find any PWM signal on P3 which I initially, based on the voltage, thought looked like 66% duty cycle PWM. Neither I could find anything on P4.

All that being said, is there anyone out there who further explored the fan situation, got a hold of schematics and performed experimentation to improve it? I am particularly curious about PWM.

P.S. For those who will recommend replacing it with Noctua, I understand that it is an option but bear in mind the 5V available Noctua fans will spin at max RPM and they are still loud at max RPM.

~Cheers~

There is no PWM or fan control circuitry on any Apache designs except for a certain run of Orion MKII boards. On those boards the fan control circuit had a design flaw that caused more noise instead of less

If you want a variable speed fan you will have to DIY a solution.

That connector connects to the Raspberry pi half of the unit, and there is the possibility of a device driver on the pi controlling a fan. As far as I know it has never been implemented. I imagine the idea is that the device driver looks at the Raspberry pi die temperature, and adjusts to keep that constant. It will know nothing of the Saturn FPGA temperature (which is available electronically) and the PA heatsink temperature (which isn't).

So Scott's answer is correct - DIY solution needed.

Remind me. Was the G2 designed so that the fan continues to run after the G2 is turned off? Currently the external power supply must be turned off. Why?
Is there a simple explanation?

Thanks,
73 Torrey N9PY

Mine doesn't!

If you shut down the processor from the desktop, then yes it continues to run. If you press and hold the power button for 2 seconds or so, then the fan cuts off. Or mine does.

w-u-2-o wrote:There is no PWM or fan control circuitry on any Apache designs except for a certain run of Orion MKII boards. On those boards the fan control circuit had a design flaw that caused more noise instead of less

If you want a variable speed fan you will have to DIY a solution.

That is a bit disappointing. A friend of mine mentioned there was issues with PWM on certain boards and it was removed after. I suppose that's the Orion II boards you are mentioning.

laurencebarker wrote:That connector connects to the Raspberry pi half of the unit, and there is the possibility of a device driver on the pi controlling a fan. As far as I know it has never been implemented. I imagine the idea is that the device driver looks at the Raspberry pi die temperature, and adjusts to keep that constant. It will know nothing of the Saturn FPGA temperature (which is available electronically) and the PA heatsink temperature (which isn't).

So Scott's answer is correct - DIY solution needed.

That's really unfortunate. It is definitely not ideal to only look at the Pi die temp for fan control. How exactly is FPGA temperature available electronically maybe there is a way for it to be fed back to the Pi?

See also this thread: viewtopic.php?f=17&t=2682

w-u-2-o wrote:There is no PWM or fan control circuitry on any Apache designs except for a certain run of Orion MKII boards. On those boards the fan control circuit had a design flaw that caused more noise instead of less

If you want a variable speed fan you will have to DIY a solution.

Scott, just to confirm if I am not misunderstanding. Raspberry Pi compute module 4 IO Board for example has a built in EMC2301. If I understand correctly, you are saying there was similar controller available at some point but was taken out because it was causing noise?

laurencebarker wrote:That connector connects to the Raspberry pi half of the unit, and there is the possibility of a device driver on the pi controlling a fan. As far as I know it has never been implemented. I imagine the idea is that the device driver looks at the Raspberry pi die temperature, and adjusts to keep that constant. It will know nothing of the Saturn FPGA temperature (which is available electronically) and the PA heatsink temperature (which isn't).
So Scott's answer is correct - DIY solution needed.

Laurance, can you elaborate how exactly does the fan connector connect to the Pi? I am still trying to wrap my head around the problem to figure out if there is any viable solution.

Thank you, gentlemen, for the input on this.

~Cheers~

For a DIY solution, Paul, W1GHZ might have an option.

http://www.w1ghz.org/PCBproj/Fan_controller.pdf

Mike. N1JEZ

F128 wrote:Scott, just to confirm if I am not misunderstanding. Raspberry Pi compute module 4 IO Board for example has a built in EMC2301. If I understand correctly, you are saying there was similar controller available at some point but was taken out because it was causing noise?

Yes, that's correct. A previously posted, refer you to this thread where both the circuit and its design flaws are described: viewtopic.php?f=17&t=2682[/quote]

laurencebarker wrote:That connector connects to the Raspberry pi half of the unit, and there is the possibility of a device driver on the pi controlling a fan. As far as I know it has never been implemented. I imagine the idea is that the device driver looks at the Raspberry pi die temperature, and adjusts to keep that constant. It will know nothing of the Saturn FPGA temperature (which is available electronically) and the PA heatsink temperature (which isn't).
So Scott's answer is correct - DIY solution needed.

F128 wrote:Laurance, can you elaborate how exactly does the fan connector connect to the Pi? I am still trying to wrap my head around the problem to figure out if there is any viable solution.

The Pi fan controller ciruit is copied from the Pi Compute Module IO Board, schematics can be found at that link. It operates under the control of the Pi OS. The Pi OS does know the temperature of the Pi CPU but that is no help because the design driver is the temperature of the RF power amplifier devices.

With the required changes to the FPGA firmware and Thetis software, the OS could be made to know the temperature of the Saturn FPGA, but again that is no help for the same reason, which is that it's the temperature of the RF power amp devices that matter.

A temperature sensor could be provisioned on the RF board that connects to the Pi. There are a million web articles about how to do this for remote sensing of temperature with the Pi. Indeed, there could be several temp. sensors if you like. Then an OS script could be written that would control a fan connected to the Pi fan connector.

That said, a completely independent approach is probably easiest. See my post below.

Mike_N1JEZ wrote:For a DIY solution, Paul, W1GHZ might have an option.

http://www.w1ghz.org/PCBproj/Fan_controller.pdf

Mike. N1JEZ

You can buy a much more capable solution for less than you can build that one. For instance:

https://www.amazon.com/pwm-temperature- ... controller

Buy whichever board floats your boat, add a nice Noctua fan, adjust to taste, and you are done.

I can confirm that the PWM controller is there because it is in the Raspberry pi reference designs - not because it was intended for this application. If we hadn't included it we'd have to remove the device driver, making it harder to use the Raspberry pi operating system builds. It isn't suitable for controlling the temperature of the PA heatsink. It would be theoretically possible to make it read FPGA die temperature - it is a simple register read to get it, and it is in the documentation - but it wouldn't help and you'd need to rewrite the linux device driver.

laurencebarker wrote:I can confirm that the PWM controller is there because it is in the Raspberry pi reference designs - not because it was intended for this application. If we hadn't included it we'd have to remove the device driver, making it harder to use the Raspberry pi operating system builds. It isn't suitable for controlling the temperature of the PA heatsink. It would be theoretically possible to make it read FPGA die temperature - it is a simple register read to get it, and it is in the documentation - but it wouldn't help and you'd need to rewrite the linux device driver.

You might be looking at this the wrong way, Laurence. It's not about using the Pi OS features as designed, it's about exploiting the hardware that is already there.

It would be completely trivial to purchase a thermocouple or sensor board (aka "hat") designed for the Pi GPIO interface, then write a shell script or some simple Python code to operate the on-board fan controller in whatever manner was desired. There's even example code out there, so it would be little more than cut & paste coding.

Here's an example of a very capable board that is not expensive at only $40, plus sensors:

https://pi-plates.com/thermoplate/

I'm not familiar with the insides of a G2, a ribbon cable might be necessary to attach to Saturn board connector U60 (an odd name for a connector) which is the 40 pin Pi IO port.

I replaced the noisy fan with a Noctua one and it is much quieter even though it runs all the time.

NF-A12x25 5V

F128 wrote:Hello Gentlemen,

I am exploring options to have some sort of meaningful fan operation for the Anan G2 that improves noise without compromising operation. It looks like the stock fan in use is a 5VDC and is constantly connected to 5V which spins it at maximum RPM all the time, which is annoyingly noisy, in fact much louder than my legal limit LDMOS amp

Since I noticed J53 on the Saturn board is a 4-pin connector and out of curiosity I measured the pinout to find:

P1: GND
P2: 5V
P3: 3.3V
P4: 0.01V

I also measured with an oscilloscope but could not find any PWM signal on P3 which I initially, based on the voltage, thought looked like 66% duty cycle PWM. Neither I could find anything on P4.

All that being said, is there anyone out there who further explored the fan situation, got a hold of schematics and performed experimentation to improve it? I am particularly curious about PWM.

P.S. For those who will recommend replacing it with Noctua, I understand that it is an option but bear in mind the 5V available Noctua fans will spin at max RPM and they are still loud at max RPM.

~Cheers~

Hi,

I installed the Noctua NF-A12x25 5V PWM fan in my G2 Ultra and connected it with the 4-pin connector. As you wrote, this did not reduce the volume. Since I could not find any other solution, I throttled the 5V to 3.6V using a step-down voltage regulator module (link: https://www.amazon.de/dp/B0C4YY4368?ref ... asin_title). Now the volume is significantly lower.

Mathias, DD5MD

Hello,

which one is more suitable:

NF-A12x25 5V OR NF-A12x25 5V PVM?

73

Volker
DL4ZBG

dl4zbg wrote:which one is more suitable: NF-A12x25 5V OR NF-A12x25 5V PVM?

Only difference is 3 or 4 cable pin as shown here
https://faqs.noctua.at/en/support/solut ... 1000081757

As you can not control the fan speed the non-PWM is just good.

73 Ronald PH7R