Server chronicles - testing fan proposal

The HP ProLiant DL360p Gen8 has 8 fans to cool down the two CPUs and other hardware.

These fans are high speed PWM controlled and are prone to fail over time.

The only thing to do is to replace the defective fan with a new one, but it would be nice to understand exactly what the problem is and check if the replacement fan is working before to open the server.

Flipper Zero - setting up a development environment

Flipper Zero is defined as "hacking" tool, i.e. is a collection of HW and SW that allows to explore and interact with technology around us.

How can we work on it ?

Video tests with Raspberry Pi

Tests to check quality and feasibility to use a Raspberry Pi to track objects in movement.
In order to create an object follower, is important to see if the quality of the video is good enough.

5V Solar Power Supply - testing results

As described in the 5V Solar Power Supply - testing system article, here some results after running tests for a while.

First of all, I did connected the data logger to the battery terminal and let it run for few days, with the load connected, drawing around 350mA.
Here a graph, starting with the battery charged.
Once the battery was depleted, with the load connected there is no way to have it recharged enough.
The weather was good at least for the last two days, so plenty of sun, but even with the 3W solar panel the battery was unable to be charged enough.

This indicates anyway the need to disconnect the DC/DC converter in order to have the battery charged.

5V Solar Power Supply - testing system

In order to test the solar power supply without connecting the Raspberry (also because used to monitor the solar charger itself), here how I'm proceding.

5V Solar Power Supply for Raspberry Pi - first results

The introductory article was about describing the feasibility of the project.
Let's see some preliminary results.

The measurements

Until now I did run the system with an external voltmeter to see if the solar panel was able to keep up with the system.
It seems that so far the answer is no.

The graph shows the battery level measured in few days.
Note that basically there was no real load to the battery, only the Dc/Dc converter, so very few mA were drawn from the battery.
The days were pretty cloudy all the time, so no full sun for long time. In the best scenario (see the latest days) the battery level barely reached the 2.9V, well well below the minimum of 3.5V, and again I stress the fact there was no real load on the system.

The solar panel used so far is a 6V 2W, capable to provide in full sun a current of 330mA.
Definitively not enough to keep the battery charged AND powering something else (the Raspberry Pi has an estimated drawing current around 300-350mA).

Until there was sun the system did run happily, but as soon as the solar panel was not in full sun, the system had to use more and more the battery to power up the project and of course there was no battery charging there.

The battery is a nominal 3.7V 6600mAh and can last just few hours when not charged, so the system starting with a loaded battery in a sunny day can run almost all the day, but that only the first day.

What to do ?

There are few things to try.

1. Use a bigger and powerful solar panel.
   A bigger solar panel, still 6V but maybe 6 or 9 W, probably can give more energy even with less light.
   In full sun the power should be enough to charge the battery AND power the system.
   Adafruit has a 6V 5.6 W solar panel, capable to give up to 930 mA in full sun and presumably more juice even in less than ideal conditions.
    
2. Implement an intelligent power management
   We can assume to don't have the system running 24 hours a day.
   An external circuit (like the RPOf) can monitor the charge of the battery and shutdown the Raspberry when the battery level is below a specific threshold.
   When the battery threshold level is above a specific value the circuit can power on back the Raspberry.
3. Reduce as much as possible the Raspberry Pi current needs.
   It can be done disabling every thing not needed, like the HDMI port or the LEDs.
   Even better adopting a Raspberry Pi 3, eliminating also the need to power the USB ports (currently the USB ports are in use for the WiFi dongle).

Very probably all the solutions will be needed.

Sensor modification

To better handle the solar management, a modification is needed on the light sensor.
Currently the light sensor is based on a photoresistor.
It can indicate if there is light or dark but is not enough to determine the "quality" of the light.
In other words the value returned is not enough to determine if the light is enough to generate ppower via the solar panel or not.
To do so a better light sensor is needed.  So in the near future I'll use a solar light sensor rather than the generic light sensor.

Airflow sensor test

I used a little board with a MSP430-2013 and a LCD display  to develop the basic code to test and read the airflow sensor.

The circuit

Here a picture of the prototype

The software

The airflow sensor reading is based on counter and a timer.
The MSP430 pin connected to the airflow sensor is set in interrupt mode (rising) and after a check on the pin state to be sure is not a spike, a counter is incremented.
Every time a related timer expire, the value of the counter is copied into another variable (to be displayed) and the counter is reset.
I use another timer to display the read value on the display, after a integer-to-ascii conversion.
The code described is not reading the DS1820, only the airflow sensor.

Roomba 4xx test batteries - manual

Having a lot of Roomba batteries for the series 4xx (third and fourth generation), before to trash them in the recycle kit I just received,  I needed a way to see if a battery is good or bad.

To test a battery there are some steps to follow :

• measure the voltage of the battery - if below 15V the battery can have problems
• discharge the battery
• charge the battery
• discharge the battery on a calculated load - it should last a fixed amount of time depending the type of the battery. At least 1 hour for the older NiCd

Using the external iRobot charger modified (for manual), I did set up this environment.

The dis-charger is a 12V lamp.
I'm experimenting with a 12V 50W and a 12V 34W lamps.

Building

Building the contraption is easy.
The only trick is to pay attention to the polarity of the wires.
It would be great to use red/black cables, but as usual I use whatever I have around in my junk box.

 Here I'm using a 12V 50W lamp, but as soon as I find a socket for the other lamp, I'll try to use the 12V 34W lamp.
The ideal it would be to use a load that simulates the Roomba load. People doing this before me estimated that a 12V 25W lamp is a perfect load.
One of these days I'll try in different places to find the lamp, for now this is what I have around.

For mechanical reason, I put the switch (Switch A) in plastic box (of course re-used many many times for different projects).

The switch is actually a 3 position switch, so the label indicates ch (Charger), off (Off :) ) and dis (Dis-charger).
Because of that, there is really not need of the Switch B, since is enough to put the switch A in the Off position to measure the battery voltage with no load.
But for "logical" reasons, the Switch B still exists.
Let say that "Switch A in Off position",  is equivalent to say "Switch B open".

Here the contraption connected to the modified external charger.

Operations

In order to see if a battery is OK or not, here Step by Step list of things to do.

1. set the Switch A on Off (or set  the Switch A  to connect the battery to the dis-charger and set the Switch B of the dis-charger  open).
   In this way is possible to measure the voltage of the battery without load.
2. put the battery on the battery holder of the iRobot external charger.
   
   Low reading at this stage can already indicate if a battery is dead.
   A healthy battery should give a reading around at least 14.4V.
3. set the switch A on dis (Dis-charger) or set the Switch B On.
   Simply keep connect the battery to the dis-charger until the lamp is on.
   When the lamp goes off :
   - if lasted at least an hour for the NiCD and two hours for the NiMH the battery is OK (assuming a load of 12V 25W, less for bigger loads)
   - if last less the battery is not good anymore, however at least one charge cycle must be performed before to decide the battery is no good.
   Go to the point 4 if the battery needs to do at least a charge cycle
4. toggle the Switch A to connect the battery to the charger.
   If the battery has a chance to be OK, the battery charger is engaged (green LED ON).
   The charge operation can last up to three hours, depending the type of battery.
   - If the green LED flash, the battery must be dead. Discard it.
   - If the green LED goes OFF in few minutes, the battery could be dead.
     Check if the battery is warm. If not the charger refused to charge it. The battery is dead.
   - If the green LED goes OFF after at least 1 hour (black battery) or 2 hours or more (yellow battery), go back to the point 1