Profiles can be edited by connecting a computer to the unit via USB and starting a termnial program. On my Linux laptop, I used:
minicom -b 115200 -P /dev/ttyUSB0
There are two types of configuration: global and profile.
Some settings are always used, no matter what profile is selected.
The ical command is used to calibrate the current measurements. It's
deault value is 133 milliOhms. By measuring the current with a multimeter,
you can calibrate this value to match your specific hardware:
ical 0.130
Would set the resistance to 130 milliOhms
With a battery connected, measure the voltage at the vcal test point and enter
it with the vcal command. For example:
vcal 25.4
Note there is a related vdrop parameter but this parameter is per-profile to
accommodate different breakout boards (e.g. whether they have protection diodes).
You can control how often to take voltage sag measurements.
First is vsag_interval_seconds which determines how often to take a baseline
measurement. For example:
vsag_interval_seconds 15
Next is vsag_settle_ms which determines how long to allow the battery to recover before taking a measurement. For example:
vsag_settle_ms 1000
Use fan_celsius <min_percent> <min_celsius> <max_celsius> to control any connected fan(s)
At min_celsius, the fan(s) will run at min_percent and smoothly ramp to
100% speed at and beyond max_celsius. Example:
fan_t 20 40 80
Use fan_watts <min_percent> <min_watts> <max_watts> to control fans by power output.
The settings for fan_t and fan_p are active at the same time. Whichever
leads to the higher fan speed at a given moment will be used.
Note that on may fans, using too low of a percent leads to no rotation.
Use finish_display <mah_ratio> to configure the number of seconds
that the finish stats are shown before the unit shuts down. Example:
finish_display 0.5
In the example above, if 1000 mAh were pulled from the battery then the display would be active for 1000 * 0.5 = 500 seconds. If 100 mAh were pulled, then the display would be active for 100 * 0.5 = 50 seconds.
The system looks to open up the FETs as much as possible while keeping under the current limits of:
- Max voltage sag
- Max current
- Max power
- Max temperature
Thus the goal is to reach one of these maximums while staying under the maximum for all others. Which maximum is reached will vary by battery, fan settings, room temperature, etc.
Coming up with and tuning an optimal PID algorithm that meets the above requirements would be daunting due to the large number of free variables. For example we would need to combine all four terms above to define an "error" as a starting point, then would still need to determine the PID constants.
Fortunately, we do not need to system to converge in minimum time. By relaxing this requirement, we can get away with less complex calculations.
The tuning algorithm is an annealing type algorithm where we use the terms "velocity" and "acceleration" as a mental model. As an analogy, imagine trying to adjust the volume knob on a friend's stereo. You would likely start with some gross adjustment until you overshoot the mark, then hunt with increasingly fine movements until you are satisfied. If a "loud" commercial comes on, you will need to adjust it again - again starting a bit fast and fine tuning.
With that in mind, here are the parameters of the algorithm:
- max_velocity The maximum (and starting) velocity in percent / second
- min_velocity The minumum allow change velocity in percent / second.
- deceleration_factor the amount to decrease the velocity on an overshoot (going past the intended mark): from 0 to max_velocity
- acceleration the amount to increase the velocity (not reaching the intended mark) in percent / second. from 0.0 to max_velocity
Example:
- fet_max_velocity = 30
- fet_min_velocity = 0.05
- fet_deceleration = 0.5
- fet_acceleration = 0.1
say the per sample time is 100ms
The PWM will scale from 0 percent to 100% at 3% per sample, taking 33 samples or 3.3 seconds to get there.
Say instead we hit a limit on the 10th sample. PWN at this point would be 3 * 10 = 30%. The velocity is reversed and decelerated, becoming to (-30 + 0.5) * 0.1 = -2.95% per sample
On the next sample, PWM will change to 30 - 2.95 = 27.05% Say that now everything is below threshold. This means that we may have backed off too far. Thus we reverse and decelerate again. The new velocity becomes (29.5 - 0.5) * 0.1 = 2.9% per sample.
One the next sample, PWM will change to 27.05 + 2.9 = 29.95. Say that we are still too low. In that case, the acceleration factor kicks in and the new velocity is not inverted, becoming (29 + 0.1) * 0.1 = 2.91% per sample.
and so on.
You can create a profile with the new command and a name. For example:
new
Profiles are named "New" as a starting point. Alternately you can duplicate an existing profile under a new name to use it's settings as a starting point:
duplicate 0
would take whatever profile is at index zero and copy it under the next available index.
Change the name of a profile with the name command. e.g.
name 0 "LIPO 3.8 percell"
Use the delete command with the profile index. e.g.
delete 1
Move a profile from one index to a different one via move <source_idx> <dest_idx>. e.g.
move 3 0
Use the list command to view profile names and indexes:
list
Use the show command to view settings for particular indexes or to dump all indexes:
show
show 0 2
show 1 2 3
Use vdrop <profile_index> <voltage> to set the voltage difference between the connected
battery and the vcal test point. This will take into consideration various FETs and
diodes that exist between the two points. Example:
vdrop 1 0.75
Use cell_count <profile_index> <count> to adjust the number of cells. If set at zero,
the number of cells will be automatically calculated at start by dividing the voltage
by the per_cell_target_volts (see below). If the automatic calcualtion would lead to incorrect
results in your situation, you'll need to fix the value.
Note that you can also use a value of 1 if you don't want to think in terms of cells
but only the full voltage. Examples:
cell_count 1 0
cell_count 1 1
cell_count 1 6
Sets the per-cell target voltage. When the target voltage is reached, the unit will shut off. For example:
per_cell_target_volts 1 3.8
Would set the target voltage to 3.8 if cell_count is 1, 7.6 if cell_count is 2,
etc.
per_cell_target_volts 1 0.0
Will run the drainer until it can no longer power itself, regardless of the cell count.
If the target voltage is set below the damage_voltage, the user will be told that
they are about to destroy the connected batteries and will ask for confirmation.
The default settings is for LIPO batteries and is set to 3.3 volts. Change it
with the damage_voltage command:
damage_voltage 1 3.0
Use max_amps <profile_index> <current> to set the maximum allowed current. Example
max_amps 1 10.0
Use per_cell_max_vsag <profile_index> <sag> to determine the maximum amount of
per-cell voltage sag that is allowed. For example:
per_cell_max_vsag 1 0.4
Would allow up to 0.4 * 4 = 1.6V of sag on a 4 cell pack.
Use max_celsius <profile_index> <temp> to change the maximum allowed heatsink termerature. Example:
max_celsius 1 80.0
Use max_watts <profile_indx> <watts> to change the maximum allowed wattage (voltage * current). Example:
max_watts 1 150
The console provides a fake_mv command. If used, the voltage sense logic will provide that value instead
of the actual voltage reading. It is useful for doing testing with no battery connected (since you can't
otherwise make it to the charging screen) Using this setting with a battery connected is not recommended.