|
@@ -29,6 +29,7 @@ message FromSmartKnob {
|
|
|
}
|
|
}
|
|
|
}
|
|
}
|
|
|
|
|
|
|
|
|
|
+/** Lets the host know that a ToSmartknob message was received and should not be retried. */
|
|
|
message Ack {
|
|
message Ack {
|
|
|
uint32 nonce = 1;
|
|
uint32 nonce = 1;
|
|
|
}
|
|
}
|
|
@@ -38,17 +39,70 @@ message Log {
|
|
|
}
|
|
}
|
|
|
|
|
|
|
|
message SmartKnobState {
|
|
message SmartKnobState {
|
|
|
|
|
+ /** Current integer position of the knob. (Detent resolution is at integer positions) */
|
|
|
int32 current_position = 1;
|
|
int32 current_position = 1;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Current fractional position. Typically will only range from (-snap_point, snap_point)
|
|
|
|
|
+ * since further rotation will result in the integer position changing, but may exceed
|
|
|
|
|
+ * those values if snap_point_bias is non-zero, or if the knob is at a bound. When the
|
|
|
|
|
+ * knob is at a bound, this value can grow endlessly as the knob is rotated further past
|
|
|
|
|
+ * the bound.
|
|
|
|
|
+ *
|
|
|
|
|
+ * When visualizing sub_position_unit, you will likely want to apply a rubber-band easing
|
|
|
|
|
+ * function past the bounds; a sublinear relationship will help suggest that a bound has
|
|
|
|
|
+ * been reached.
|
|
|
|
|
+ */
|
|
|
float sub_position_unit = 2;
|
|
float sub_position_unit = 2;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Current SmartKnobConfig in effect at the time of this State snapshot.
|
|
|
|
|
+ *
|
|
|
|
|
+ * Beware that this config contains position and sub_position_unit values, not to be
|
|
|
|
|
+ * confused with the top level current_position and sub_position_unit values in this State
|
|
|
|
|
+ * message. The position values in the embedded config message will almost never be useful
|
|
|
|
|
+ * to you; you probably want to be reading the top level values from the State message.
|
|
|
|
|
+ */
|
|
|
SmartKnobConfig config = 3;
|
|
SmartKnobConfig config = 3;
|
|
|
|
|
|
|
|
- /** Value that changes each time the knob is pressed */
|
|
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Value that changes each time the knob is pressed. Does not change when a press is released.
|
|
|
|
|
+ *
|
|
|
|
|
+ * Why this press state a "nonce" rather than a simple boolean representing the current
|
|
|
|
|
+ * "pressed" state? It makes the protocol more robust to dropped/lost State messages; if
|
|
|
|
|
+ * the knob was pressed/released quickly and State messages happened to be dropped during
|
|
|
|
|
+ * that time, the press would be completely lost. Using a nonce allows the host to recognize
|
|
|
|
|
+ * that a press has taken place at some point even if the State was lost during the press
|
|
|
|
|
+ * itself. Is this overkill? Probably, let's revisit in future protocol versions.
|
|
|
|
|
+ */
|
|
|
uint32 press_nonce = 4 [(nanopb).int_size = IS_8];
|
|
uint32 press_nonce = 4 [(nanopb).int_size = IS_8];
|
|
|
}
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
message SmartKnobConfig {
|
|
message SmartKnobConfig {
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Set the integer position.
|
|
|
|
|
+ *
|
|
|
|
|
+ * Note: in order to make SmartKnobConfig apply idempotently, the current position
|
|
|
|
|
+ * will only be set to this value when it changes compared to a previous config (and
|
|
|
|
|
+ * NOT compared to the current state!). So by default, if you send a config position
|
|
|
|
|
+ * of 5 and the current position is 3, the position may remain at 3 if the config
|
|
|
|
|
+ * change to 5 was previously handled. If you need to force a position update, see
|
|
|
|
|
+ * position_nonce.
|
|
|
|
|
+ */
|
|
|
int32 position = 1;
|
|
int32 position = 1;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Set the fractional position. Typical range: (-snap_point, snap_point).
|
|
|
|
|
+ *
|
|
|
|
|
+ * Actual range is technically unbounded, but in practice this value will be compared
|
|
|
|
|
+ * against snap_point on the next control loop, so any value beyond the snap_point will
|
|
|
|
|
+ * generally result in an integer position change (unless position is already at a
|
|
|
|
|
+ * limit).
|
|
|
|
|
+ *
|
|
|
|
|
+ * Note: idempotency implications noted in the documentation for `position` apply here
|
|
|
|
|
+ * as well
|
|
|
|
|
+ */
|
|
|
float sub_position_unit = 2;
|
|
float sub_position_unit = 2;
|
|
|
|
|
|
|
|
/**
|
|
/**
|
|
@@ -60,14 +114,83 @@ message SmartKnobConfig {
|
|
|
*/
|
|
*/
|
|
|
uint32 position_nonce = 3 [(nanopb).int_size = IS_8];
|
|
uint32 position_nonce = 3 [(nanopb).int_size = IS_8];
|
|
|
|
|
|
|
|
|
|
+ /** Minimum position allowed. */
|
|
|
int32 min_position = 4;
|
|
int32 min_position = 4;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Maximum position allowed.
|
|
|
|
|
+ *
|
|
|
|
|
+ * If this is the same as min_position, there will only be one allowed position.
|
|
|
|
|
+ *
|
|
|
|
|
+ * If this is less than min_position, bounds will be disabled.
|
|
|
|
|
+ */
|
|
|
int32 max_position = 5;
|
|
int32 max_position = 5;
|
|
|
|
|
+
|
|
|
|
|
+ /** The angular "width" of each position/detent, in radians. */
|
|
|
float position_width_radians = 6;
|
|
float position_width_radians = 6;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Strength of detents to apply. Typical range: [0, 1].
|
|
|
|
|
+ *
|
|
|
|
|
+ * A value of 0 disables detents.
|
|
|
|
|
+ *
|
|
|
|
|
+ * Values greater than 1 are not recommended and may lead to unstable behavior.
|
|
|
|
|
+ */
|
|
|
float detent_strength_unit = 7;
|
|
float detent_strength_unit = 7;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Strength of endstop torque to apply at min/max bounds. Typical range: [0, 1].
|
|
|
|
|
+ *
|
|
|
|
|
+ * A value of 0 disables endstop torque, but does not make position unbounded, meaning
|
|
|
|
|
+ * the knob will not try to return to the valid region. For unbounded rotation, use
|
|
|
|
|
+ * min_position and max_position.
|
|
|
|
|
+ *
|
|
|
|
|
+ * Values greater than 1 are not recommended and may lead to unstable behavior.
|
|
|
|
|
+ */
|
|
|
float endstop_strength_unit = 8;
|
|
float endstop_strength_unit = 8;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Fractional (sub-position) threshold where the position will increment/decrement.
|
|
|
|
|
+ * Typical range: (0.5, 1.5).
|
|
|
|
|
+ *
|
|
|
|
|
+ * This defines how hysteresis is applied to positions, which is why values >
|
|
|
|
|
+ */
|
|
|
float snap_point = 9;
|
|
float snap_point = 9;
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Arbitrary 50-byte string representing this "config". This can be used to identify major
|
|
|
|
|
+ * config/mode changes. The value will be echoed back to the host via a future State's
|
|
|
|
|
+ * embedded config field so the host can use this value to determine the mode that was
|
|
|
|
|
+ * in effect at the time of the State snapshot instead of having to infer it from the
|
|
|
|
|
+ * other config fields.
|
|
|
|
|
+ */
|
|
|
string text = 10 [(nanopb).max_length = 50];
|
|
string text = 10 [(nanopb).max_length = 50];
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * For a "magnetic" detent mode - where not all positions should have detents - this
|
|
|
|
|
+ * specifies which positions (up to 5) have detents enabled. The knob will feel like it
|
|
|
|
|
+ * is "magnetically" attracted to those positions, and will rotate smoothy past all
|
|
|
|
|
+ * other positions.
|
|
|
|
|
+ *
|
|
|
|
|
+ * If you want to have more than 5 magnetic detent positions, you will need to dynamically
|
|
|
|
|
+ * update this list as the knob is rotated. A recommended approach is to always send the
|
|
|
|
|
+ * _nearest_ 5 detent positions, and send a new Config message whenever the list of
|
|
|
|
|
+ * positions nearest the current position (as reported via State messages) changes.
|
|
|
|
|
+ *
|
|
|
|
|
+ * This approach enables effectively unbounded detent positions while keeping Config
|
|
|
|
|
+ * bounded in size, and is resilient against tightly-packed detents with fast rotation
|
|
|
|
|
+ * since multiple detent positions can be sent in advance; a full round-trip Config-State
|
|
|
|
|
+ * isn't needed between each detent in order to keep up.
|
|
|
|
|
+ */
|
|
|
repeated int32 detent_positions = 11 [(nanopb).max_count = 5];
|
|
repeated int32 detent_positions = 11 [(nanopb).max_count = 5];
|
|
|
|
|
+
|
|
|
|
|
+ /**
|
|
|
|
|
+ * Advanced feature for shifting the defined snap_point away from the center (position 0)
|
|
|
|
|
+ * for implementing asymmetric detents. Typical value: 0 (symmetric detent force).
|
|
|
|
|
+ *
|
|
|
|
|
+ * This can be used to create detents that will hold the position when carefully released,
|
|
|
|
|
+ * but can be easily disturbed to return "home" towards position 0.
|
|
|
|
|
+ */
|
|
|
float snap_point_bias = 12;
|
|
float snap_point_bias = 12;
|
|
|
|
|
|
|
|
/**
|
|
/**
|