mit-drl / e-vent Goto Github PK

Make a display to show pressure at all times

We need to speed up loop frequencies to at least 100Hz.

• Be smarter about not doing UI stuff so often
• Make slow things more efficient if possible
• Wait for timers when in the last loop before a trigger, to more accurately slice loop times

hi i am facing issue related to motor control after homing motor start moving in one direction and it'speed is more the expected speed

There's currently a "sanity check" in the setPosition function. It's not clear that this is needed, or that it couldn't be triggered by noise. Furthermore, it's not clear that the way it's handled (break) is the proper fault response.

We should look into robustifying this by clearly defining the error condition we need to detect, and ensuring the response is appropriate (perhaps alarm?)

The sensitivity of the volume knob makes it request for confirmation even without being touched.

• Backwards encoder wiring
• Backward motor wiring
• Disconnected pressure sensor
• No roboclaw connected

After receiving feedback from clinicians, we have settled on a new display format that will allow us to incorporate more information in a more compact layout.

See below for images showing the new functionality. The key feedback we received includes:

• Show the status/alarm on the top (A) and the icons / setting change on the bottom (C)
• The icons for Pip/Plat/Peep using waveforms were understood and preferred to arrows
• the 1:x should be shown for IE ratio, and also included in the I:E (decided to compress it into a single glyph for space)
• I:E ratio should be allowed to go as low as 1:0.4
• The time-triggered vs patient-triggered icons were understood, and should be placed above the pressures (showin below in top-right corner)
• AC knob should be labeled "Ptr"
• Alarm sound icon should only show when an alarm is sounding (B), otherwise, it should be blank (A)
• When an alarm is occurring but muted, the sound icon should indicate this, and also count down seconds until muting will stop (C).

A: Normal Running

B: Alarming

C: Change Settings

Account for constant acceleration by choosing max velocities that give the correct average velocity given the trapezoidal velocity profile we execute.

If I understand correctly, there are 3 coefficients that determine the particular profile for different ambu bags but I could use a better explanations for each number so I can make things work with my ambu bag. These coefficients are in constants.h

// Bag Calibration for AMBU Adult bag
const struct {float a, b, c;} COEFFS{1.29083271e-03, 4.72985182e-01, -7.35403067e+01};

The on/off button turns off the system, but does not restart it; instead the confirm button has this function.

Check for driver disconnection. Recover gracefully.

Look for issues hitting/leaving limit switch and alarm. Consider

• Too high current
• Too many encoder counts
• Too much time

void Display::writeBPM(const int& bpm) {
const int bpm_c = constrain(bpm, 0, 99);
char buff[12];
sprintf(buff, "%2s=%2s ", getLabel(BPM).c_str(), toString(VOLUME, bpm_c).c_str());
write(elements_[BPM].row, elements_[BPM].col, buff);
}

here, the sprintf line is wrong, it should be:

sprintf(buff, "%2s=%2s ", getLabel(BPM).c_str(), toString(BPM, bpm_c).c_str());

We should include some level of software and hardware verification where we test for example:

• software unit testing
• all required hardware is plugged in
• perhaps one cycle of actuation with expected values measured to close the loop
• any failing tests triggers a visible alarm, only all passing tests triggers a visible "green light"

Try to shorten event.ino .
Reorganize and refactor.
Try to move standalone functions to other files.
Add documentation / comments as needed. // for example like this.

Useful for debugging and automated testing.

• Check that it reliably works
• Doesn't break when weird stuff gets sent in

On 1.0-003, pressing and holding the OFF button does not trigger a turn off.
This could be hardware or software.

Show an icon to indicate if the alarms are snoozed

• Remove static text to make more space
• Not hide critical info during alarms
• Add icons for snooze/ac control

Feature request to clear all alarms after holding the alarm mute button for 2 seconds

Machine suddenly stops for no apparent reason.
Observed on 1.0-003 while connected via USB to Ubuntu laptop.

Currently, there's no way to see what the value is for the pots. We could use manually printed labels near the pot knobs, but perhaps using the LED display to show updated pot values whenever they are adjusted would be a more robust way to ensure the user knows what value is being set.

Currently, the motor position is controlled through a handcrafted trapezoidal velocity curve to ensure accelerations are within limit.

However, this may not be the most appropriate, especially considering we may need to set the position at a higher frequency.

• Perhaps a PI or PID controller over acceleration or velocity would be more advisable
• Alternatively, perhaps a position control servo or motor controller with position control and max_velocity control would be more robust

We should do some testing / research

Do a better job measuring plateau instead of simply taking the last pressure value before exhale.

After a few minutes of run time my LCD screen starts to display garbage. Is anybody else experiencing this? The bug occurs after homing is complete. I enter the EX_STATE and then since I am just emulating pressure feedback manually on my desk several alarms occur. After a few minutes the screen displays nonsense.