开云体育

Erich Shulz mentioned in message #2131 that pairing an arduino with a laptop could improve UI performance. I think laptops, or expanded computing options in general is worth further consideration for future models so I decided to move it to its own topic. To start with, I want to say this is a discussion for future design, not a critique of the current design. I believe the current arduino based system is the best answer to the original design goal of a rapidly designed, rapidly built, and rapidly deployed system which was inexpensive and didn't impinge on the already strained medical supply chain. Improved processing power is something I've been thinking about from the beginning. The current design with the arduino has many limitations, however, it does what it was designed to do. From what I've seen, our device provides the best balance of performance vs simplicity and cost. I don't know of any other system that provides nearly the same functionality while still being build-able from home. Though as others have suggested, there's no reason to settle on one system, there is no one size fits all. For the sake of discussion, I'll outline my thoughts on the subject here Erich, I agree that using a laptop would provide vastly more computing power, but would be a much more expensive and complex system. It also begs the question, why bother with an arduino at all? If you have a laptop connected, we just run everything off that? Another possibility is to include the code for laptop interface in the arudio and let the user decide whether or not to use it. It's something to investigate as the software evolves. One concern I have up front is the fact that connecting or disconnecting a serial device (such as a laptop) automatically reboots the arduino which could be a problem if the laptop is ever disconnected or fell asleep etc. Another possibility that has been brought up here and with other groups is a raspberry pi (RPi), either alone or pared with an Arduino. The Pi Zero can be obtained for around $5 US and has virtually all the capabilities of a laptop including keyboard and mouse support, usb input/output, wifi, bluetooth, HDMI output, and well designed and documented touch screen comparability. I have a personal remote data cloud and ad blocking server running on a Pi zero in my home, they're amazing little devices. And they run on 5v just like the Arduino. However, the programming is a bit more complex, and since it's an actual computer running linux there are many background and underlying processes that need to be taken into account as they could lead to instability. The programming considerations are greater than simple arduino/C++ coding. Additionally, since it's running an OS, it must be shut down properly to avoid system corruption. That being said, there are small relatively cheep hardware systems that can detect loss of mains power and then use a battery to keep the Pi running long enough to properly shut down. This could be a standalone system, or tied into the battery backup already implemented in the current vent design. There were also questions early on about RPi's availability and supply chain. I think for these reasons and others, it made sense to avoid RPi's for the first round, but I think they're a great candidate for V2. I'm also curious what impact a paired computer be it laptop or RPi would have on the arduino speed. Moving the heavy processing off the arduino would obviously help, but on the other hand, serial communication is a bit slow and becomes blocking if the buffer is overloaded. So the process of moving data off the arduino might cause more delays than it fixes. Another option is the use of a more powerful and/or dual core project board. For example ESP-32 boards are about the same price as arduino nanos but include two cores, each far more powerful than the nano's atmega280. They can be programmed with arduino IDE and use very similar code functions. It could be possible to dedicate one core to vent functions and the other to display/io functions. However, dual core programming adds a whole new level of complexity in design, and might make it less accessible as fewer users will be able to understand/modify the code if needed. These are just my thoughts as a hobbyist, I'm curious to hear what others think. Particularly professional programmers. Steve S

To fill everyone in on the crucial regulatory aspects of this design in the United States: The team at UF has received materials from the FDA outlining their regulatory and testing requirements. The FDA is loosening their requirements JUST FOR THE PERIOD OF COVID19 THERAPY -- and the team at UF has not fully digested the FDA information but they have people working on it. So they are on top of THAT aspect as well. Thanks, Gordon Gibby

I had to make a board to hold the transducers (depending on WHICH ONES YOU HAVE AVAIABLE) -- adafruit and a chinese clone of the BMx280 And then I learned I had to have a battery bckup/switchover. So I built a prototype circuit for that. Needs about 14-15 VDC from AC powered supply Shottky diodes switch to whichever supply is best. LED's inform you which one is providing Approx 60 mA trickle charging for a 3AHr sealed let acid, which should run this board for half a day. With a charged marine 12V battery you should be able to go a week. Gordon

I've been doing some testing of the impact of various parts of the code on the loop rate and it appears that using I2C displays slows the code substantially. When using I2C display the loop time is aprox 25ms with the vent off and jumps up to 250-300ms when the vent is running. Using the standard interface, the loop time is aprox 2ms with the vent off and 25ms with the vent on. The numbers are about the same regardless of display size. I've tested two i2c libraries with the same results. There is a third that is supposed to be faster which I will be testing this evening, but at this point I think we need to avoid using i2c for display, it's just too slow. Of note, I haven't yet found any other factors that substantially impact speed, including changing from the ssense to adafruit bmp280 libraries Steve S

Hey team, my transducer board just arrived, time to get my flow sensor up and running. Does anyone have the dimensions for the flow sensor with the 200pa sensor. There was lots of discussion about details early on, but haven't seen the final plans. What is the orifice diameter? Are we still seating it in a 1" diameter PVC or just the 3/4"? Are the taps right next to the orrifice plate or are we doing the 2 diameters upstream and 8 diameters downstream? Thanks Steve S

I think there's a problem with the way we're assessing airway pressure if the vent is going to be used for more than a few hours. Since the airway pressure sensors are absolute sensors, the measurement has to be corrected for the atmospheric pressure but the atmospheric pressure is only measured on vent startup. I'm concerned that measurement at a single time point won't be sufficient since weather substantially influences atmospheric pressure. Additionally, ambient pressure could change if a patient is moved in or out of a negative pressure room. As an example, according to NOAA the atmospheric pressure at the Minneapolis airport dropped by 0.45inHg (from 30.14inHg to 29.69inHg) over the past three days, that's over 15cm H2O. I think the best way to address this is to replace the BMx 180/280 with a differential pressure sensor with one port connected to the airway and the other port open to the atmosphere. An alternative would be to add a second BMx 180/280 sensor outside the airway to measure atmospheric pressure. Steve S

Query 1:I write a code http://tiny.cc/r85xnz but problem is serial output not giving constant length string although I have fixed the string length.What should I change it to produce proper and constant length string. Query 2:How can I read sensor data and controlling servo independently. In this code When servo delayed ,the sensor no read the data or printing.I also tried to using timer0 and timer1 as servo using timer1 but code did not work.Any hint or suggestion would be appreciated.I am added the output from my arduino code.

After 2,100,000 cycles the still functioning Rainbirds are off the test bench, replaced by two Orbit Jar Top valves. The test cycle time has been further reduced to 250 milliseconds. We should be able to accumulate more than 300,000 cycles a day. Bob Benedict KD8CGH