Hackaday Prize Entry: Open Sip And Puff

A sip-and-puff device is an assistive technology used by people who cannot use their hands. Being a quasi-medical device, you can imagine this technology is extremely expensive, incapable of being modified, and basically a black box that can’t do anything except what it was designed for. For his Hackaday Prize entry, [Jason] is building his own sip-and-puff interface that’s cheaper and more capable than the available commercial versions.

Sip-and-puff devices can be mapped to control a wheelchair, click a mouse, or press a key on a keyboard. You can do a lot with USB, so for this open sip-and-puff device, [Jason] is using the ever-popular ATmega32U4 microcontroller.

USB is only one part of the problem, and to measure the sips and puffs of air through a plastic hose, [Jason] is using a pressure sensor from Freescale/NXP. While this is very similar to what would be found in the off-the-shelf version of a sip-and-puff device, it’s rather hard to interface with. The current version of the board is using an instrument amplifier, and the mechanical connection between the pressure sensor and the board is slightly bizarre. [Jason] has a few ideas for a better sensor, and for the rest of the Hackaday Prize he’s going to work on redesigning this device with simplicity in mind.

29 thoughts on “Hackaday Prize Entry: Open Sip And Puff

    1. Yep, nice!

      Have a partner with MS and have been thinking about something like this.

      I’ve hacked a “attendant call” thingie for her, but it is not idea, requires rather constant adjustment to deal with her limited range of motion.

      Going to look up the SIP device and see, if perhaps, it can function as just a replacement switch for the call unit itself.

      Will be watching this.

      Thanks again, stuff like this helps folks who are not rolling in $$$ for the “medical” stuff that is so out of reach for so many that need things like this.

  1. One of the main issue with medical electronics is that they have to have a very strict isolation between things that might get in contact with mains voltage and the patient. Routing USB directly to the sensor is typically not an option if your sensor attaches to the patient’s body (been there…).

    Now, for Sip’n’Puff, there shouldn’t be any (conducting) connection between patient body and electronics, and a device following Jason’s principle is probably certifiable, but be warned that certifying anything for medical use is a) full of terrible surprises and b) expensive.

    1. The question is, if it is really arguable to treat this as medical device. You don’t treat computer mice as medical device, and you also do not treat spoons and forks – which also go into the mouth – as medical devices. Of course consumer safety is necessary with all products. But this is in my opinion very different to a device with a deliberate electrical connection to the human body.

      1. You do have to raise the bar a bit for any device intended to be used by anyone less than able bodied. Firstly perception there is a problem may be impaired, secondly ability to move body parts out of contact with faulty device may be impaired.

        Imagine for an example, you rig a mouse as a respiration sensor, that goes in a divot you knocked in your mattress. You sleep on top of it every night… does it seem like a good plan that the USB shield connects to the case ground on the computer, which is tied to the buildings ground. The cord is chafing nightly, unnoticed, one weekend, when you were going to stay in bed until 10, your neighbour gets an early start, but in passing between your houses with the lawn tractor, snags out the ground wire… Now this sounds unlikely and contrived, but betting that you never have a ground fault over 33% of your life is not real good odds, turn that into a sitch where you’re connected to something near 100% of your life and you’d want to get very anal about it.

        1. Yeah but being on the end of a plastic tube is pretty fool proof. I suppose you might want to double insulate it’s case, although it’s already got as much insulation as the keyboard under my fingers. If it were some sort of worn electrical sensor, you’d totally need to do proper isolation.

          1. With respect to the plastic tube, you are correct in that it will insulate the patient from the electronics (with the correct non-conductive tube that is). However, you now have to think about contamination of the tube and the sensor with respect to the patient’s breath. How to keep it clean so nothing can grow in it and thus make your patient sick or potentially kill them.

            This is a major challenge in hospitals, especially with anything involving respiration since human breath is quite humid. I won’t even go into what I’ve seen growing inside some equipment over the years because it wasn’t cleaned properly (this was typically ‘patient owned’ items, but not all of it) and also why there is a lot of thought and testing with respect to medical devices and contamination. The primary reason for Hospital Acquired Infections is things aren’t properly cleaned, and this can happen with ‘patient owned’ equipment as well.

          2. Biomed Bob the entire POINT of making this an open and repairable device is “What do you do when the device becomes fouled”. I won’t pretend to know what the “medical grade” versions use as a solution but in the open source’s case it’s slap another $8 pressure sensor into the socket and take another length of tube off the roll, though with a storage temperature rated for 125c presumably a low temp steam autoclave would work as well.
            Now RW, Biomed Bob if you have any excuses other than “Oh we have so much to think about when we design these things” and “It’s a free country we are allowed to make money on our work (by locking financially stressed people into a product)” for the rampant Bad Faith in medical devices I’d love to hear them, otherwise we all know about the existence of edge cases and that the human mouth is a dirty place.

          3. I’m not trying to say this isn’t a good idea, or that it shouldn’t be possible for the average person to repair them. All I was trying to point out was design considerations to think about from a safety standpoint. You are correct about the $8 sensor (I’d suspect you could find something cheaper that works better, I have on other items) but at the same time we get faced with some really nasty things from ‘patient owned’ equipment… such as deat roaches INSIDE ventilator tubing. (No, I really don’t want to talk about that one)

            Done properly the issue of contamination can be mitigated or addressed, either through an easily replaced sensor or by placing filtering material in line with a disposable portion of the tubing that gets changed regularly. I’ve seen what looks like a metal filter element on some Sip-n-Puff systems (think the old gas carburetor filter on cars) and we used to make our own tubing out of medical grade rolls rather than buy the manufacturers ‘tubing’ sets at ridiculous prices.

            As to what RW was trying to point out with respect to electrical safety was the liability issue with medical equipment. Yes, his example seems extreme but do I really need to bring up the McDonald’s ‘hot coffee’ lawsuit? Common sense often goes out the window when someone gets hurt and looks at how they can make a buck off of it. While they do their best to get the bulk of the extreme circumstances that come up, they also have to go beyond that to protect themselves from multimillion dollar lawsuits because someone got hurt from a device that got abused.

    2. The hospital I work for has hosted several Hacking Health events, pairing hackers/makers/coders with researchers/lawyers/etc. The idea, I think, is in part to flatten the learning curve a bit in terms of working within the super-restrictive and otherwise exclusionary medical environment. [Jason] might want to see if there’s a chapter in his neck of the woods.

    3. This is like any device that is on hackaday. You use it at your own risk.
      Just like a robot hand or eye reader.
      But you are right it is so stupidly expensive. and Yes I was working in the hospitals as well as the Electrician.
      Mind you the hospital it self try to get away with all they can do to save money. and some of those things are not getting properly certified Hospital equipment. I have had lots of fights with management over this.

      Great job…….

  2. Now that the lawyers have put their underpants (CYA) on us, we can use the technology. Now that the mamby pamby educators are speaking, what do u call a long sip? Suck! OMG what a bad word, why we’ll have to obfuscate the word and earn a degree.
    There are so many things the “educated” world throws at the handicapped, it’s …I won’t go there. Imagine that you are learning to speak, and suddenly they are throwing Latin at you. Huh!
    Obviously , I’d like to use this rig with midi on a performance setup.

    1. There’s already MIDI wind instruments, with breath sensors. I bet it’d probably be easier just buying one, by the time you’ve got the fine control worked out. This is a simple blow / suck sensor, so it’s a lot easier.

  3. Slightly OT but related: How would you go about using a single ported, relative pressure sensor to fuel (gasoline) level measurement?
    In principle it looks simple: Hydrostatic pressure acts upon a sensor et voila. Apparently though gasoline will slowly get through the protective silicone coating and damage the strain gauge die. Is there any simple method of separating such a sensor from corrosive media? I thought of some thin membrane with two ports, one leading to the sensor and filled with syntethic oil, another connected to the tank.

    1. They make media-isolated pressure transducers with a stainless steel diaphragm. They’re more expensive ( naturally ) than the non-isolated versions but there are plenty of them out there.

    2. Also look for fuel pressure isolators. They’re very common in the racing world because a lot of racers want mechanical fuel pressure gauges in the car – and for obvious reasons, racing sanctioning bodies don’t like a lot of extra tubing full of gasoline running through the dashboard. Gasoline pressure in, air pressure out.

  4. On the topic of the original issue – he mentions two issues with the pressure transducer being used: The interface circuit, and the difficulty replacing it. Both could be solved at the same time with an automotive MAP (Manifold absolute pressure) sensor. These pressure transducers typically have a 0-5V output you can feed straight into an ADC on a microcontroller (just needs a low pass filter), and have a connector on them for easy replacement. And if you break one, you can get a replacement at an auto parts store anywhere, right away.

    1. For a non turbo application they actually measure vacuum up to atmospheric. Then when you get a 2 bar one, you can read 1 atm over, but the range is from 2.5 to 5V so not terribly sensitive to smaller changes.

      1. They are normally referenced to atmosphere so you could potentially use a sealed pressure reference.

        Normal human breath though, pressure differential of about 0.3 psi, full force of healthy untrained human lung about 1.4 psi against restriction, trumpet player can maybe hit 2 psi. Anyway, on the 2 bar sensor, you get about 0.16 volts per PSI.

        I believe you can use a different reference voltage on them, since you supply 5V and ground and read value out, so a good regular MAP should read near 5V out when not hooked up to running motor. Whereas your 2 bar should read 2.5… so if you hooked them to say 3V, should get 3V and 1.5V… but that’s not very helpful unless you want to read large pressure swings. So if you want to spread the scale it’s potentially possible to go the other way and stick 12V in, but not sure how much of that they’ll survive.

        There are a few 1.5 bar MAPs around, not sure what applications used them though.

          1. MPX4115AP measures 0-115kPa; MPX4250AP measures 0-250kPa. These output a voltage from approximately 0.2V to 4.8V from a 5V supply and are mostly ratiometric within a small margin of error range… 1-2%, I believe. They also have a rather fast response rate of 1mS to settle at a final value.

            The 4115 might present an issue, however, if the person finds their self in a location below mean sea level as there is only a slight margin for over-pressure near the upper end of its range.

  5. Regarding the “slightly bizarre” connection to the pressure transducer, when my work mounts them, we use either a row of through holes that line up with the pins and bend them at the small indexing notch on pin 1, or a vertical mounted female header row surface mounted on the board, again bending the pins to drop into the header.

    1. I was going to comment that the mounting for that particular sensor may seem odd, but that actually provides for an easy replacement of the sensor should it become fouled- unscrew it, slide it out of the header and replace.

  6. This is the basis of Boba FETs helmet. This is cutting edge research to go with augmented reality and user interface. Wouldn’t it be easier to match the movements to the eye, then use a sensor on the persons temple to be the right mouse click and the sensitivity of the central nervous system itself to carry out complex operation?
    This is more than a medical device. This is the future of the new Mandalorean Empire we’re talking about here! We just need the personal cloaking device, jet pack, heavy armor, shoulder mounted laser gun from Predator and some funky claws, maybe space suit worthy climate control and radiation protection, and bam! We have a hunter predator armored suit worthy of any sci fi movie.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s