Open Source Ventilator question

Hey y’all, not sure how off-topic this is, but did you read about this? Coronavirus: 3D printers save hospital with valves

It sounds like if Covid-19 continues on the current course, there may be a need for open source medical devices:

Does anybody have experience with something like this in your area?


In case you’re interested in playing along at home:

As a bit of an experiment for how quickly our open production could spin up a new piece of hardware, I’ve been doing some work on this (officially on the clock for about 3 hours, unofficially after work yesterday and before work this morning for a few more hours), and here’s kind of the thought process behind why this is a good idea and which direction to go next. Let me know if this is something that’s interesting to you :slight_smile:

I’m heavily involved in the effort at working on opens-source solutions, where we have collected many designs. You can join the slack here: (link will expire in a few days.)

I’ve created a list of known projects here:


About the first one:

1 Like

TIL, in Michigan, if not the entire US, you actually need a prescription to buy a BVM. I tried both my human doctor and my dog’s veterinarian, but both were closed so I’ll have to try that part again tomorrow :frowning:

Another good article on the topic:

1 Like

Hi @robertlread,
Do you know if any of the ventilators in the list has been tested in real pacients?
I’m from Argentina and there are a lot of makers working on open source ventilators but I don’t think they are going in the right direction.
Best regards

Hi @robertlread, @pablocremades,

I think we can make a NIV quite easily, and yet satisfy safety concerns, if we keep it simple and transparent. Here is a drawing. (The mask is not shown, will probably be 3D printed or cannibalised from somewhere else).

Air is pumped from aquarium air pump (1). Probably 3 or 4 are needed in parallel to provide the required volume. This air is collected by downward displacement of water, in flask (2) in water trough (3). Excess air is spilled, pressure is maintained by the head h (5) and cannot build up. 2 normally-closed solenoid valves (4) and (6) control inspiration and expiration; when one is open, the other is closed. The solenoid valves are controlled by one-shot timers (not shown) such that each triggers the other. One one-shot controls the inspiration period, the other, the expiration period, and both are adjustable.

During the inspiration phase valve (4) opens and air that has been collected in the flask (2) is delivered to the mouthpiece. During expiration valve (6) opens for the patient to breathe out.

Using hardware timers instead of an Arduino means there is no software that needs to be audited. The operation of all components is transparent and easily tested by non-geek staff.

Safety alarms can be added. A watchdog timer can be constructed from a liquid level switch in flask (2) to reset a timer module. If the timer expires due to the flask not filling, a buzzer sounds.


It is worth noting that just something that pumps some air is not enough. A poor quality ventilator may do more harm than good. It is worth reading the specifications set out by the UK government for a “minimally acceptable” ventilator:


Hi @julianstirling,

Thanks for posting that document, I hadn’t seen it before.

I’m coming from the viewpoint that the main problem is acceptance. Many simple solutions are possible but if it fails user acceptance then nothing else matters. A big part of acceptance I believe is simplicity and transparency.

To that end, in that document, where it gives a pressure in cmH2O, we can prescribe that pressure with a water column of the given height. Where it mentions a volume in cm3, we can show that volume of air displacing water in a graduated flask. By using mechanical components, the operation and function are obvious. Exception is the electronic timer modules, but these conceptually simple to understand and test.

With these components, we can achieve mandatory mode, PRVC, SIMV-PC, pressure control (volume control might be tough), CPAP, FiO2 to some extent. I’m still thinking about peak/plateau pressure, that might not be achievable.

I’m not under any delusions. This isn’t the sort of thing GOSH does. But I’m putting it out there anyway just in case.


Interesting discussion taking place here as well:

1 Like