Sprint Day 2: What developing world substitutions are there for flow meters?

[tkutscha]

You can get a cheap acrylic air flow meter (<$20) and use a series of optical sensors with an MCU to read it out live.  I initially thought of putting a magnet ball in there and using hall-effect senors, but changing the weight of the ball would probably upset calibration.  Since you're looking for 0-80 LPM, you could put two in series to get accuracy at the low and high ends.

If you have a bunch of them, you could point a cheap camera at all of them to read out the ball heights with image processing software.  Maybe a super cheap webcam and some image filters.  Make the ball luminescent or glow under an UV lamp to make it stand out more?

Looking on Amazon: $16 for a flow meter, $10 for a cheap USB cam, and a raspberry Pi to look at several of them ($35).

[Simon]

sounds really fail safe and medical.....

[dietert1]

If i remember right, others have been able to measure heart rate of by-passers using video, so video is a powerful tool.

A flow meter i built 15 years ago for respiration measurements in a university institute was based on a difference pressure sensor and an "airflow resistor" that can be made with a 3-D printer and/or some tissue. Those meters i made had an instrumentation amplifier to read the sensor bridge, a CMOS analog multiplexer to provide 4 different steps of gain and a 12 bit ADC in a MSP430 microcontroller.

Recently i thought a lot about whether such a setup may be enough to control respiration support in a ventilator. In the documents circulating right now they propose monitoring pressure, maybe monitoring flow could be as efficient. Yet again those projects are not electronics, but interdisciplinary.

Regards, Dieter

[Simon]

A flow meter i built 15 years ago for respiration measurements in a university institute was based on a difference pressure sensor and an "airflow resistor" that can be made with a 3-D printer and/or some tissue. Those meters i made had an instrumentation amplifier to read the sensor bridge, a CMOS analog multiplexer to provide 4 different steps of gain and a 12 bit ADC in a MSP430 microcontroller.

Recently i thought a lot about whether such a setup may be enough to control respiration support in a ventilator. In the documents circulating right now they propose monitoring pressure, maybe monitoring flow could be as efficient. Yet again those projects are not electronics, but interdisciplinary.

Which is what I suggested above. The main issue is getting the pump to work in sync with patient. If you don't their lungs won't survive being on any of these contraptions for weeks.

[dietert1]

Anyway a medical machine needs some redundancy. In fact the proposals include mechanical safety valves and the like.

At that time the respiration project was some kind of reuse. Our pulse oximeter modules include logging an ECG signal - again for redundancy and also to determine PTT, a parameter in sleep diagnostics. In those meters i mentioned the pressure signal replaced the ECG signal, so with very few changes we arrived at a solution combining pulse oximetry with airflow meter. It's a USB device with isolation.

Regards, Dieter

[EEVblog]

Another open source ventilator, in Danish.

https://www.dr.dk/nyheder/regionale/nordjylland/danske-forskere-har-brugt-blot-14-dage-paa-lave-en-simpel-respirator

[donotdespisethesnake]

So did we give up on this?

FWIW, if anyone can deliver a hardware prototype, I'd have a go at writing the software. I've no experience with medical, but have worked in other safety critical areas.

Edit: The main code for the Medtronics PB560 runs to 53,000 SLOC. That's around 135 programmer-months, a little over 12 years of effort. I hope that gives some useful perspective...

Edit2: sorry I miscounted. There are nearer 96,000 lines of code. So double that timescale! (excluding blank, comment only lines but including C source this time )

[dietert1]

At SEAT car factory in Martorell in Spain they needed one week to setup a production of 300 devices per day. They are using on stock wiper motors plus SMPS blocks to operate rubber ventilation bags. The device has about 110 parts. Volume per cycle can be adjusted mechanically, speed gets adjusted by supply voltage. Apparently medical doctors preferred something that doesn't need supervision.
Don't know how they control the exhaust valve and i did not read about backup batteries. That virus changes everything.

https://www.seat-mediacenter.com/newspage/allnews/company/2020/From-making-cars-to-ventilators.html

Regards, Dieter

[magic]

So did we give up on this?

Have you seen the Project Leader and Scrum Master anywhere recently?

I hope it was an early April Fools attempt. If the guy was serious, I kinda feel sorry for him

Last Active:    April 10, 2020, 04:23:06

[MK14]

So did we give up on this?

Have you seen the Project Leader and Scrum Master anywhere recently?

I hope it was an early April Fools attempt. If the guy was serious, I kinda feel sorry for him

Last Active:    April 10, 2020, 04:23:06

The last post(s), they ever made (unless they come back), was dated 1st April 2020.

https://www.eevblog.com/forum/projects/sprint-day-0-what-open-source-ventilator-projects-exist/msg2994090/#msg2994090

The virus situation, isn't something which people should make April 1st joke threads about, so I hope their apparent disappearance on April 1st (last post, to date), is just a coincidence.

[MK14]

At SEAT car factory in Martorell in Spain they needed one week to setup a production of 300 devices per day. They are using on stock wiper motors plus SMPS blocks to operate rubber ventilation bags. The device has about 110 parts. Volume per cycle can be adjusted mechanically, speed gets adjusted by supply voltage. Apparently medical doctors preferred something that doesn't need supervision.
Don't know how they control the exhaust valve and i did not read about backup batteries. That virus changes everything.

https://www.seat-mediacenter.com/newspage/allnews/company/2020/From-making-cars-to-ventilators.html

Regards, Dieter

That is amazing!

But they had the needed ingredients, to do it that quickly.

The right engineers, in the some building/organisation, who knew each other (presumably or mostly).
The test dummy thing, to test out the prototypes.

A massive manufacturing plant, with recently suspended production.

To try to do it as a thread (here), open source, didn't seem like a good idea. The chap/OP (sorry), didn't seem to have the right skills set, to make it happen.

[Simon]

At SEAT car factory in Martorell in Spain they needed one week to setup a production of 300 devices per day. They are using on stock wiper motors plus SMPS blocks to operate rubber ventilation bags. The device has about 110 parts. Volume per cycle can be adjusted mechanically, speed gets adjusted by supply voltage. Apparently medical doctors preferred something that doesn't need supervision.
Don't know how they control the exhaust valve and i did not read about backup batteries. That virus changes everything.

https://www.seat-mediacenter.com/newspage/allnews/company/2020/From-making-cars-to-ventilators.html

Regards, Dieter

that still sounds like the type they don't want, no mention of how it interacts with patients without destroying their lungs unless it is used for limited time only.

[EEVblog]

To try to do it as a thread (here), open source, didn't seem like a good idea. The chap/OP (sorry), didn't seem to have the right skills set, to make it happen.

And dozens of other teams aren't even close.
Hopefully this whole worldwide endeavor shows that going from prototype to a practical working and safe (and approved) medical device is extremely difficult and time consuming.

[Gyro]

At SEAT car factory in Martorell in Spain they needed one week to setup a production of 300 devices per day. They are using on stock wiper motors plus SMPS blocks to operate rubber ventilation bags. The device has about 110 parts. Volume per cycle can be adjusted mechanically, speed gets adjusted by supply voltage. Apparently medical doctors preferred something that doesn't need supervision.
Don't know how they control the exhaust valve and i did not read about backup batteries. That virus changes everything.

https://www.seat-mediacenter.com/newspage/allnews/company/2020/From-making-cars-to-ventilators.html

Regards, Dieter

I'm surprised that wiper motors have the durability to run 24/7. They run in good airflow under the scuttle panel and normally don't have a high long term duty cycle. Maybe they can get away with it due to reduced mechanical load (?).

There again, the only ones I've examined are those that have failed due to exposure to the elements and age.

[MK14]

To try to do it as a thread (here), open source, didn't seem like a good idea. The chap/OP (sorry), didn't seem to have the right skills set, to make it happen.

And dozens of other teams aren't even close.
Hopefully this whole worldwide endeavor shows that going from prototype to a practical working and safe (and approved) medical device is extremely difficult and time consuming.

There are so many potential pitfalls, which are not so obvious to the less experienced.
I'd be especially worried about getting hold of the correct parts, on time.
Typically, you can order a few hundred of a wide range of parts, from the usual electronics parts suppliers, such as Digi-key.
E.g. They might have 153 in stock, of the microcontroller, you have used, for the ventilator.

But, if you need 25,000 of them (or even a few thousand), you may well find there is a huge lead time, of many months.
Substituting other microcontrollers, could well lead to very time consuming software rewrites and retesting again. Especially if the hardware I/O configuration (PWMs etc), is completely different, between the MCUs.
Even just a change of case. could mean re-spinning the PCB.
Mechanical parts (especially if custom), are probably even worse.
I think it usually takes a very long time, perhaps many, many years, for new medical devices, to complete all their extensive safety (critical) tests and approvals.

[Simon]

i recently had an interview with a company that is designing a medical device for virus testing (not covid-19 related). the test involves heating the sample. This fact alone means that their software has to be independently assessed and the device needs non software fail-safes. the risk is that if the container overheats and explodes those nearby could be contaminated. That is all the risk. Rasberry pi's are not qualifiable.

It's not as simple as people think.

[OwO]

You can sense airspeed as a substitute for flow rate, and airspeed is easily sensed with a pitot-static system (2 pressure sensors).

[MK14]

You can sense airspeed as a substitute for flow rate, and airspeed is easily sensed with a pitot-static system (2 pressure sensors).

Good idea, in principal.

But setting up and developing such systems (properly), can be very time consuming. There are so many different flow rates, pressures, temperatures and ratios of air to oxygen levels.

Designing for a high percentage of Oxygen (from what some hospital sources seem to be talking about, as regards treatments), is probably another factor. Potentially making it take a lot longer to sort out the ventilator.

[Simon]

All flow measurement tends to revolve around speed measurement. It's how car mass airflow meters work. there are plenty of differential air pressure sensors that have a logarithmic output to counteract the exponential increase in pressure differential across a restriction.

The challenge in the ventilator design is that it correctly interacts with the human body so that it works with the body, the result of working against the body is lung damage.

[MK14]

All flow measurement tends to revolve around speed measurement. It's how car mass airflow meters work. there are plenty of differential air pressure sensors that have a logarithmic output to counteract the exponential increase in pressure differential across a restriction.

The challenge in the ventilator design is that it correctly interacts with the human body so that it works with the body, the result of working against the body is lung damage.

Although it is indeed challenging to make sure the ventilator correctly interacts with patients.

You also (ideally), want to make sure that the proposed hardware, performs safely (to specifications), despite changes to various parameters. When you start listing them, you can end up with a huge list.
Supply voltages, component tolerances, ambient temperature, mechanical variations, Oxygen percentage, air humidity, operator error, faulty Oxygen supply, etc etc.
Covering all those different things, can take years of research and development.
Otherwise, the machine could end up harming patients.

Presumably some sort of compromise to the usual required safety standards, has been taking place. I don't know how far they have gone, in that respect. Maybe in the future, there will be reports about it.

What engineers (electronic/mechanical) really need, is some kind of definitive specification, as to what is required and what will be accepted. Without such information, it can be very difficult to come up with proposals.
But it could be argued, that even a spec takes time to write and maintain.
I.e. The OP needed to realise this, and come up with such a specification, very quickly.
Maybe they did, maybe they didn't. I found it too confusing, with the OP, starting multiple threads, and changing the top of thread(s), on a daily basis. It got too confusing, at least for me.

I'm still NOT clear if it was an April 1st joke or not ?

[chris_leyson]

A small turbine might be one approach as this wouldn't be effected by changes in air temperature or humidity, perhaps a small DC fan used in reverse. Small turbines were used in mines to measure airflow and here is a picture of an L. Cassella air meter No. 312. I think it's calibrated in cubic feet.

[coppice]

All flow measurement tends to revolve around speed measurement. It's how car mass airflow meters work.

Mass gas flow sensors don't measure speed. As their name says they measure mass flow. For gases mass movement is a combination of speed and pressure, and for most gas flow measurements its the measurement you really want. The actual MEMS sensors typically used these days for mass gas flow measurement are very cheap. If BOM is your major issue, building the basic MEMS device, from someone like Memsic, into your own tube should be a winner.

You could also look at the ultrasonic gas measurement solutions being built for domestic energy meters. You can measure flow quite accurately for a few dollars with those, but the mass flow MEMS sensor is probably a lot quicker to design in.

[Simon]

All flow measurement tends to revolve around speed measurement. It's how car mass airflow meters work.

Mass gas flow sensors don't measure speed. As their name says they measure mass flow.

Have a look at vehicle mas air flow meters. They are air speed meters. I use them, we take the meter out of the tube it comes in and into a smaller tube to increase sensitivity because same mass in a smaller tube has to flaw faster. They work on the basis of air cooling a temperature dependent resistor whilst compensating for ambient.

Any flow measurement based on pressure differential is a speed measurement, again the cross section of the channel needs factoring in. If you measure pressure drop over a restriction of same mas flow but in different diameters you will get different pressure results.

[coppice]

Have a look at vehicle mas air flow meters. They are air speed meters. I use them, we take the meter out of the tube it comes in and into a smaller tube to increase sensitivity because same mass in a smaller tube has to flaw faster. They work on the basis of air cooling a temperature dependent resistor whilst compensating for ambient.

The reason you have increased the sensitivity is that the MEMS sensors only sample the mass of the air close to the edge of the tube. If you reduce the tube diameter that becomes a bigger fraction of the total mass flowing through the tube. Its not a linear relationship, because the gas near the middle of the tube typically moves faster than the gas near the edge, but this is fairly consistent over time, so you can calibrate a good mass flow figure for any particular tube.

[Simon]

A small turbine might be one approach as this wouldn't be effected by changes in air temperature or humidity, perhaps a small DC fan used in reverse. Small turbines were used in mines to measure airflow and here is a picture of an L. Cassella air meter No. 312. I think it's calibrated in cubic feet.

you would have to account for "stiction" or startup inertia and friction that would bias results. I have just bought a peak flow meter to measure how strong my lungs are. The instructions state to test the freedom of movement of the marker, it's a very simple instrument that relies of free movement with low friction of the needle.