EEVblog #909 - Defibrillator Teardown

[EEVblog]

Defibrillator Teardown!
Inside the Heartsine Samaritan Pad AED heart defibrillator.

Datasheets:
http://www.mouser.com/ds/2/205/CS20-22moF1-475252.pdf
http://www.vishay.com/docs/94386/vs-30tpspbf.pdf
http://www.mouser.com/ds/2/196/irg4ph40upbf-937889.pdf
https://www.fairchildsemi.com/datasheets/FQ/FQP45N15V2.pdf
http://www.promelec.ru/pdf/ISD4004.pdf

[Andy Watson]

No DSP? I was expecting some DSP goodness in there along with a well protected ECG amplifier to determine the presence (or lack) of a pulse or fibrillation.

[Solberg]

@3:17 lol. That's not a European Conformity Mark, it's a China Export mark.

https://en.wikipedia.org/wiki/CE_marking

https://www.google.dk/search?q=china+export+differences+european+conformance&biw=1525&bih=746&source=lnms&tbm=isch&sa=X&ved=0ahUKEwiS2KXN9LbOAhWMIJoKHas7CeoQ_AUIBigB&dpr=0.9#imgrc=o5IB3xidF06snM%3A

[bktemp]

Interesting teardown.
The current switching during the biphase pulse can not work as shown: When the SCR is triggered, it can't switch of until the current goes to almost zero. So the IGBTs must be used for the first half of the biphase pulse to interrupt the current before the SCRs dump the remaining energy at reverse polarity into the PUT. That's probably why they had to use IGBTs in one leg of the H-bridge. The relays can't be used to interrupt the current (switching 1kV DC at >10A  off requires a really large switch).
But the reverse engineered schematic looks right, so I can't figure out how it works.
Is is probably impossible to fool the device by putting the electrodes into a bucket of water to make it work to measue the output waveform.

[elliottveares]

Dave, you don't Defibrillate someone having a heart attack (Myocardial Infarction)  ; you defibrillate someone having certain forms of cardiac arrest.

[Goodwill Hunting]

Dave, you don't Defibrillate someone having a heart attack (Myocardial Infarction)  ; you defibrillate someone having certain forms of cardiac arrest.

Actually, you don't defibrillate someone having cardiac arrest.  Cardiac arrest means literally no cardiac electrical activity.  In medical terms, we call it asystole.  In lay terms, it would be "flat-lining".  Defbrillating someone in asystole will accomplish nothing.

The point of defibrillation is to override a lethal rhythm, allowing the heart's natural pacemaker to restart.  The rhythm you would defibrillate would be ventricular fibrillation or v-fib.  That's the big one everyone is worried about, and it's the one that kills the most people by far.  There are other rhythms that get defibirillated, such as pulseless ventricular tachycardia, torsades de pointes, and even some atrial dysrhythmias, although the atrial dysrhythmia shocks are syncronized to occur on a specific point of the cardiac electrical cycle so we don't make the patient go into an even worse rhythm. 


As a point to the video, Dave made note of the Joules going from 100 to 120 to 150 (or something like that).  The concept here is that the first shock is at the lowest Joule.  Following that is 120 seconds of CPR, followed by the AED analyzing the rhythm again.  If the AED determines a shock is warranted, it will step up to the middle Joule because the lower Joule did not work.  Then, 120 seconds of CPR and the AED will analyze and, if warranted, shock using the highest Joule.  Any further shocks will be at the highest Joule.   Each manufacturer of biphasic defibirillators uses or can use different Joule setting depending on what they get certified with.  It is not important for a user to remember the exact Joule setting, because each machine may be different.

Monophasic defibrillators were all the same and used 200J, 300J, and 360J.  These have been mostly phased out, although you will still find monophasic defibirillators in older and less advanced healthcare settings.  These monophasic defibrillators used more Joules because they were less effective.


For anyone worried that they need to understand any of this stuff to use an AED:  Your concerns are unwarranted.  The AED talks to you.  You open it and it will tell you to apply the pads and how/where to do it.  It will tell you to press the anaylyze button.  It will tell you to start chest compressions.  You don't need to know anything to use an AED, and you might save a life.

[filssavi]

the reverse engineered schematic looks right, so I can't figure out how it works.

My guess is that the series inductor is part of a resonant tank that lets the 2 series Tyristors turn off, then the igbts/big SCR is triggered and the negative pulse is delivered, that said I haven't done any circuit analysis and i might be talking out of my a**  so don't quote me on this

[max666]

That turning off of the SCR's is in deed a big crux in Dave's explanation. Is it possible the polarity of sternum and apex is wrong?

[alxpo]

Very disappointing. I waited when Dave explain main magic of the circuit - closing of SCRs in proper time, but it was futile.
The version with wrong polarity and first pulse through IXYS and IGBTs looks plausible.

[LazyJack]

Regarding the clock drift of the RTC. It's not really an issue. If you need the timestamped log of what happened, you can read the current time from the device and correlate the log to that. I assume that you want the log shortly after it was activated.
Provided this is really the purpose of the RTC in this device.

EDIT: Oh, just saw that Mike had already commented just about the same on Youtube, before I wrote this here.

[mtdoc]

Nice teardown Dave.  It would be interesting to know more about how it does the rhythm analysis to determine if there is a shockable rhythm - but I suppose most of that magic happens in software.

Dave, you don't Defibrillate someone having a heart attack (Myocardial Infarction)  ; you defibrillate someone having certain forms of cardiac arrest.

While technically correct, since most cases (about 70%) of cardiac arrest are due to myocardial infarction - it does turn out that most uses of defibrillators occur as the immediate result of an MI. 

So, I think it is understandable that for the general public - the 2 terms are used synonymously - even if it is not quite correct.  "Heart attack" is really a layman's term in any case.

And FWIW, the vast majority of MI's do not result in cardiac arrest.

[elliottveares]

Dave, you don't Defibrillate someone having a heart attack (Myocardial Infarction)  ; you defibrillate someone having certain forms of cardiac arrest.

Actually, you don't defibrillate someone having cardiac arrest.  Cardiac arrest means literally no cardiac electrical activity.  In medical terms, we call it asystole.  In lay terms, it would be "flat-lining".  Defbrillating someone in asystole will accomplish nothing.

Ventricle Fibrillation and Pulse-less Ventricle Tachycardia are both forms of Shockable Cardiac arrest as cardiac arrest means to have no heart output (to have no pulse).  Asytole as you said is also a form of non-shockable cardiac arrest what can not be treated with a defibrillator.

[max_torque]

I did a limited teardown on a Pico Differential probe a while ago:


https://www.eevblog.com/forum/projects/what-am-i-mystery-ic-competition!/


Links in that thread to more info on the circuit architecture used etc.

[marshaul]

Yes, please, Dave! Please do a teardown/reverse engineering of your differential probe.

[mikeselectricstuff]

No DSP? I was expecting some DSP goodness in there along with a well protected ECG amplifier to determine the presence (or lack) of a pulse or fibrillation.

It's a very low bandwidth signal, and doesn't need to be analysed in real time, so shouldn't need a DSP to process.

[Andy Watson]

No DSP? I was expecting some DSP goodness in there along with a well protected ECG amplifier to determine the presence (or lack) of a pulse or fibrillation.

It's a very low bandwidth signal, and doesn't need to be analysed in real time, so shouldn't need a DSP to process.

It is low bandwidth but I would expect it to be analysed in real-time. You don't want to shock a person with a normal pulse.

[Maxlor]

No DSP? I was expecting some DSP goodness in there along with a well protected ECG amplifier to determine the presence (or lack) of a pulse or fibrillation.

It's a very low bandwidth signal, and doesn't need to be analysed in real time, so shouldn't need a DSP to process.

It is low bandwidth but I would expect it to be analysed in real-time. You don't want to shock a person with a normal pulse.

That LPC2210 is a 60MHz ARM7 part; it should be fast enough for the signal analysis of a single channel ECG. Data recording happens at 100-500Hz typically, so after a few seconds of recording, there's not a big amount of number crunching that has to happen. Mind you, the first ICDs (implantable defibrillators) came out in the 1980s, and they managed to figure out whether and when to shock

[Maxlor]

So what I don't get with Dave's diagram (aside from the part that he says Thyristors can't turn off, but then they're supposed to?) why do the current path parts use so many different topologies? SCRs of various sizes, IGBTs, one path with an inductor one without; why not use only IGBTs? Or why not use more of those big SCRs to replace those pairs of SCRs in series?

Btw, I work for a company that builds med tech products, although being in R&D, I'm not directly involved with the production stuff, so I'm not very familiar with the regulations; My impression though is that they focus more on process than on parts. What I do know is that we're not required to have certification for every single component; it would be infeasible if not impossible. What we are doing however is machine-inspect every single part down to small passives like ceramic caps, and this data including photos from all sides is kept in storage. This is so that in the event that there is a defect in a part that went undetected, we can go back and check whether maybe the whole series might have been affected, and find out which products exactly those parts have gone into. So yep, full traceability for every single 0201 passive. Just in case you wondered why health costs are so high

[bktemp]

So what I don't get with Dave's diagram (aside from the part that he says Thyristors can't turn off, but then they're supposed to?) why do the current path parts use so many different topologies? SCRs of various sizes, IGBTs, one path with an inductor one without; why not use only IGBTs? Or why not use more of those big SCRs to replace those pairs of SCRs in series?

SCRs are probably easier, cheaper and more robust, because you give them a short pulse (can be easily coupled using a small transformer) and they keep conducting. For IGBTs you need to keep the gate for the full on time at 15V otherwise they will go into linear operation and fail.
The IGBTs are probably necessary to switch the current off before reversing the polarity. That can't be done using SCRs.
The inductor makes sense, because SCRs don't like a rapid current rise at turn on. You often see inductors in series with SCRs when they switch a resistive load.

[filssavi]

now that others have brought it up, wrong polarity on the electrodes is much more plausible than a resonant switch...

Still as maxlor said the whole switching topology is strangely a mix of SCR and transistors, maybe SCR having a much simpler structure is more reliable, and they used them in as many places as they could get away with, so the only transistors (IGBT) are the ones that must be also turned off not just turned on, what i dont really get is why the 2 SCR's in series in one place and a single bigger one in the other, i would be tempted to say cost but i don't know it doesn't seem to make much sense to me

[VK3DRB]

The medical device industry is a very tough environment for good reason - safety. The Cornell Dubilier caps were used because they are deemed as critical components and of course would need to be of the highest quality and reliability. Everything about them would need to be reported for IEC-60601 compliance, along with other devices.

There is a massive amount of paperwork, processes, testing and time to get a product to market. Hardware development is one thing (IEC-60601) but software development is another major task (IEC-62304). Writing code or creating a circuit that works is only one small part of the huge bureaucratic processes to get a product developed. Testing and documentation is very extensive. As are the QA processes. And IEC 61000 compliance for EMI/ESD. It is not the easiest career.

At the heart of everything is risk analyses - identifying and mitigating risks. The manufacturer can be audited at any time by the FDA, TGA or whoever else is relevant to the geography you are selling product. The manufacturer owns the product risk. If a manufacturer has falsified any documents, he can end up being shut down or even doing time (esp if someone died as a result of a dodgy product).

Dave had issues opening the pads. They would have been hermetically sealed and sterilised. And that is another complex controlled process. The defrib would have had multiple feedback mechanisms to control and detect that everything was working correctly. I would suspect even with the power "off" there would be battery monitoring going on in the background. You would not want to need to use one of these to discover the batteries are flat.

[Paul Moir]

The medical device industry is a very tough environment for good reason - safety. The Cornell Dubilier caps were used because they are deemed as critical components and of course would need to be of the highest quality and reliability.

Couple other things about them: a +10%/-10% spec is a little unusual for an electrolytic, but obviously required for this thing since you don't want too much energy available & can't overload that smaller SCR.  Also there doesn't appear to be balancing resistors so perhaps they're matched for leakage.  Maybe that's what the dots are indicating.  Throw in a form factor requirement and you might be married to one manufacturer.

The diodes appear to be straight up 1000v PIV so they're not pulling anything fancy there.
 

[zl2wrw]

A quick word of warning if you ever have to use an AED on a casualty for real - many models "chip" the electrodes, so that they can only be used for one patient. Usually they do this by checking the state of the electrodes (new or used) when you wake the AED up (typically by opening the lid).

The catch is that after putting the electrodes on your casualty, and finding that they do not require a shock, or after a shock successfully restores a useful heart rhythm, you definitely should NOT "tidy up" and unplug the electrodes or close the lid on the AED (leave that the ambo crew). This is because unplugging the electrodes or closing the lid can put the AED back into deep sleep mode, and should your casualties condition worsen, requiring a shock, a prematurely shut-down AED will likely refuse to work because on wake up, it will detect that its electrodes are "used" 

[DrSchweizer]

Hi Dave

A defibrillator isn't used "to restart a heart that had a heart attack", its used to resynchronize the electrical activity of all cells when that was lost. So when someone has by example a heart attack and in regions with insufficient oxygen supply the cells start to act chaotically and do not spread the electricity uniformly over the heart but every cell does whatever it wants the heart stops to pump (called ventricular fibrillation) and now a defibrillator will cause all cells to "discharge" at once and restart a normal rhythm.

I hope that was explained well enough ...

Regards from Switzerland


-Schlomo (cardiologist)

[Halcyon]

This particular (as most AEDs) will actually detect CPR being performed and provide visual/audible feedback. Every compression on the chest is detected and measured by the device. The AED will only allow a shock when it detects a certain irregular heartbeat. It won't help anyone who has no cardiac output.