The Amp Hour Electronics Podcast

A weekly show about the trends in the electronic industry.

You are here: Home / Radio Show / #335 – When the TV watches you

#335 – When the TV watches you

Play

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | RSS

 

Next week, we’ll be welcoming Bunnie back to the show! Ask your questions here. Listen to the first time we had him on (ep 84!) in the past.

Thanks to FlashPro for the image

Comments

  1. One of the reasons for having an open source hardware laptop is that you don’t have to relay on closed source bios. For example modern intel PCH (AKA south-bridge) have an embedded MCU for remote management, which runs indipendently of the main OS, with a close source firmware. Clearly not something to be super-happy about.
    See for example this slides: http://2012.ruxconbreakpoint.com/assets/Uploads/bpx/Breakpoint%202012%20Skochinsky.pdf
    Btw, nice show as always, you guys rule!

    • SHK, Thank you for sharing that information, I was completely unaware. I will checkout those ruxcon slides. I am curious if there is a video presentation that accompanies them.

  2. Usually you are asked to lay down and be still when they are taking your ECG for diagnostic purpose such as to rule out a heart attack or capture an arrhythmia. One exception to that is when you undergo an exercise ECG in which case you are usually running on a treadmill with an ECG hooked up to you.The difference in between the two lies in what you are trying to look at and subsequently what you can get away with in terms of limiting the bandwidth and more aggressively filtering. An ECG that requires you to be still has the greatest sensitivity at capturing any unusual electrical cardiac events and thus has a wider bandwidth with less filtering. In contrast, an ECG that is taken during exercise is far more limited in bandwidth and certain details of the ECG itself are attenuated to the point that they may not even show up. Again, all depends on what you are looking for and if missing details is okay with you.

    Artifacts on an ECG happen all the time and the question of whether or not a particular spike is from noise or is actually being generated from the heart itself is a very common. Physicians get pretty good at determining if something is noise or actually important by practicing on thousands of EKGs from real patients during their training. There are also a few tricks physicians can use to help them in addition to just experience such as other vitals as well as waveform morphology analysis.

    As far as why the ECG is so hard to get clean goes… it all comes down to magnitude. Generally speaking, the signal-to-noise ratio (SNR) of an ECG signal is actually less than 1. This is because electrical noise generated by the muscles in the chest and abdomen actually have a higher magnitude than the heart muscle due to size. Add to that your chest muscles actually attenuate the electrical activity of your heart as seen by the ECG because the signal from the heart has to go through the muscle to reach the electrodes. I want to also point out that the average magnitude of the highest peaking signal on an ECG is about 0.001v. Induced AC-line noise can easily be one or two orders of magnitude higher especially when the front-end is high impedance as in this case.

    Regarding electrode placement, your limbs can be roughly be seen as an extension of your body in terms of electrical conductivity (wires) so it is acceptable to place electrodes on your body or on your limbs. There is medical literature out there that shows that the vector is slightly altered by doing so but whether or not it is clinical significant/relevant is up for debate. Regardless, the torso electrode version was selected because it is faster to put electrodes on the torso versus each limb as you usually just have to remove/manipulate one article of clothing. In addition, there is less attenuation of the signal as it doesn’t have to go through the resistance provided by the limbs. Some even say there is less noise since limb muscles should have minimal affect on torso electrodes as they are not in the pathway of the electrical circuit when only torso leads are utilized.

  3. For the algae analysis the term you were like looking for is “electrochemical impedance spectroscopy”. The reason why you use AC for measurements is that certain processes are activated/occur at certain frequencies. You can also plot impedance vs frequency and the make up a model for the curve (usually a bunch of half circles) to extract parameters such as junction resistance/capacitance and what not. This was basically my project at uni where i looked at characterising perovskite based solar cells. Also can send you the software for this analysis if you want Chris

  5. The “Humans Need Not Apply” video promotes some really bad economics. The horse analogy doesn’t make sense because horses were never actors in the economy. Horses were used as tools or machines and obsoleted by better tools and machines as has been going on for thousands of years. Horses talking about their future employment makes no more sense than slide-rules talking about theirs. Neither were ever employed to begin with. There is an economic concept called “comparative advantage” that debunks most of this modern Luddite stuff (which isn’t much different than the old Luddite stuff). Short term disruptions can effect employment in specific sectors but over the long term, there will continue to be jobs available unless we eventually reach a point where the desire for ever more stuff starts to wane…

  6. That was pretty thin on the premises of it! It was breaking news when you recorded it, but BestBuy’s VIZIO TVs were revealed to have some pretty invasive habits. NPR has a nice followup bit where a fellow paid to read the Terms Of Service notes that ‘We never sell your data’ is followed soon enough with a guarantee that their use of your ISP to send usage and other data for people who interact with the set up to any/all motherships is not something they’ll pay extra to do.
    Moreover Dave’s assertion that espionage takes >>0.4W is facile in more ways than decoding 3 to 5 QAM (ATSC-ish?) 1080p channels in that same power (with ASICs, not the general solution with noise instrumentation afforded by 300 W Analog DSP) is common. It’s the cablecos’ indecision (and maybe some hardening) that blows 30-50W on the backend keeping broadcast menus and forced updates via broadband available in spades.
    Not to say that if Dave were in Cleveland he’d have great lighting even to CMOS and IR imaging devices (fire safety sell,) and a camera that understands his ‘inflatable thong’ is not half of an inflatable bikini, but he would and I think cameras waiting or ‘processing’ for a still moment are commodities on the IP hoof.
    There was a pop business press^W^W^W hbr.org article by Barton, Manyika and Williamson about AT&T, Amazon and others being forward-looking companies that recover better than others about (Steve Case linked it, for example) but I can’t help but think those are consumer-agnostic firms with wretched opening offers and ethics.

    If you wanted to get into worthy paranoid domains, you could consider work like the Pita Hack put subtly into sub-bandgap computers in ring isolation; and optioning smelly actors below TPC and other microcode hypervisor rings that live in supposedly crufty WiFi IP, and of course cross them in amusing ways with baseband attacks, s.t. somewhere in the mists between All Possible Bootimages being known before their birth (as co-primes of some firmware fuse key) and chipset-compatible battery chipsets exposing and subverting your true potential in the cloud (and by crashing browsers at awful times,) maybe you’d look up mechanical CAD extensions to EMACS.

    Honestly, I had hoped to hear some build reports on OLIMEX things; here Farnell’s been supporting and shipping 80386 System Software Design Manuals since…then, and you mention OLIMEX like it’s on even footing. eh…I heard the MMU was separate, etc. etc. That’s fine if it should be, but the mints are different by a quarter century and there’s a gap in backstory I might not have been getting from Chinese blogs. Yet.

    Similarly the ‘Chinese Factory Implements Robots’ is supposed to unleash the Tranche of Debug Jeering. What happened?

  7. Good show. With Amazon Alexa/Echo in the kitchen, and Google Home in the living room, we get two different ladies answering at different times.
    ===========
    Yes – ECG’s do show different noise sources, in different frequency domains. But what you had on the show was much worse than what we see with commercial units in the clinic. Some details of what is being looked are pretty easy to see even with lots of noise – others are more difficult to eek out. I wouldn’t be able to fully evaluate the morphology of the your P-wave or QRS, but some basic things looks ok .

    Take a look at where things were at the beginning of the field to respect the guys who had to do it the hard way: https://en.wikipedia.org/wiki/Willem_Einthoven

    Think about getting ECG’s in babies – they tend not to hold still, breath about 20x per minute (shifts the baseline), with heart rates of 120’s – 140’s range (depending on age) – and we may be looking at the morphology of the rising edge of the QRS complex, which lasts 75-80 ms (for the whole complex).

    Just perused PubMed, and found this article:

    Dickinson DF. The normal ECG in childhood and adolescence. Heart. 2005;91(12):1626-1630. doi:10.1136/hrt.2004.057307

    at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1769212/

    It mentions sampling at 1000 Hz, with 250 Hz BW, as minimums. The waveform look blocky even with the best traces you showed – not sure where that problem was from. I do see on the crowdsupply site that the.response curves seem to show a -3db BW of <80 Hz.

    ==================
    Hearing aids – Current hearing aids have much better battery lives compared to what you were mentioning. The ones I got last year (compliments of good military medical coverage) last me 10-12 days, averaging about 12 hours per day on a single 312 size battery per side (1.4 V, 150 mAh capacity). The hearing aids (Oticon Alta2 Pro Ti) have DSP signal processing, though i couldn't find exact details of what hardware it is running on. I'm assuming a custom FPGA.

    • You are comparing apples to oranges if you are trying to compare a $30 ECG module with a multi-thousand dollar commercial unit that also has the benefit of digital processing. I seriously doubt the market that this ECG module targets would include hobbyists if it was designed to compete on that level.

      P-wave morphology is best seen in lead V1 rather than lead II so I don’t understand why you would be scrutinizing it to the detail you are hinting at in the lead arrangement David was using. QRS morphology applies as well in terms of precordial leads rather than limb leads (which David was using).

      Sampling at 1Khz with a 250Hz bandwidth is used more to capture diagnostic quality data (such as pacer spikes or subtle conduction abnormalities) – again – not the intention and design of this device at the presented price point. It would be crazy to sell this unit for $30 with the promise of it being diagnostic quality as the required bare minimum component cost wouldn’t even be covered let alone all the certifications and insurance required in case it leads to a misdiagnosis.