Fail of the Week: Dave Jones and the Case of the Terrible Tablet

Nothing spices up a quiet afternoon like the righteous indignance of an upset engineer, especially if that engineer is none other than [Dave Jones], on his EEVblog YouTube Channel. This week [Dave] has good reason to be upset. A viewer sent him what looked to be a nondescript 2010 era tablet from a company called Esinomed. From the outside it looked like a standard issue medical device. Opening up the back panel tells a completely different story though. This thing is quite possibly the worst hack job [Dave] (and we) have ever seen. This is obviously some kind of sales demo or trade show model. Even with that in mind, this thing is a fail.

wtf-solderThe tablet is based upon an off-the-shelf embedded PC motherboard and touchscreen controller. [Dave] took some offense at the hacked up USB connector on the touchscreen. We have to disagree with [Dave] a bit here, as the video seems to show that a standard mini-b connector wouldn’t have fit inside the tablet’s case. There’s no excuse for the USB cable shield draped over the bare touch controller board though. Things go downhill from there. The tablet’s power supply is best described as a bizarre mess. Rather than use a premade DC to DC converter, whoever built this spun their own switch mode power supply on a home etched board. The etching job looks good, but everything else, including the solder job, is beyond terrible. All the jumps and oddly placed components make it look like a random board from the junk bin was used to build this supply.

The story gets even worse with the batteries. The tablet has horribly hand soldered NiMH cells shoved here, there and everywhere. Most of the cells show split shrink wrap – a sure sign they have been overheated. It’s hard to tell from the video, but it appears as if a few cells have their top mounted vent holes covered with solder. That’s a great way to turn a simple rechargeable battery into a pipe bomb. Batteries can be safely hand soldered – Radio Controlled modelers did it for decades before LiPo cells took over.

We’ve all hacked projects together at the last minute; that’s one of the things we celebrate here on Hackaday. However, since this is a commercial medical device (with serial number 11 no less) we have to stamp this one as a fail.

73 thoughts on “Fail of the Week: Dave Jones and the Case of the Terrible Tablet

    1. An company I used to work for fell victim to this all the time. One of the worse was a dip coater that was 80/20 rail a stepper motor, threaded rod and an arduino, a few buttons, and a screen in a project box, for…$5000. They also bought a faraday cup for $800 that was to heavy for the scale they were using, so I made one out of a few soda cans and ptfe. Lab equipment is a good business to be in.

        1. The biggest problem is the vicious grants cycle. Research grants often go to what is ‘sexy’ or in vogue instead of the more left field interests. Therefore the breakthroughs are rarer. This is compounded by the requirement for researchers to constantly publish because this is the metric they are compared against for promotion, regardless of quality or social relevance. Outside of the grant system there is only business to fund research, because for some reason people with 8+ years of education also want to eat. Business research isn’t always bad, look at what Bell labs produced…..

      1. Considering a B2B hourly rate of $200 (that’s a concentrative estimate), that’s 25hrs for production, testing, shipping and support. Actually not too bad considering what you’ve described.

      2. lab equipment indeed is a good business especially if the technician are not too smart. a company tried to sell us a “camera adapter” for 3k it was basically an aluminium tube with no optical element whatsoever… shameless… thankfully we were able to gracefully refuse their “offer”

        1. A CE certificate is basically “I solemnly promise…”

          You can slap a CE sticker anywhere, and basically lie that the product conforms to EU directives, because there’s no checks until someone launches an investigation.

          1. Yup – Bit of a hit with garnet paper, some flux, and a good 40 watt hammerhead iron, and you can build your own stick packs without tabs. I wouldn’t trust my life to them, but I regularly would trust a few hundred dollars and countless hours of stick building model airplanes.

        1. Why does no one mention that it’s an incredibly bad idea to solder directly to the bottom of a battery? There’s a reason why the tabs are spot-welded on, the whole “do not expose to direct heat” thing that’s usually printed on the side of every battery…

      1. When not printed in all caps on a case label it seems to be capitalised as eSinoMed. e for “Electronic”, Sino for “Made for the China market”, Med for… medical. It’s an entertaining coincidence but I don’t think it’s anything more.

  1. the company name: Esinomed is the reverse of Demonise whawhaaaa
    with a lot of good will maybe in the best case it can read as demo-nise (as demo nice)
    But it reads more like demon-ise

    1. Could be. But are hospitals or whatever allowed to use non-approved people for repair? There’s lots of extra safety rules on medical stuff. That’s assuming it was used on patients, and not just some bit of research equipment that didn’t directly affect human life.

  2. Looks like a prototype or demo or early rev of http://wp12160106.server-he.de/esinomed/en/medizin-displays-pc/med-op-stationen/index.html

    I suspect the prime driver here were various compliance issues. These devices have to meet requirements about how easy they are to wipe down with disinfectant and so on. This explains the disinterest in jacks and connectors. Or holes of any kind. Obviously they were still working out overhearing issues.

    So, yeah. The kind of good engineering we want takes not just a back seat, but isn’t even in the vehicle, when compliance issues are the only thing you think you can meet by a deadline.

    I’ve worked in the factory and automation, medical devices, and auto business on the software side. It’s a very strange world where the obvious is never obvious and sometimes bad is good.

    Compliance and requirements lead to some interesting unintended consequences, sometimes.

    1. Can you give an example?

      I fail to see how your single example of wipe-down\clean-up would drive the design choices seen in this device, be it a prototype or otherwise.

      Regulations are there because business men will murder people for $, then call it a accident.

      1. I’m not sure what you want an example of. This device was clearly intended for a market that required a sealed case. Whether or not this specific device was ever intended to to be used in a real environment, it certainly looks like they were hacking together a device that was trying to follow that basic design of something that, once in production, would be easy to move around, not require space around it for heat management, and easy to disinfect.

        Given a short enough delivery date (for a demo or a prototype or (heaven forfend!) the real thing, who knows — we certainly don’t in this case) and other constraints, almost everything else will be dropped to make the basic requirements so that the device design can make it to the next step. It is more important to make a meeting, even on short notice, than to have well engineered circuits. The thing just needs to turn on, at most.

        I’m not defending this, only saying it is a fact that something has to give, and if the choice is delivering a shit prototype that meets the basic requirements (it may not even need to actually do anything much at the beginning) and not shipping a shit prototype, the answer is clear.

        Regulations and standards are not just CYA words. They define levels of compliance that regulate how you can even get into a room with stakeholders in order to have a conversation about future potential sales and deliverables. And the next meeting and level of compliance. Your demo does not have to meet or exceed the criteria. But it must show the direction you are going in to meet or exceed those criteria.

        The fact is that we don’t know what criteria this thing had to meet, but it has all the hallmarks of a hastily put together idea that hits some number of sometimes non-obvious criteria for the purpose of showing up at a meeting. But what is likely is that this box would never have had to run for long, if at all. Obviously, it *is* overheating because the batteries are exploding. And it is also obviously not a production run or even close to being production ready. It looks just like a rev 1 alpha suitable for initial stakeholders to point at while discussing powerpoint slides.

        I mean, I’ve been there, in one of those meetings, demoing a shit product that everyone knows is shit. The intention is that as you go through the process, more criteria are applied until you meet or exceed process, government, regional application criteria.

    2. The autopsied device would be a fine example of how NOT to design a wipe-don\clean-up…the textured surface would be difficult to clean and the screen is not flush with the case, forming corners and ridges that are perfect breeding grounds for evil things…

  3. He didn’t say if this was used in the field or what its origin was, just that it came from some guy in a mailbag. It could have been a mockup or a repair. For example why would anyone stuff in unconnected batteries for free? This Dave Jones seems to have a problem with beaking off before getting all the facts. Watch his other video called “14yo Hobbyist Arrested For Bringing DIY Clock To School”.

    1. This thing is full of terrible design choices even for a spit and bubblegum prototype. Dave is right about the amateur hour engineering, which speaks badly of the company. I don’t care if it was an internal prototype and wasn’t meant to be seen by the public.

      1. You’re also going with the same bad assumption as Dave Jones by saying this “speaks badly of the company” when you have NO evidence that the company did this at all. Also, replay his exact words 3:17-3:22. Yes, we all have eyes. But making the assumption that this was delivered to a client? That’s another baseless assumption. We have no clue of the origin of this. Like others have said here, it could even be a hoax. Anyone can wire something up and send it to Dave Jones for his negative commentary. Like the clock kid, we should not jump to conclusions before all the evidence is made available.

  4. Dave has his own maybe a bit expressive way to say things but i have to agree: This is a fail. Serial number 11 and medical device with such batterie und power supply is just awful. And the case doesn’t loke like it is easy to clean too.

    If the company still exists somebody should write them and ask for information about this, even if i don’t think they would respond…

    1. I started my last product’s serial number at 101. There were only twelve (never released) until a jump to 201 and actual production. Around 278 or so it jumped to 301.

      Serial numbers mean whatever you want them to mean (despite the literal interpretation of the phrase), and there’s no telling what “serial number 11” actually means.

      I’m not saying this isn’t a terrible design, but it could still just be a poorly designed demo unit.

    1. None of their products look like this, there’s a chance (even a good one) that some one got an embedded touch display from an existing product and hacked it into a x86 tablet.
      I did the same with a Point of Sale display, the display was working, it could be wired to a USB connection so I hacked it into PC to control gate / garage access.
      Throwing a cheap single board PC (some x86 VIA crap) in my case was an easy solution because you can simply hack something into it without having to worry about low-level coding.
      In my case simply using a .NET application with a web browser controller for the video feed from a Wifi D-Link camera and a couple of buttons that triggered relays over the COM port.
      Did it needed x86, windows xp and .NET for this? nope, but I wasn’t going to find a way to relay a video feed over anything other than a standard PC, even if i had time to spend figuring how to do it on a DSP or an embedded CPU why would I even bother?

    1. From the description: 800MHz AMD Geode NX800LX Nano-ITX Motherboard.

      I would guess 2007-ish vintage.
      MediaGX/Geode, Via, and Atom Mini-ITX and Nano-ITX X86-based SBC’s were once (relatively) inexpensive low power “fanless” embedded board choices before the era of Arduino and Raspberry Pi. They’re now a mostly forgotten page in history.

  5. Looks like a “Can we cram all this into a box this size?” extremely early prototype for starting software development. Could be a thing they hacked together as a trade show demo to show where the company was going with a near future product.

    “Hey, that pile of crap got us a large pile of orders! Now we can afford to make it a real product!” Esinomed might be pooing a brick over this leaking out, making plans for what to say to any investors that might question their product development process. “It’s a proof of concept we put together for internal use to see if it was practical to put the technology into this form factor.”

    Sort of reminds one of the original Macintosh hardware http://www.folklore.org/StoryView.py?project=Macintosh&story=PC_Board_Esthetics.txt

    1. Indeed, this smells to me more like a heroic last night effort to save an important product demonstration rather then a polished product that would be sold to hospitals.

      It’s 2008 according to the latest sn.
      The PCB looks crisp and well designed but (very) badly soldered.
      Regular of the shelf components like a powerbrick and a std usb to mini usb cable were’nt available to the builder.

      p.s. that guys voice is grating on my nerves.

      1. This has serial number 11 on it though. That would imply this is a terrible 11th iteration or an even worse last minute bodge on a series of products. As for the PCB, why bother with a second layer when there is only two traces, the logical thing to do is run wires instead and save the effort.

        I have made last minute rush jobs and quick prototypes, it is not difficult to make clean connections & secure things properly. It would actually make it easier because im sure they spent another day trying to troubleshoot that board.

  6. I think people are overreacting a bit. These are probably proof of concept devices to see if there is any interest. These would be given to sales guys to take to prospective clients and see if they would be interested in something like it. Just needs to work well enough to get the idea. I have seen so many mockups crammed together like this, it is not that unusual.

  7. My experience in this kind of ‘cludge’ is that a lot of these ‘prototypes’ are nothing more than a scrap-bin raid while teaching/entertaining school kids that intern for a few weeks.

    In the UK we have ‘work experience’ where a kid requests to work with us for 2-4 weeks.
    As many things are far over their heads, we’ll find something fun for them to do and take home at the end.

    This looks exactly like that – a few old bits in the recycle bin jigged together. They learn to build a simple PSU, some basic soldering skills. Grab a panel here, returns or demo unit there. Voila, a toy.
    It’s even possible a non-hardware guy hashed it together for a software test using such scraps, knowing just enough to be dangerous.
    In my current place, the work experience and apprentices get to play with the tech we later will R&D with, so they get to do a lot of unboxing and plugging in.

    In dev i’ve worked prior, we would upgrade to the latest prototype or use release models, rather than keep the old unit, as revisions or build quirks are worked with, rather than discovered in clients devices.
    Often the old prototype would be gutted and the case reused for another project, so you could have a mish-mash of gear with screen-printed model numbers, but flipped over would have another model written in marker – annoying when you think you’ve picked model X but is enumerating as model Y.

    1. I’d say you would probably do on par with this except on the batteries, they generally turn out like this, It would depend on equipment and prep though.
      Looking at the solder job, it has poor fillets on the board, looks like a few (are not) cold solder joints.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.