Back then, the good stuff might have been good because the motor was big.
The modern good stuff probably has a brushless motor that can prevent stalls. It doesn't matter if it's big or small, the computer won't let it overheat, so they can make it smaller to save money that they can spend on whatever part is the actual failure mode.
The trouble is finding manufacturers that actually do that, rather than just using a tiny motor with no computer and plastic gears with no torque sensing. If you design things with old fashioned methods, you need to use old fashioned heavy parts.
Nylon gears in contact with ozone, and switches and connectors of all kinds seem to be the problem on everything I've seen. Which is why I always say I would gladly use 10k lines of code to save one moving part or one switch.
- Mechanical: "Here's the physical specs, no need to account for A, B, or C, as it's been handled in software"
- Electrical: "Here are the physical constraints and electrical spec. No need to account for X, Y, or Z, as it's being handled in software."
- Firmware: "You need to handle A, B, C, X, Y, and Z, but if you run out of time or don't have the budget to make up a test rig for validating all those cases are handled correctly, no worries we'll just ship it anyway."
You'll see stuff like a lack of battery level monitoring, or relays without current sensing.
I'm not sure why we don't do things the other way, have firmware people present at every step. Some things can't be fixed later.
If mechanical puts a weak link gear designed to fail before the motor, firmware might not have enough margin to not wear it out with auto-stop.
This creates a whole new reliability problem. Software reliability is notorious for its interface failures and there can be many more of those modes (where hardware typically has a few wear-out style failures). I think about my relatively high-end TV that seems to always struggle with software updates without any glitches. It seems like we've just been lulled into accepting software glitches as the price of having software control everything. Luckily, most of the software is not safety critical.
I think this is one of the dangers of relying on software to close reliability gaps. Yes, you can change it quickly and cheaply and it's versatile, but it's often very hard to identify all the failure modes.
For non safety critical stuff it's much better to crash every week than to wear out in 5 years.
Maybe I'm becoming a curmudgeon or maybe it's my bias as a former reliability engineer, but the fact that we've normalized this drives me crazy. Imagine your frustration if your car needed to be "power cycled" on a regular basis just to keep it moving forward, yet we've become accustomed to that in our day-to-day interactions with electronics.
It's the bane of my existence that I have to continuously power cycle a device to do something seemingly simply like change a tv streaming service or scan a document. Interface failures are rampant.
you'd be surprised how often watchdogs get their GIGO conditions triggered and things perform a cleanup reboot while you drive
But I've seen so many things fail due to bad switches, connectors, and other simple parts, to me it's a great tradeoff to occasionally reboot something as opposed to having it totally fail at some random time within a few years.
I wouldn't trust my life to it, but I'll trust my watch party to it.
Plus, a consumer now only needs to power cycle or factory reboot, so untrained people can keep stuff going.
I'd much rather they need reboots than need skilled work like they used to, that just wouldn't be practical in a digital first world.
Oddly enough having failures be so normalized makes them less problematic, because the victim doesn't get many complaints. People who have to wait just assume that's how it is, nobody will tell you you shouldn't have used a computer for that
Before all these complex systems we used pen and paper and things subject to human error, and we had to deal with that quite a bit.
Computers suck compared to what they could be but they are amazing compared to a lot of older stuff.
My SO convinced me to get rid of my old laser, USB interface printer. That one always worked but was old. My new WIFI enabled one requires logging into an app to use, and the USB interface maybe works 10% of the time. The app and hardware constantly require power cycling before it works. Many times it never does.
One thing becomes apparent in quality engineering: often, adding more components/interfaces undermines reliability because there are often exponentially more failure modes. It feels like we’re being trained to no longer expect reliability, and the beneficiaries are not the consumers but the companies that create unnecessary interfaces to monetize data.
If you, say, use a microcontroller to soft start a motor instead of making it bigger, or use a capsense pad instead of a switch(Yes they can be obnoxious, but at least they last!), you are unlikely to get into a permanent bricked state because you're not writing to flash at all.
Your failure modes might increase in number as software complexity increases, but for the most part they stay ephemeral.
I'm not sure exactly why experiences are so varied on things like smart TVs and WiFi printers. I have the "Smart" version of almost everything, and it usually works very well, but some people seem to have generally bad experiences.
I think some of it might be because they only test on fairly generic consumer setups with the latest stock Android, no firewall, no VPN, Bluetooth never turned off, no custom DNS, gmail for the email account, etc, and more tech savvy people tend to have a few things that modern tech doesn't like.
It always seems like people who tinker have problems other people don't. I also notice iOS users seem to have trouble with random gadgets and peripherals not being too happy, but a my iPhone using friends aren't a big sample size.
If I was designing a product, I'd definitely be more careful, but as a user at home, I tend to avoid anything heavily customized, so I don't usually see as many of the bugs.
First MS Windows is the only way to get Netflix and similar to stream the 4k HDR Dolby Atmos stuff outside of smart TV applications. It will silently stream the lower quality without telling you. So if you do not tinker, you must use the smart TV crap to get any chance of watching content at the capability of your hardware.
Then comes the bugs. Randomly the TV and Sound system decide they cannot transfer sound anymore. It's just as if it is on mute.
To be clear the connection is pc -> tv-> sound system. Using 48gps HDMI cables.
The only solution is to unplug both the TV and the sound system. Reboot is not enough. Unplug one is not enough. It has to be both.
Btw if you do not have the right HDMI cable you will also get lower resolution and refresh rate. Without any recourse. No information reported anywhere until you learn that pressing 7 times the green button on the LG TV remote finally gives you the details on the active HDMI connection.
Oh and by default the sound setting of the TV will also decide atmos and convert it to 5.1 PCM before forwarding to the sound system, while you really want a passthrough to the sound system.
Basically nothing really works. So if you do not tinker I am not sure you even get the advertised quality and features. With that said ignorance is bliss.
These are relatively common phrases now that would have baffled anyone 30 years ago.
They sent me a replacement that works great and is entirely ignorant of the outside world (and will stay that way.)
Or if there's an activation step they never planned to need recovery from failure of?
If you have the hardware programmer you can most often update the flash directly.
I don't expect everybody to have all the hardware programmers on hand. But I would gladly go to a repair shop to get a quick firmware flash to unbrik some device. Of course this is probably not worth it for manufacturers. It's so infuriating because most device even have a relatively standard programming port already (though you need to open open the device to access it)
If products would all just use a Wemos D1 or something on their board, none of this would be an issue.
But sadly software defined products need(or think they need) proprietary firmware and Arduino isn't set up for strong DRM.
And the thing is, somehow every vendor seems to skip out on ensuring a working out-of-band upgrade process - despite U-Boot bringing everything you need with it to provide recovery via whatever means you want. Ethernet, USB, a serial debug port... hell, I did A/B partition support and bootloader checks with a watchdog years ago, hacked together in an afternoon but able to first attempt booting from the fresh firmware, if that failed three times it would attempt the B partition containing the old image, and if that failed it would attempt booting from a recovery partition that was intelligent enough to bring up the base board with enough networking to be able to bring up a web server and accept a signed firmware image.
Or, if that's too complex, a fucking button that the user has to hold and it will boot into the recovery image directly.
It's not rocket science, it's not even much effort, all you need is a fucking week at least and a couple of cents for a switch. Pure penny pinching, and I hate companies that produce hardware that can turn to e-waste because of it.
One can still hope.
Oh yeah. Like Surefeed smart cat feeders. They don't use end reed or contact switches to control the lid motor - they use current sensing on the motor and stop it when the current surges due to reaching the end position. The problem is, they also use cheap-ass plastic gears on an apparently pretty strong motor... my old Surefeed from five years ago is still going strong, but the newer revision (which is, admittedly, easier to clean) failed after not even a year because the motor had shred the gears and could no longer detect if the lid was fully closed or opened, making noise for so long that the cat refused to use it.
Fucking hell, these things cost 150€+ apiece. Two contact switches and wiring would have cost them less than ten cents. Now they have to pay for the cost of handling the warranty claim... but I'd rather like to have a well-designed product in the first place.
(Oh, and it doesn't like rechargeable C cell batteries either, because they apparently saved the dollar that would have given them a boost voltage regulator, so you're stuck buying disposable alkaline batteries)
They probably wouldn't even need a reed switch, just a wad of foam for some buffering plus a more sensitive stop algorithm. Or better yet, don't give the motor enough current to shred gears even at full stall.
I've got two sure microchip cat doors. They've been reliable. Same problem, hard to clean. When one of our cats figured out how to break through the door sure sent me a 3d printed plastic part that clipped on over the latch. I think their customer care is pretty good.
The new one isn't much better tbh...
> When one of our cats figured out how to break through the door sure sent me a 3d printed plastic part that clipped on over the latch.
I mean, my cats are smart enough to open doors, cupboards and drawers, but how on earth can a cat break through a cat flap?! Brute strength?
Sintered metal gears, last only a little longer than plastic and then fail catastrophically destroying half of the device so that you can't repair it. But at least they are much cheaper than proper machined gears.
After disassembly, it turned out to be perhaps the only plastic gear in the thing. It occurred to me that it was perhaps designed to fail rather than the mixer continuing to break the bones of the person who got their hand caught in the paddle.
So, I throw the engineer a bone for that one. I was bummed though to have to order a replacement gear from some obscure place in Florida for like $75 or whatever.
I bought an official replacement from KitchenAid for $12 including shipping.
This is a good call out of potential other reasons an engineer team might make something cheaper & easier to break. It would be awesome to read a summary of the trade-offs the teams discuss in their process.
($YOU may be. I'm not.)
Except that nowadays I tend to just guess the latter and I'm statistically correct anyhow.
Superman has a quote about living in a world of cardboard, where he must control his every move lest he rip through everything. I'm beginning to feel that way myself with these consumer products. If you take them even 5% out of spec expect them to catastrophically fail. They'll catastrophically fail even if you use them entirely within spec. I'd happily pay a bit extra for something that wouldn't but there's no way for the manufacturer to successfully convey to me that this is such a thing even if they wanted to. The metaphorical water is so metaphorically muddied that it's just dry dust now anyhow, with fake 5 star reviews and decades of refinement on marketing messages about how this vegetable peeler peels vegetables so well that it will even enhance your marriage prospects.