“You can’t control what you can’t measure” goes the old chestnut. But that’s a little bit negative, in my opinion. Instead, think of the benefits of sprinkling sensors around everywhere: you gain insight where you simply didn’t have it beforehand.
We were thinking about this in the context of the recent video on pressure advance in 3D printers. Essentially, the unmelted filament acts as a springy piston, and that springiness means that the pressure built up in the melted plastic lags the feedrate of new filament. We usually calibrate this out with a guesstimate constant, but it can be different for every different filament. Measuring that pressure directly with a strain gauge in the hot end makes more sense.
But then there are knock-on benefits of having a sensor in the hot end. You can use the strain gauge as Prusa does to run the nozzle gently into the bed and set the z-axis height. Or you can use over-pressure as a sign that the nozzle is clogged. It’s quite possible that you can use it to signal other things that can go wrong as well, but you can’t tell until you put the sensor on in the first place.
Of course, you don’t want to put a pressure sensor where you want to know the temperature, or vice-versa. But as a general rule, the more you can measure, the more you can discover about the way your system is running. How many strain gauges are too many?
Much of the time that goes backwards: you can’t measure what you can’t control. Identifying causes from effects and all that.
Easy to measure items that are not controlled. That is a way to find out if it is controlled or how good the control is.
i feel like the weather is something i can measure but cannot control
You can at least control the weather at the sensor by choosing where to put it. Otherwise you wouldn’t know if the result is representative of the weather or some local variable.
“in my opinion. Instead, think of the benefits of sprinkling sensors around everywhere”
look at what that concept has done to the car industry. There are sensors for just about everything now, and when one gets a bit wonkey it puts your car into limp mode. Don’t even get me started on all the CAN bus stuff when your CAN bus taillight shorts out and brings everything on the BUS down.
https://en.wikipedia.org/wiki/Goodhart%27s_law
Rings surprisingly true in many circumstances. For example, in automobiles, someone figured “Let’s put switches under the seats and a light in the dashboard, so people are reminded to put on their seat belts.”
Then someone else came along and went, “Well we got to make that light beep and flash, because people might ignore it otherwise.” Then another one came along and said, “It’s still too easy to ignore, let’s program the car so it won’t actually go faster than 10 mph if somebody hasn’t got their seat belt on.”
And so, from a friendly reminder that you should wear your seat belt came an obtrusive anti-feature where setting your shopping bag on the passenger seat sets off blaring alarms and disables the entire car, by people whose singular concern was that you have to wear your seat belt no matter what.
What you measure, you eventually want to control.
Control, of course is by itself a neutral concept. It needs a modifier:
When I control my stuff, more is better.
If someone else controls my stuff, less is better.
Let’s not forget that control implies force, yet force without distance goes nowhere. You can have all the control you can muster and get no effect, like trying to argue with a HOA about when you’re allowed to mow the lawn (never) and how long the grass should be (short). In such cases, losing control may be the saner option, or at least preserve your sanity.
But then again, force over distance equals work, so when you actually manage to exercise control for effect it comes at the expense of spending effort. Therefore, in both cases, the less control I have to apply, the happier I am.
I’ll add that the security of the sensor is also important. If the sensor can be hacked to create specific situations and actions it is not useful. Returning to CAN bus, some cars were forced to open the locked doors by lighting a fire under the car or by connecting to the CAN bus by smashing one of the lights. Also some big american truck was disabled when water damaged the stop lights and messedup the electronics (search HaD).
I believe that the sensors controller should be a restricted one and able to send only specific measurements and not commands, and the main controller should verify what values is receiving.
Efficiency is fragile – perfection has no room for error.
here’s a little story about hacking the outputs of an automotive oxygen sensor to make it fake the behavior of a working catalytic converter, all it took was two resistors and a capacitor, and after that the check engine light blinked off and the car (a Honda Civic hatchback) passed inspection:
http://asecular.com/blog.php?090314
The irony:
https://www.seatbeltextenderpros.com/end-ding-seat-belt-alarm-stop/
Suckers. Just reach under the seat and unplug the switch.
Works for driver’s seat anyway. The passenger side doubles as a weight sensor for the variable-strength-explosion airbag on that side, so you can’t disable that one quite so easily without the airbag system throwing an error.
Yep. More modern cars throw a fault code if you unplug the sensors, so the easier option is still to put in a dummy plug until someone actually needs the seat belt.
Point being, the system is a completely disproportionate response to the problem of someone not wearing their seat belt. If they want to die in a crash, that’s their own fault, and the fault of the driver for not checking. But, because someone decided it’s a good thing to at least remind them of their safety, the point kept escalating towards enforcing the case to the point of forcing the car owners to bypass the safety to avoid the constant harassment and annoyance of it.
Replying to Dude. It’s not really the case that “If they want to die in a crash, that’s their own fault” as a result of not wearing a seatbelt. When your car stops suddenly, such as in a crash, you don’t. Your body becomes a meat missile travelling at the same speed the vehicle was. Unrestrained you can now cause serious damage to other people in the car (who may have wisely been wearing their seatbelt, but its effectiveness has been rendered moot), or you exit swiftly through a nearby window and cause death or damage (not necessarily your own) outside the vehicle.
Wear your seatbelt.
That goes doubly on a plane. I do not want to get up close and personal with your corpse if we hit clear air turbulence. Kthxbye.
Yes, hence “and the fault of the driver for not checking”. If the seat belts are not on, the car should not be moving, but the responsibility is on the driver – not on the car. The moment a car becomes your nanny that keeps you from killing yourself out of stupidity is the moment you shouldn’t be driving a car.
i have the seatbelt fastened in the passenger seat of my subaru forester at all times for this very reason. my dog neville loves to ride shotgun, but then his weight triggers the sensor and i have that obnoxious beeping
Yeah, except no vibration sensor for bearings, so you wouldn’t have to guess.
It’s hard to isolate vibration for where it’s coming from, and what level or spectrum of vibration is actually indicative of a fault before it actually becomes a problem. Placing sensors everywhere might tell you that your wheel bearing is going out, but by the time it becomes unambiguous you’ll probably hear and feel it yourself.
Don’t know about the US but here in Europe we have this thing called roadworthiness inspection where every year a licensed mechanic does all sorts of tests, inspections and computer measurements to check if vehicle is… roadworthy.
If you do this in a reputable workshop, mechanic will give you an early heads up about things that will need attention within a year or two.
I don’t know why you replied with that, but I can tell you from experience that these licensed mechanics let a whole lot of shit pass if they like you and your car… if they don’t, they’ll find a reason to flunk it. There’s always a reason.
But the problem with automating the measurement still stands: just because you can measure some vibration doesn’t mean there’s a fault developing, or what kind of a fault it is, because there’s no stable baseline to compare with and it’s all case-by-case. It’s actually extremely hard to differentiate until the problem becomes so severe that even the driver can tell there’s something off with the way their car sounds and behaves.
But I like being grumpy online. I’m not trying to prove you anything about myself.
When I was looking to buy newer second hand car, I was specifically looking into reliability and repairability, including what cars do with canbus etc. A colleague had an Opel he got rid of out of frustration. If the front bumper sensors are disconnected, his car wouldn’t start anymore and had to be reset by a dealer. He couldn’t put LED lights in his tail lights because of CAN. His car wouldn’t start anymore. His radio had a fault which resulted in his cruise control not working. It’s infuriating. I bought a first gen Toyota Auris. Was almost new. Almost no canbus. First thing I did was replace the radio. If I had to replace this car I would buy the exact same make and model.
a sensor is not the interlock it controls, nor the redundancy of the transport it’s readings use.
I wish there was an ADD sensor, would make it easier to navigate social situations 😂