With the rapid obsolescence of technology, replacement boards will be difficult or impossible to obtain in the future and nobody will know how to repair/rework such ancient electronic tech by then either, making for a lower initial cost and feature laden but disposable product. Yet unlike with LEDs and computers, it doesn't make sense to upgrade appliances just because newer tech models come out.
You're assuming that the controller board is the weakest link in a new appliance. I can tell you otherwise. Modern electronics generally become obsolete long before they stop working. I have digital alarm clocks over 30 years old still working! If properly designed and built, the controller board will outlast the mechanical portion of the appliance it's in. The failures some talk about are due to this stupid obsession of getting the price on appliances you might buy once or twice in a lifetime as low as possible. Refrigerators cost the same in actual dollars as they did 30 years ago. Obviously something had to be compromised for that. A person in the business mentioned to my sister that today's washing machines, air conditioners, and refrigerators probably won't last 10 years whereas the older ones are still going strong. We're still using a 35 year old washing machine and 30 year old refrigerator. The problem isn't voltage spikes destroying controller boards, it's planned obsolescense. Appliances break because they're designed to on the assumption nobody wants them for more than a decade.
Analog displays are superior because you do not have to read them! They are best for conveying info at a glance where the precise value does not matter (eg when you only need to know needle in red zone = bad). Digital readouts are far slower because you need to read the number on the gauge, then interpret what the number means and make a judgement call on whether that is OK or not. This is the case no matter how large the display is or how elegant it looks. The other problem is digital cannot show rapid changes in a way that can be comprehended easily--a digital speedometer will necessarily jump 4, 8, or 12mph at a time when accelerating or decelerating rapidly, and what information does such a blur of numbers really tell you? It's the same with an analog multimeter, which is better at conveying simple "high" vs "low" or changing values rapidly because the exact value does not need to be read to do so.
I could understand the value of analog for certain things where you're basically scanning for good/no good, and perhaps a few values in between. Speed isn't one of those things. If you want to know your speed to within 1 mph then digital will be faster to read than analog. I cycle and I find that to be true as I remember well the old days of cable-driven dial bike speedometers. If you need to know speed to within 5 mph it's probably a wash. And as for acceleration, a properly designed digital speedo won't jump by large increments. There are three ways to design a digital speedometer. All three obviously count pulses. The oldest and most primitive way simple counts the number of pulses in a given time period. The standard for most cars is 4000 pulses per mile. So if you want to display 60 mph that means the display must count 60 pulses between updates. 60 mph=4000 pulses per minute. So this means updating the display every 0.9 seconds. Obviously this method is unsatisfactory since you will in fact see the large jumps you mentioned as you accelerate. This method was the one used in the 1980s, and the reason why digital speedos were much maligned.
The second method which came into vogue next was PLL (phase-locked loop). Basically you run a variable-speed oscillator, divide it by some factor, and vary the frequency so the divided wave matches the frequency of the pulse train coming from the car. The display methodology is the same as before-count pulses in some interval and that's the speed. The advantage here is that instead of 4000 pulses per mile you can get any number you want. This means the speedometer can update much more often, and can even display speed to tenths of a mph. The drawback is it takes time for the PLL to lock on to the frequency of the pulse train from the speedometer sensor, so when speed changes rapidly the display might not accurately track speed. This method is great for things where speed changes slowly most of the time, and OK (i.e. no worse than the first method) when speed changes rapidly. It's still used on some bike computers (and OK for that as speed when cycling seldom changes by more than 3 or 4 mph in a second, if that) although method three, which I describe next, is becoming much more common.
The third method, which didn't come into vogue until we could make small microprocessors fast enough, is to simply count the time interval between pulses, and then compute the speed. This method is very flexible (you can program the number of pulses per mile, and get speed displays accurate to the tenth of a mph). More importantly the display can be updated with every pulse (in other words tens of times per second), and be accurate at every update (even in a rapidly accelerating car the speed changes very little in 1/60th of a second). Therefore the display tracks the speed exactly. Granted, if you're displaying to tenths of a mph those numbers will be a blur but you'll still be able to clearly follow the 1s digit and especially the 10s digit even if accelerating rapidly. And if you're interesting in gauging acceleration accurately, you can put a separate bar graph for that above or below the speedometer display, perhaps reading from -10 to 10 miles per hour per second (or even in g's if you want to be cool). Acceleration bar graphs have been used on railway vehicles for quite some time. I'm quite amazed they haven't found their way into cars.
How many times do you need to know how many tenths of a mph you are going anyway?
All the time in cycling as it lets me know if I'm accelerating/decelerating slightly or not. In fact, it's necessary if I want to consistently maintain a set speed (and no, you can't track accelerations that small on an analog speedo, especially with the needle bouncing around on bumps). And in driving if you want to sit right on the limit without exceeding it, what better way than a digital speedo displaying to tenths of an mph?
I'll also add that I love the odometer capabilities of modern bike computers even more than the speedometers. You can usually set the wheel circumference to millimeter resolution. Since a bike tire is typically 2100 or so mm this means if you measure accurately you can set circumference with an error of no more than 0.5 mm of the actual circumference, or about 1 part in 4200. So you can measure miles with an error of no more than about 15 inches per mile. Or put another way, the cumulative error when riding a century would be half a city block or less!
And virtually all cars use digital speedometers nowadays, what with the OBD2 vehicle speed sensors (VSS) being digital. Only the display is analog, the better for us analog humans to read.
I know that. The problem is the analog speedo itself is basically an analog voltmeter which converts the pulses into an analog display. As such it's subject to the same inaccuracies and variation with time as any analog instrument. And the needle needs to be damped to keep it from fluctuating wildly at every small bump. That means it takes time to settle into the new reading when the speed changes (in other words, it's not even that great at tracking acceleration, ostensibly one of its advantages). A properly calibrated digital speedo will be just as accurate 100 years later provided the tire size has remained the same. And some cars are finally using radar-based speedometers which avoid the inaccuracies of tire-driven based ones. Railway locomotives have been using those for at least the last decade. My main point here is that not only are analog displays inherently flawed if you want precise readings, but the fuzzy readings they give can be off by 5%, 10%, even more. I don't know about you, but if I'm driving (I don't BTW but I'm saying if I did) I'd want to know that if the speedometer says I'm doing 110 mph then I really want to be doing 110 mph, not 96 or 117 or even 110.5. Stick to the analog displays for things like temperature or oil pressure or even a tachometer where the exact value is unimportant, but give me a digital speedo any day. And you do know you can display a dial-type speedo with a supplementary exact digital speed on a display screen, don't you? This gives you an analog display, but with the inherent accuracy of a numeric display. Why isn't this used more instead of actual dials?
BTW, I've been an electronics engineer for over 20 years and I have yet to need or want an analog type voltmeter. I can interpret what a DMM tells me well enough to not need to see a needle bouncing back and forth. And if I need to go further than that, I hook up my scope to see the actual waveform (yes, that type of analog display definitely has value).
Getting back to the main topic, the problem isn't so much computerized appliances and digital displays as it is one of
poor design. Analog designers have had how many decades to get it right? I'll bet in ten or twenty years nobody will be complaining about computerized appliances, and nobody will be signing the praises of dials any more. The controls will be both intuitive and highly flexible, taking into account things learned over several decades. Just look at what's happened with operating systems from the old DOS days of command line interface as an example. A modern GUI O/S offers ease of use for those who just want to use their computer, but also infinite customization options for the more technically oriented.