Hacker News Reader

Hacker News

Comment by acapybara | Hacker News Reader

Parent Full thread

acapybara

Our Subaru has a physical volume knob.

Its position is only a "gentle suggestion" to the software to move volume up or down.

At times, especially when starting the vehicle, there is major latency in the software processing the knob movement..

Also, if the knob is moved too fast, the sw gets the wrong reading.

This results in volume being way too high for serveral seconds as well as many other frustrating, dangerous situations.

UX in cars made in the 60s were better than this.

I don't think slapping on fake physical knobs onto laggy buggy software is a solution at all.

View on HN

Clearly that is a problem with Subaru (and Honda from someone else?) and not physical buttons themselves. Just because some vehicles have a piss poor implementation doesn't mean the idea of physical is wrong.

I have a 2015/16 Ford and Chevy with the infotainment screens - both have physical buttons and both respond instantly to my changes. They also both have an additional volume control on the steering wheel itself which also responds instantly.

I'm not advocating everyone go buy Ford/Chevy by any means, but physical buttons - when implemented correctly - are superior to touchscreen controls.

My 22 Ford has problems with the volume. Sometimes it just keeps pumping to max and you have to fight with it for up to 30 seconds as it's blasting your ears out.

That's kinda the problem; these 'Infotainment' systems keep getting upgraded/etc for marketing purposes but don't have enough care put in the upgrades to make sure stuff works right.

"Move Fast and Break things" is not a good motto for anything that does in a car IMO.

The problem with cars is that’s it’s move slow and break things.

These comments with exactly the same problems I have are really making me LOL. Thanks. (I'm battling fever bouts post chemo so I need a laugh)

> and Honda from someone else?

fwiw my '20 Honda Civic doesn't have this problem. Volume knob feedback is instant.

> Just because some vehicles have a piss poor implementation doesn't mean the idea of physical is wrong.

Yes it does. We constantly ban entire classes of goods because some of them can be badly built or badly used by customers knowingly using them wrong. Sydney even banned a few hundred bars in 2013 because tenants were using alcohol then misbehaved on the public street.

Of course, a proven track of horrible implementations of OS for physical knobs should ban those companies from using them for 10 years. Come back to us when you can control how your gear behaves, we should tell them, this is the textbook implementation of no-upfront-regulation-until-you-misbehave.

This is great satire. I just can't understand how people go home at night knowing that this thing they built could easily have got right but they just got it so wrong. I think the big "system constraint" isn't so much the technical system but the business management system that clearly has to please "stakeholders" allows these flaws to keep continuing.

As a Sydneysider you know we like those shame pages for bad restaurants and the like, maybe we need one for crappy UX

i'm dropping my subaru for this among other reasons.

1. throttle response, move your foot to 20% and experience between no throttle and wide open as the 'turbos' kick in in my outback

2. every time we get near an intersection with bollards, it thinks it's a kid and slams on the brakes

3. you're not allowed to persist auto start disabling

4. you will be nagged for looking anywhere other than dead straight, especially if looking around a curve, you'll get beeped at

5. sometimes when driving the main 'touchscreen' just dies, takes about 5 minutes to reboot, and the entire time i have no climate or audio control, luckily the adaptive cruise control seems to stay working, you just can't reenable it...

Wow on 2-5, but I thought 1 was 'normal' for cars with a turbo charger?

Turbo lag is normal, but is is rarely more than a couple seconds. Even my F350 when pulling a heavy trailer I don't notice it - unless I'm in too high a gear for my speed. (I've only had a turbo with a manual transmission - I've driven some automatics that are slow enough to shift that you might think it was bad turbo lag, but those cars didn't even have a turbo)

I think it's mostly transmissions and other non-turbo effects that cause these contemporary lag complaints.

With all the fleet mileage requirements, there are much higher gear ratios for cruise and coast modes, so the engine can be bogged way down near idle speeds. Also, there are more software based throttle controls and interlocks. Some of this has to do with heavy-handed reaction to prior "unintended acceleration" panics, and some is just part of all the modern traction/stability control and efficiency programming.

The programming can get utterly confused and fail to dip back into a ground state where it is ready to accelerate. People doing various combinations of heel-and-toe or other transitions between braking and acceleration, including rolling stops, may get their car into a different mode than they would want, and the recovery from this seems like "turbo lag" or worse.

I wouldn't say that is a thing anymore since the 1980s. My old 2003 truck a 5.9l turbo diesel was quite responsive but at nearly 6 litres spewed a lot of exhaust to spin up the turbine. Smaller cars may suffer a bit more although some sporty cars may have two stages; separate small and large turbos. The small is good at lower rpm and the large better at high rpm.

I thought 3 was federal law and was universal amongst US cars.

I personally don’t mind it, but thought it was indicative of how unpopular that feature was when a software update moved the disable button to my Subaru’s dock.

FWIW my Subaru doesn’t experience the other issues mentioned besides 1 & 5. 1 is like any turbo I’ve driven before so maybe my expectations are off but 5 is crazy annoying.

Interesting. Out of these 5 issues I've only experienced the auto start annoyance on my Subaru crosstrek 2021.

Is this limited to their higher end models?

Surely 5 is a warranty claim. I actually went through 3 head units on my 2014 Subaru XV with a very similar sounding issue.

Subaru was awesome. The cvt, the ridiculous headsets, and thin paint mean this 18 xt forester is sadly my last one I think.

The 5 speed outback sport I had before was absolutely a joy to drive.

We just bought a 2023 6-speed Crosstrek and it’s amazing. Also, fewer (color) screens because it’s practically the base.

Sadly the last year one can get it as a manual.

I quite like the CVT, but yeah my Outback paint is pretty much not up to the task of protecting from the Aussie sun, the branches and nuts that fall out of the, and birds and bats that poop all over it. I actually like cars that look nice but I'm now almost at the state of mind as long as it stops the body from rusting it is good enough.

It’s not that I don’t like the cvt, but if it breaks I may as well buy a new car… my friends outback had its cvt self destruct and yeah, no thanks.

1. Subaru's engines are notorious for being low powered. A 2.0 with 152hp and a 2.5 with 182hp? That's... NOTHING. Even with turbos that is shamefully low. Pair that with a slow-acting CVT and you have a recipe for disaster.

I do wonder if the mapping of the pedal is direct. Because one can do it in 2 ways: First one being: putting your foot down 20% means open up the air intake valve and derive fuel requested from that. Or 20% is give me 20% engine power, which then can actually continuously remap what that means in terms of how much we open up the air intake valve and how much fuel we inject as the turbos spin up.

3. I have never seen a car that allows you to do it as the user. I program it in my BMW.

5. If the car is new enough I'd start the lemon process.

Huh, that sounds terrible. I got two Subarus (nether if them outback) and never had any of these problems. Maybe a bad model year or some controller issues?

This seems to be the head of the Subaru thread.

If you have any hope at all for Subaru to get better at software development, abandon it here. Subaru is a high watermark in my career for:

- "us-vs-them" fighting between "IT" and "Business"

- Siloing of teams within IT

- Team bloat / retention of non-operative team members

More please! Would love to know more

Here's some Subaru love from Damaskas Hollodan:

https://www.youtube.com/watch?v=MXYKVfoJ2Gs

I love the gentle suggestion. I'll use that next time when I try explain to my wife about how it does annoy me as much as it does her. I would have really hoped the engineers that made clearly poor decisions (I cannot believe it is a fundamental system constraint) were brought to reckon on it somehow. Obviously not enough people at Subaru (or their OEM) cared. And you can't even (or I haven't worked out how to ) just have the media start muted every time you start the car. Often for a trip you just want to few moments of quiet contemplation thinking about what you have forgotten to take with you - and not the DJ yelling at you.

I have a 2018 Outback for reference

You sort of hit the nail on the head there. That's not a physical button. It's a physical button tied to terrible software. The software is the problem, and if the button either changed the volume in an analogue way (ideal), or Subaru had engineers who were writing better software (both less ideal and much less likely to happen), there would be no issue.

I'd rather have good digital than analogue. The aftermarket head unit in my car has an encoder that responds instantly, and accurately. So I can either click it once for a small change or turn it quickly more for a big change.

I'd much rather that than an analogue potentiometer that is guaranteed to be scratchy in a couple of years. Also you won't get steering wheel controls to work with a true analogue knob.

I guess that's what we get when a rotary encoder is much much less expensive than a 4-6 gang potentiometer. And I also feel your pain with the car "booting up" phenomenon. I don't even have a particularly tech-heavy vehicle and upon starting the car the entire infotainment system feels like booting up a packard bell in 1996.

Is the problem that the same systems are used for car media and potentially other systems and that encoders "can" be more flexible if they need to be?

I can't recall seeing a single car that used a potentiometer for the volume control, and I've had cars from the early 90's all the way to today. They've all used encoders.

Do you have some examples of makes/models that used potentiometers for volume control?

> I can't recall seeing a single car that used a potentiometer for the volume control, and I've had cars from the early 90's all the way to today. They've all used encoders.

That's because the switch happened earlier than that. Go back earlier than the "DIN size" head units of the '80s and '90s to the "shaft" style radios with two large knobs flanking a center section with an analog frequency display and maybe some preset buttons if you're lucky.

> I guess that's what we get when a rotary encoder is much much less expensive than a 4-6 gang potentiometer.

Also a rotary encoder doesn't age in the same way as a potentiometer. A potentiometer can wear out the area being used most heavily while simultaneously developing oxidation on the areas not being used. Eventually this leads to crackling, dead areas on the dial, and other misbehavior.

A rotary encoder on the other hand doesn't wear out in a practical sense. The only part that even could wear is the bearing or bushing supporting the rotating assembly, and if that's specced appropriately for the application it's effectively a lifetime component. It's possible to build a crappy rotary encoder that falls apart earlier than desired, and of course they can still be damaged by abuse, but a well built one should outlast the useful life of the device it's installed in by multiple orders of magnitude.

Relative rotary encoders are much cheaper than absolute rotary encoders, and I expect that's the problem here. If the encoders were absolute, one could make them work just like a potentiometer (because potentiometers are absolute). Relative encoders cannot remember where they were last set because they only sense dp/dt, rather than position itself. So it's up to the software to remember the last position, and everybody knows car companies won't pay software engineers tech company salaries, so by definition car companies get B-level and C-level programmers, and the driver gets weird misbehaving audio in the car.

While you are correct, if the knob was only used for a single task, most car stereos I've used in the digital era give multiple roles to any knob(s) they may have. In the common configuration with one knob on each side the default modes are usually volume nearest the driver and either manual tuning or rotation between presets nearest the passenger, but if you go in to the menus one will usually scroll the menu while the other one changes settings. You don't want to have your volume or station change just because you needed to tweak a setting.

Old school wire-wound potentiometers are nearly indestructible.

Rotary encoders do fail, they produce the effect of control being unresponsive or “reversed”

Yes, but they don't really "wear out" to get that way, just some debris ends up blocking either the sensor itself or some amount of the openings in the wheel. It's a random event rather than an inherent result of age or wear. It's also often possible to fix simply by cleaning the debris out. Unless it's been physically damaged it's unlikely to need replacement.

The problem is not a rotary encoder, but how the software handles the signals. That's what interrupts are for, the command should make it into the software queue basically instantly, ensuring commands aren't lost.

Shut up with your Silicon Valley nerd talk!

Good old reliable polling. That's where it's at. 2 second response for a volume knob should be quick enough for anyone!

Oh great, why don't you just implement automotive user interfaces in PHP? ;)

"I'm not a real programmer. I throw together things until it works then I move on. The real programmers will say "Yeah it works but you're leaking memory everywhere. Perhaps we should fix that." I’ll just restart Apache every 10 requests." -Rasmus Lerdorf

"We have things like protected properties. We have abstract methods. We have all this stuff that your computer science teacher told you you should be using. I don't care about this crap at all." -Rasmus Lerdorf

You can't really use a user facing pot for volume control when you've also got volume control buttons on the steering wheel and keywords for voice control.

Well you can and some kinds of high-end audio equipment are actually built to do this by putting servo motors on the pots so that settings can be stored and retrieved automatically. But that's probably too expensive for a car company to consider.

Packard bell comment made me laugh :)

On my BMW if I move the volume knob 180 degrees to the right spread out over 10 seconds it will increment it by x.

If I do the same movement but in 1 second it'll only move up the volume by x/y where y > 2.

That doesn't explain the initial latency. It's probably being posted to a queue which no-one is handling.

Same with my Hyundai. Volume knob is unpredictable above a slow pace.

It's a great car otherwise, but that makes the flaw stick out even more.

Last year's top trim Tucson was the most uncanny valley of cars I've ever driven. All the controls are there, and sort of work, but everything is slightly off and inconvenient in an almost imperceptible way. That and it managed to break GPS with Android Auto, so plugging the phone in resulted in a worse reception and the car frequently lost itself, jumped to parallel roads and rerouted.

Chiming in with a Toyota Corolla with a physical volume knob.

Most of the time it works fairly well, I can perceive some latency but I'm pretty sensitive to it -- my wife can't tell at all.

But when the car is first starting up the infotainment is clearly overloaded and the volume knob is suddenly a gentle suggestion. This is particularly upsetting if it decides to switch to a different input like Bluetooth then starts blasting at the last volume level of that input, and it takes a good 2 seconds to respond to the counterclockwise turn input.

My 2015 Ford Fiesta had a more reliable volume knob, and it was as bargain bin as they come (manual windows!).

sounds like a different type of hell, didn't realise it could get this bad!

Thing is... most of these are quite small bugs. The software development team working on this stuff for just a week or two more would have fixed 80+% of these.

Problem is they probably had project managers breathing down their necks about "don't spend time working on non-release-blocking tickets".

Why did the project managers introduce the bugs in the first place?

good i hate this so much, same in a honda odyssey

Reply on news.ycombinator.com