Remember the chip shortage? We sure do, mainly because as far as we can tell, it’s still going on, at least judging by the fact that you can’t get a Raspberry Pi for love or money. But that must just be noise, because according to a report in the Straits Times, the chip shortage is not only over, it’s reversed course enough that there’s now a glut of semiconductors out there. The article claims that the root cause of this is slowing demand for products like smartphones, an industry that’s seeing wave after wave of orders to semiconductor manufacturers like TSMC canceled. Chips for PCs are apparently in abundance now too, as the spasm of panic buying machine for remote working during the pandemic winds down. Automakers are still feeling the pinch, though, so much so that Toyota is now shipping only one smart key with new cars, instead of the usual two. So there seems to be some way to go before balance is restored to the market, but whatever — just call us when Amazon no longer has to offer financing on an 8 GB Pi.
The long, sad farewell to the Mars InSight lander continued this week as NASA released the spacecraft’s final selfie. Once you look at the picture, the eventual cause of death for the mission is pretty clear — or rather, really dusty. The whole vehicle is absolutely caked in dust, and with its twin 2-meter diameter solar panels obscured, it’s only a matter of time before the batteries can’t be charged anymore. InSight’s team is preparing for the end by making sure every last bit of data is gathered and downlinked, even while they massage the power management systems to keep the lander’s seismometer running as long as possible. The team is even knocking down the mockup lander ForeSight, which was kept on a model Martian surface and used to plan moves of the robotic arm before sending instructions up to InSight.
We’ve been following Lufthansa’s off-again, on-again relationship with Apple AirTags with some amusement, mainly because it seems way more likely that the airline is trying to manage perceptions of its luggage handling prowess rather than mitigate the risk of a CR2032 coin cell bringing down a flight. And while Lufthansa finally relented, it seems that other airlines are now interested in alienating their customers too. Air New Zealand just banned AirTags in checked baggage, although curiously their version of the TSA hasn’t been told to remove any of the devices it finds. And just to give you an idea of why airlines might actually be doing this, check out this story about some AirTag-equipped camera equipment that was allegedly stolen from checked baggage and tracked to a private residence in Alaska.
And finally, we featured a really cool engine simulator a while back that really seems to have caught people’s imagination. That’s understandable, because as cursed as the internal combustion engine may be, there’s no denying that they can sound really, really awesome, and the simulator was geared to reproducing the sound of various engines based solely on their fluid dynamics. We have to admit to not really grokking the whole thing when it first came out, mainly because it was just too much fun to play with the simulation; we spent way too much time trying to get it to reproduce the chirp of the four-cylinder, air-cooled, horizontally-opposed Volkswagen engines that were our introduction to automotive mechanics back in the day. But now, the simulator’s author, Ange Yaghi, has dropped a follow-up video going into quite some detail on how the simulation works. It’s worth watching, even if just to see his homebrew physics engine going through its paces.
Yeah, there is just too much weight/cost to include a brush and motor to sweep dust off Martian solar panels. It’s much cheaper to end the mission and ask Congress to fund the next mission.
Guess they’ll have to install a feather duster on the next one?
could’ve had brushes attached to the back of them and then just close and reopen to clean them.
They were opened by a spring loaded mechanism getting released. They can’t be closed. It’s also doubtful a few brushed would help much with such heavy statically charged dust.
If the dust is heavy and static is an issue they need a large nylon brush like a snowbrush used here in the midwest for very heavy snow on cars, also make brush either with a electrostatic ion gun (zero blaster) built in or some radio isotope in it like radio shacks old static eliminator device. and no excuses if power comes from solar electric, 1. overbuild with enough panels to compensate for 50% dirty power drop, or PLAN A WAY TO CLEAN THEM EVERY DAY OR SO! (bet they would love a windex wipes on a roll!!!!! Cleaning is NOT OPTIONAL FOLKS rover power cells get DIRTY!! DUH !
I would assume the dust would also get into the motor and jam things. Best to just send a human there and have it wipe it clean.
More like “somebody at NASA ran the numbers and found that any system they could build to clean the solar cells would be too expensive for the benefits it might bring.”
Mars isn’t like your kitchen floor – or even the roof of your house.
A brush in the “hands” of the robot arm sounds simple, but is it really?
– You have to clean the brush after using it. How?
– You have to be sure that the robot arm won’t glitch and poke a hole in the panel.
– You have to be sure that the robot arm can stand the stresses when reaching back towards the rover instead of reaching away from the rover.
The rover missions have all outlived their design goals by years – without having a way to clean the solar cells. The missions are designed with a specific minimum life span and specific goals for science and exploration. Anything above that is a win.
The chip shortage has eased for some parts, and especially for parts on the most advanced processing nodes as the demand for PCs and smartphones has dropped. But there are still shortages of many other kinds of chips, including SoCs for embedded systems; those are built on older process nodes such as 40nm or 28nm or 14nm, not 4 or 6 or 7nm like those PC and smartphone processors. Excess high end capacity doesn’t help the makers of those chips, and a further problem is that chip fabs are planning to sunset some of the older process nodes. (TSMC, for example, plans to discontinue their 40nm process.)
The Raspberry Pi makers can’t get enough of the Broadcom SoCs used on their boards. (The only exception is their RP2040 microcontroller that is used on the Pico, which is fabbed by TSMC on 40nm; the successor to that part will probably have to move to 28nm.) Broadcom only has so much capacity, and making Raspberry Pi SoCs is a side project for them, so they’re probably keeping as much of their limited capacity as possible to making higher profit products.
The popular STM32 series of Arm microcontrollers is another product line where shortages continue to hold back manufacturers. Lead times of a year are still being quoted for many of their popular parts. Although there are other suppliers of Arm microcontrollers, switching to a different product is not trivial; those other parts may use the same ISA but the peripherals are very different.
Thanks for the additional information.
It gives a wider perspective on the industry.
There’s a separate problem that if you have recurring orders for a part, and you’re seeing the backlog stretch out, you order more to make up for the backlog, which is a positive feedback loop that results in driving huge increases in production to make up for the apparent backlog, and the result is that suddenly you have a vast excess of production because you’ve filled the actual order depth and everyone cancels their excess orders.
This is a long-standing issue in electronics and probably every manufacturing process. It’s exaggerated in electronics because the lag time in bringing new manufacturing resources online is measured in years.
I quite like the various STM32-parts and so it has been quite disheartening to see their prices just go right through the roof, if they’re available at all! I’m a hobbyist, so I can’t afford to buy a lot of 10 000 pieces at $30/pop, nor would I have any use for that many anyways.
Yes, the chip shortage is by no means “over.” Some parts may be available now, but there are many, many more that are still not available for the next year or so from the delivery dates I’ve seen. As a working RF Engineer designing boards, it is still very much my main problem to find parts or re-designing existing boards to use parts that are currently available. I couldn’t believe how bold the news article was to say it’s “over.” And yes, ST ARM Microcontrollers are near impossible to find. Go to any distributor web site and the truth is evident.
It’s “over” in the same sense the pandemic is “over”.
stm32 (f4 and f7) are most of what quadcopters use for flight controllers, these days. prices have doubled or more and no sign of coming down. I have boards from a few yrs ago that i’m trying to not crash (literally) so i can keep them alive as long as possible.
We use microcontrollers from three different manufacturers (Atmel, Renesas, ST) and experience shortages for all of those. Depending on packaging earliest deliveries are quoted as 2024.
It doesn’t help that the auto industry uses parts on process nodes that chipmakers are abandoning in favor of more profitable smaller process nodes and are unwilling to change their chips to something more modern…
More modern doesn’t always mean better.
For automakers, the figure of merit is generally robustness. Temperature range, noise immunity and robustness against damage by electrical transients are all superior at the larger node sizes. And this goes beyond mere node size. Automotive and other enhanced-reliabiligy parts require different layout rules, because wide temperature ranges imply significant change in material dimensions (changing capacitance, among other things). Electrical transient immunity also comes with its own requirements in terms of area of conductor, capacitance between conductors and hard lower limits on semicondutor contact patch sizes.
A smaller node is generally incompatible with these requirements.
Sadly, simply using a larger node doesn’t provide the necessary robustness, either. It also requires specifically-designed design rules and lithography. Many of the fabs working with legacy process nodes aren’t working with automotive-grade design and layout rules, and switching their operations to make automotive-grade chips would take nearly as long (and cost nearly as much) as building a new fab. If this wasn’t the case, most of the automotive-grade parts shortages would never have occurred.
God I love this blog
That engine simulator is truly something! The comments on the youtube videos are telling.
It does seem to be quite centric on 6 cylinder and above simulation… I wonder how it would do with single and twin cylinder sounds, such as motorbikes? What would be interesting is a simulation of the unusual firing pattern (from what I’ve heard – no pun) of the ‘harley davidson’ twin cylinder… would it sound real? Could it even do it (it seems hard-configured for numerical division of 720 degrees divided by the number of cylinders)?