Posts by Electronics'R'Us

The case for handcrafted software in a mass-produced world

Small...

In the murky world of microcontrollers and sensors there are often situations where an OS is highly undesirable. Bare metal applications abound in this (vast) world.

One of my projects was for a very precise 2 axis tilt sensor and the sensor was analog (MEMS devices drift a lot and quite quickly). The drive circuit was reasonably simple but to get the stability of readings some post processing of the data following the ADC conversions was necessary (think 64 tap or more FIR filtering).

The issue here is that the time between each conversion starting must be as close to identical as possible and interrupts is the way to do it; the kicker is that nothing must get in the way of those interrupts to introduce timing jitter on the conversions (even a small amount of jitter can royally screw up the filtering because the mathematics depends on the time between samples to be identical).

Add to that the issue of designing timing windows (sampling, processing, re-initialising data structures, reading the ADC data and hardware functions such as DMA to name a few) and it becomes clear that cycle accurate timing is required (for the sampling at least).

The best code in this situation is small and tight apart from the startup initialisation (where it doesn't really matter). No assembly was required (an accurate oscilloscope is your friend for this stuff). That means no layering of the code (or at least extremely little but none is best). Goodbye several deep layers for function calls.

In that particular case, everything was done based on hardware timers and interrupts that were provably not interfering with each other and results communicated upstream to an application processor (from a DMA buffer).

It might surprise El Reg just how many of those types of system exist.

Sorry, Moxie. Blaming Agile for software stagnation puts the wrong villain in the wrong play

I sort of agree with some of his points

And disagree with others.

In the original article, he railed against the use of frameworks [1].

I don't oppose frameworks in general but I do think they are overused or used when it is really not appropriate to do so. Do we really need an enormous framework such as electron for simple applications? Honest question.

When he was going on about understanding the underlying hardware (knowing what the computer is actually doing) I disagree. COBOL programmers didn't really need to understand what the underlying hardware was doing but when I am bringing up a single board computer, it is imperative that I thoroughly understand precisely what is actually going on.

Try initialising DDRx without understanding all the register settings (there are a lot) or initialising the DMA subsystem. That way lies not only madness but also a major loss of hair and perhaps the consumption of large quantities of 'refreshments'.

It is courses for horses, really. Not everyone writing code needs to understand the internal details of the parts we are using as it would have no measurable gain [2]. For those who do need too know, it is interesting that fully understanding a modern microcontroller from a family we have never used before can take several days or weeks depending on the architecture. There is typically 4000 to 5000 pages of documentation in total for a modern microcontroller.

1. The standard C library can be viewed as a framework (as can numerous others); they all provide an abstraction layer to a greater or lesser degree. Obviously C exposes far more of the hardware level than python libraries for example.

2. That does not mean I would not encourage them to look into what the machine is really doing, just that for many tasks it is unnecessary.

Google's ex-CEO U-turns after saying staff 'going home early' killed winning

Office when it adds value

I have no objection to going into the office [1] when it adds some value.

A lot of my work is design and analysis (see handle) and generally I am far more productive at home [2].

When I do need to go in (as I did a couple of days ago to decide just where to add some fixes / mitigations for EMC failures) there needs to be a solid reason. A few weeks ago I spent the entire week at the EMC test house (living out of a hotel room) to run the most risky tests (RE-102 and CE---102 for those who may be interested) so that if (when) there were failures I could analyse them and have time to develop a solution. Travelled home on the weekend.

Most of the time I spend the majority of my week in my home office [3] with the occasional day at the relatively local office.

No commute means I am far more relaxed and ready to go when I start the day.

1. $COMPANY has 3 sites of which I regularly go to 2, one of which is full day worth of travel so I usually stay for a week. The local office is open plan (yuck) so it is not the place to get things done that require concentration.

2. Design and analysis requires concentration rather than hanging around a canteen / water cooler / <hangout of your choice> and I am quite ok in my own company when I am doing that. If I need to chat to other members of the team that is simple (Teams video call - it works well enough for that).

3. I don't work excessive hours as that can easily lead to errors. At the normal end of the working day I shut down the work laptop and close the door to the office. Occasional late work is fine but it should noit be the norm. Being a bit disciplined in that is useful.

AI stole my job and my work, and the boss didn't know – or care

Re: Past parallels

After over 5 decades in the electronics and associated business (and still going strong), I have yet to find any form of automation that can replace a skilled, experienced, professional.[1].

I recently oversaw some EMC testing for some naval kit and, as always, there were some failures on the initial run. [2]. Understanding the root cause of those failures is a darker art than even being able to understand Intercal.

Some fixes are more obvious than others but the why and underpinning theory are often not easy to find. I might, just for a giggle, ask one of these 'miracle machines' what the answer may be considering it will have zero knowledge of the internals of this rather large, multiple box system. Oh - did I mention cable runs of up to 50 metres?

1. Many years ago (30 or so) I designed an automated test system for a product line. The rationale is quite simple in that touch time is more expensive than line time so it can make sense in some circumstances but you need to be doing sufficient volume to amortise the cost of the hardware and development.

2. Anyone that tells you their non-trivial system passed all EMC testing (particularly if you are testing against some of the MIIL-STDs) first time is probably, lets see - stretching the truth.

LLM-driven C-to-Rust. Not just a good idea, a genie eager to escape

Re: Sometimes you want it to be slow

Several years ago (well over 30) before the widespread advent of microcontrollers with timers and built-in peripherals I wrote quite a few bit banged UARTS.

That means counting machine cycles for every instruction to ensure each bit was timed properly with the judicial insertion of NOOPs to maintain the timing. I think the type of tool suggested in the article would barf on those.

Then there were the integer multiply and divide routines that had to have constant timing regardless of data (so run time was data independent); those got very interesting doing 32 bit stuff on an 8 bit machine.

So for situations where the timing really matters I am not sure this type of tool would be suitable (that's an understatement I suspect).

Car makers sold people's driving habits, location data for pennies, say US senators

Number 4

Several years ago, I designed an interface to vehicle CAN bus (through a FSM gateway so it was read only).

This was for a large waste management company. It would alert on harsh braking and acceleration (and send that data to a server) but I understand the company would give the drivers an opportunity to explain why.

The rationale for the company was to reduce maintenance costs (HGV servicing is an expensive proposition and brake replacement even more so).

Company vehicle so this type of monitoring, which had a reasonable goal, was legal as far as I know.

A couple of drivers covered up the alert light and speaker, which made no difference as the data went to the server anyway.

Inquiry hears UK government misled MPs over Post Office IT scandal

Please

Stop repeating incorrect history.

Horizon is an EPOS and backend finance system for thousands of Post Office branches around the UK, first implemented by ICL, a UK technology company later bought by Fujitsu.

The entire rollout of Horizon was very much driven by Fujitsu, who owned 80% of the company at that time. They even leaned on the British government to make veiled threats about economic problems should the Horizon rollout be delayed.

History of ICL: here [Silicon.co.uk]

Fujitsu leans on government: from Computer Weekly

Meta warns bit flips, other hardware faults cause AI errors

Re: I'm a bit out of touch with the hardware design

When I look at the probability of an error within, say, the ALU, I need to know how long the data will actually be there and likewise for a register. These are usually very short.

There is a statistical chance an error can occur due to various causes but it is way lower than the figures in the article. We deal with that by having redundant channels and voting (among other safeguards) in a safety critical application which they are clearly not doing here.

Now, if they are running the thing at very hot then the chance of error goes up, usually due to timing violations within the device various data pathways.

In communications theory, all data paths have a bit error rate; what that is depends on a lot of factors but for an internal datapath I would suspect it is in ppb (parts per billion) or lower provided it is being run at a temperature that does not violate the timing requirements.

Re: I'm a bit out of touch with the hardware design

We have been dealing with this in avionics for decades.

The usual requirement for a safety critical system is:

L1 - Parity protected

L2 - ECC protected

L3 (if it exists) - ECC protected

Main memory - ECC protected

The ECC used is 'correct 1, detect 2'; a single bit error can and will be corrected, a double bit error will be detected and trapped (critical even handler).

When it comes to free neutrons (the most common cause of single event upsets in avionics) the odds of more than one bit in a word being corrupted are extremely low (far lower than what Meta claims for an error rate). There are other causes but they can all be detected.

Thermal issues are well known as well. Transistor I/O structure characteristics change over temperature (the edge rates slow down as they heat up) and typical dynamic memory devices need to speed up the refresh rate at higher temperatures due to leakage. A temperature sensor and on-the-fly parameter settings is not difficult.

FPGAs with SRAM configuration are susceptible but there are solutions for that (redundant processing internally and partial reconfiguration is one such solution used in space applications).

So use systems that have the necessary error detection and correction. Job done.

There is a timing hit at startup as all memory to be used must be initialised with the ECC syndrome bits.

Asda IT staff shuffled off to TCS amid messy tech divorce from Walmart

They have already had a meltdown

A few months ago, Asda rolled out its 'new' payroll system (this was another move away from Walmart systems) and it was an utter train wreck.

Many of their staff (who are on the low end of the pay scale) paid hundreds less than they should have been and told, in typical tone deaf management style, that the missing money would be paid at the end of the month. That got rolled back pretty quick after the rather large outcry that ensued.

So they have already had one (ongoing, apparently) software rollout that was not just bad but terrible, particularly as this can really cause major screw ups to people's tax status.

Now they are doing a full blown ERP implementation? On the cheap and rushed at that.

This is likely to be worthy of the corporate equivalent of a Darwin Award.

Apple says if you want to ship your own iOS browser engine in EU, you need to be there

Re: I absolutely adored my Mac Classic.

The closed ecosystem is nothing new for Apple.

This is from a humorous article from 1995.

Mac Airways:

The cashiers, flight attendants, and pilots all look the same, talk the same, and act the same. When you ask them questions about the flight, they reply that you don't want to know, don't need to know, and would you please return to your seat and watch the movie.

Dear Stack Overflow denizens, thanks for helping train OpenAI's billion-dollar LLMs

So the problem will just get worse

A while ago (2019) a blog entry warned of vulnerabilities in code posted to SO.

SO Blog

Novel vitrimer plastics promise greener PCBs

Re: "First, heat gun? "

Repairs are not done with heat guns (at least, not in a professional environment).

Depending on the repair, it might be a soldering iron or an IR hot plate. For larger components, preheating the board can be required.

Directed hot air feeds are definitely used, particularly for larger components and BGAs, but referring to them as heat guns is not accurate.

Details?

For PCBs, the devil is in the details.

First, heat gun? That is perhaps true for hobbyists but the standard method for modern SMT PCBs is reflow. Peak reflow temperatures are in the range of 245C to 260C for a lead free process.

If there are some through hole components, then hand soldering or even selective soldering might be required.

Some more details that are rather important:

1. Coefficients of thermal expansion (CTE). The CTE below Tg (glass transition temperature) and above vary quite widely, particularly in the Z axis.

2. Tg. The actual glass transition temperature.

3. Dk. The dissipation coefficient. This varies with frequency - just how much depends on the particular variety of material being used (no, they are not all the same).

4. Moisture absorption. In a high reliability world, this matters (and no, you cannot perfectly seal a PCB that has mounting holes).

There are many other details but by now you get my drift.

UL94 V-0 compliance is also important.

Note that the term FR-4 does not have any meaning other than the material is flame retardent.

State-by-state is the best approach for right to repair, says advocacy leader

Most laws in the USA are State laws

This is something that many who have not lived in the USA fail to understand. This is the way it was originally set up and continues to this day.

To see the effect on right to repair, Colorado is not imposing their will on any other state, because they cannot. The kicker is that in most states a company that wishes to do business there has to have a registered office within the state [1], regardless of the State of incorporation and that means the business within that state must comply with all state laws that pertain to the business.

This does, admittedly, lead to a hodge-podge of varying regulations but it does have an effect.

So if a manufacturer of something (farm equipment for example) wants to have their own company branded sales and service operation within the state then they must comply with the right to repair laws within that state. In most cases, as noted above, that requires a registered office within the state. A business that refuses to comply with the rules will have that registered office closed.

Such manufacturers are perfectly free to not do business in that state, of course, but then they lose sales and service revenue.

What the state by state approach does is to impose the will of the local legislature (which is arguably closer to ordinary people) to business that operates within the state.

Messy? Yes. Effective? Yes.

As a rather larger example of one area passing regulations that affect places far beyond, take the RoHS directive (often known as the lead free directive) within the EU. Device manufacturers now almost universally provide lead free components [2] even within the USA despite there being no such USA legislation.

Note 1. Most registered offices are a local attorney's address.

Note 2. There are some exceptions, mostly military and aerospace,

The chip that changed my world – and yours

Re: It lasted 50 years, but history finally claimed it

The announcement stated that the supplier was unable to continue fabricating the part.

That is hardly surprising; it was on a 4 micron node (IIRC) and the fabrication equipment needs maintenance and spare parts which are likely to be totally unavailable now.

It is interesting how some parts live on where you least expect it.

The venerable 8051 is often found within a wide array of controllers (running at many times the clock speed of the original); I have seen them used widely to implement state machines in USB hosts. There are numerous 8051 devices in a weapon system designed in the late 70s / early 80s, incidentally. I suspect there are many such examples.

Many problems can be solved with an 8 bit device running at a few MHz, so it is hardly surprising (in that sense) that such parts are still widely available.

Australia’s spies and cops want ‘accountable encryption’ - aka access to backdoors

They still don't get it (in one sense)

Burgess labelled encryption “clearly a good thing, a positive for our democracy and our economy” because it “protects privacy, it enables communications and transactions.”

But he noted it also provides criminals with anonymity, which is why Australia has laws that make it possible to access encrypted messages. Burgess said those laws aren’t working well because tech companies aren’t helping.

Technology of all types, in and of itself, is agnostic; it is neither good or bad. The various use cases could be seen to be somewhere in that scale, though. This is not news

Hint to all the authoritarians out there; Pandora's box is officially open.

On the subject of encryption; if anyone other than the sender and recipient know the key, it is, by definition, insecure. It is not a matter of 'tech companies aren't helping' - a properly crafted encryption system means they cannot help.

Zilog to end standalone sales of the legendary Z80 CPU

Masks

I have never used Rochester for that (and I have used them a lot). I don't think they actually buy masks but buy up all available stock at LTB.

There are die banking services available where certain companies have relationships with the silicon vendor. This is very prevalent in avionics where kit has to be available for a lifespan of (often) 40 years or more.

Boeing top brass stand down amid safety turbulence

Over 25 years of rot

The (reverse) takeover of Boeing by McDonnell Douglas (in 1997) is well known (especially to the denizens of these parts) to have been the true watershed moment.

The writing was already on the wall, and then when the company relocated HQ to Chicago and effectively put the bean counters in charge, it was clear Boeing was no longer a proper engineering company,

The 737MAX problems were a symptom of this rot that has been in place for a long time, and I am not sure how quickly it can be removed (and it needs to be removed, fast). There have been a lot of comments on the issues at Boeing over various articles and I am not going to rehash them.

It is not just a matter of recruiting new C-Suite execs; the entire corporate attitude must change to regain their reputation.

An old saying, but true: It takes years to build a reputation and seconds to lose it.

Uncle Sam, 15 US states launch antitrust war on Apple

Closed ecosystem

I remember this from the 90s; the closed system by Apple is northing new (perhaps more extensive now).

Mac Airlines

All the stewards, stewardesses, captains, baggage handlers, and ticket agents look the same, act the same, and talk the same. Every time you ask questions about details, you are told you don’t need to know, don’t want to know, and would you please return to your seat and watch the movie.

Source

UN: E-waste is growing 5x faster than it can be recycled

Surface mount parts

There are a lot of very good reasons we use surface mount parts (including BGAs with thousands of pins).

Among those reasons are signal integrity which is far simpler when you don't have a big hole in the board, let alone a component in a socket (which is, electrically, a bit of a nightmare in this context).

That said, a group of repair hobbyists (or perhaps a repair outfit) could get the necessary equipment to repair such things quite affordably.

One thing that needs to be done is to educate those people on the risks of ESD (electrostatic discharge) to modern electronics; a very small amount of it can damage a modern microprocessor or microcontroller (so small that we would not even notice that we actually had an ESD event).

A major problem is parts availability; this is where otherwise perfectly good electronics that has one defective part has to be scrapped because the replacement part is not sold through regular channels (or is not even produced any more). This is, to a certain extent, driven by the planned obsolescence model.

Most failed parts, though, are not the highly integrated bits, but are more usually the various support parts, which can be sourced quite simply in most cases.

I know of some companies that are indeed repairing expensive controller boards (and the OEMs are aghast!); when a board can cost upwards of £10k (some are much more expensive) then there is a financial incentive to repair a unit rather than buy a new item. There is at least one company that sells (and train people in the use of) diagnostics and reverse engineering equipment to facilitate this.

The ability to repair needs to have an economic incentive, in my view.

Legal eagles demand $6B in Tesla stock after overturning Musk's mega pay package

Hilarious

"The lawyers who did nothing but damage Tesla want $6 billion. Criminal."

Pot, meet kettle.

FAA gives Boeing 90 days to fix serious safety shortcomings found in report

Re: The Golden Years

The last Boeing commercial aircraft that was designed and managed by engineers is probably the 777 (not the latest iteration of which horror stories about quality abound).

I know the principal designer of the flight control computers in that aircraft (and I was involved in the technology refresh some years ago). The Boeing 777 team was both difficult and excellent.

I mean 'difficult' in a good way as quality issues did not get past them.

The train wreck that is now Boeing took hold once the bean counters from MD took over; I think most of us in the industry saw the writing on the wall quite quickly and were confirmed in our view when the corporate headquarters was relocated to Chicago.

Mamas, don't let your babies grow up to be coders, Jensen Huang warns

Re: "Jensen Huang believes"

Well, I am an EE but I have done a lot of code for embedded systems of my (or my team) design.

That is an enormous market, incidentally.

Some of the skills are setting up the underlying hardware [1] which requires a rather solid knowledge of the internals of the microcontroller you happen to be using [2]. I don't see any 'AI' (which although artificial is anything but intelligent) being able to do that in the foreseeable future.

1. The are hardware abstraction libraries available, just about all of which are poorly written, are very opaque and have horrible corner case problems. Train the model on those and hilarity will ensue. I had a library function (DMA initialisation in one case) that was several call levels deep and I replaced it with 3 lines of code. The same issue exists for just about every onboard peripheral library function provided by the vendor. For precision work, using callback functions is a major no-no but they proliferate in the various HAL libraries.

2. Modern devices have several thousand pages of documentation without considering the underlying core and assembly language and it can take a long time to fully understand all the functionality and just how to invoke their operation (even order of operations can make a significant difference).

Persistent memory to replace DRAM, but it could take a decade

Use cases

There are a lot of use cases, but the majority of them are not in general purpose computing.

1. I have done a lot of avionics and the start up time requirements can be very stringent. Shortening that can be the difference between success and failure in getting the contract.

2. Edge / IoT. Many applications in this arena need to power down, and often only power back up on an event. Without needing to maintain power to the memory during power down, power is reduced. That can be a very important factor in the design of such things.

3. Where non volatile memory is required (there are so many use cases here such as 'cause of last shutdown'). In the past, one would use a Dallas semiconductor (now Maxim, which is now part of Analog Devices) device that had a small battery. Those things cost a lot, but with the newer non volatile devices, more mainstream approaches can be used.

Note that although these could use EEPROM / Flash, the write times (and erase for that matter) take quite a while; these things operate at standard memory speed and if you are in an unexpected power off event there will be time to do an orderly shutdown and write the fact of power failure to a log file and also give that an orderly shutdown - make sure the write enables are off prior to voltages going below required states. That prevents data corruption.

Will it make it's way into mainstream computing? I have no idea.

There are plenty of places where this stuff is used.

Forgetting the history of Unix is coding us into a corner

Also...

The Unix philosophy for the various parts were do one thing and do it well which is why things such as SystemD are anathema to many.

Dell staff not alone in being squeezed to reduce remote work

Hybrid

I work from home 4 days per week normally, and $EMPLOYER is perfectly OK with that.

There are 2 offices I can go to; one is fairly local (about 35 minute drive) but that is not the team I primarily work with. The one thing of value in that office is the cross-site engineering meeting where we all get together in the offices (which connect over teams). I find that particular meeting quite valuable because I am one of the design team and the local office is an integration team so we get a bit of a chat going.

The office where the team I primarily work with is about 250 miles away and I go there when it is necessary (new hardware commissioning, troubleshooting and the like) which happens between 4 and 10 weeks apart (it really is dependent on what is going on).

When I do go there I go for a week (all expenses paid) so I am in the office that entire week - that all seems to work very well. Quite a few of the team I work with are quite inexperienced and mentoring them is easier with a teams chat than in an open plan office [1].

Some of the things I work on (apart from designing electronics) are making templates for the relatively new ECAD tool to automate the outputs; that demands peace and quiet so I can properly focus and try things out.

This is a relatively small company, so we get to be flexible, although that does mean wearing a number of hats but I am comfortable with that.

This works for everyone, so why change it?

Note 1. I am not going into the problems of open plan offices but I detest them with a passion as it can be impossible to concentrate.

Dumping us into ad tier of Prime Video when we paid for ad-free is 'unfair' – lawsuit

Re: Question is...

Several years ago (not sure when but it might be 80s) I had a book of Punch magazine [1] cartoons.

One I will never forget had two advertising people; one was saying to the other:

"I really have to believe in a product before I can lie about it"

1. Punch magazine, for the uninformed, was a wonderful somewhat satirical publication that included such things as Spy vs Spy and Let's parler Franglais.

Drowning in code: The ever-growing problem of ever-growing codebases

In the embedded world...

Most of my software is written for small microprocessors or microcontrollers. Typical maximum flash is around 1MB and perhaps 256k RAM.

Even here, bloat has taken hold with HAL (hardware abstraction libraries) although it is not extreme; they are, however, opaque.

One project (using an ARM Cortex M4) required me to do rather interesting things such as respond to external events while the processor was in sleep do a bunch (1024) ADC conversions and DMA the results to as buffer. Only at the end of the conversions / DMA was the core woken up.

The library DMA initialisation function was 6 layers deep; once I figured out what was being done, it was replaced with 3 lines of code using the same structure the library required. The library function for the ADC / DMA did not solve an issue that the DMA would just run at maximum speed (and not just after each conversion). I made that work by forcing an arbitration after each DMA which was part of the initialisation code!

My point is that most HALs are poorly written and inefficient. For ADC / DAC conversions, in particular, timing jitter will render any DSP functionality fubared so 'trusting' the HAL to get it right is simply not acceptable.

Several years ago (decades in fact) I was tasked with adding a test for an optional flash device that could be fitted in a socket (they were expensive at the time). The system resources was 32K (not a typo) of ROM, of which there was perhaps a few hundred bytes free. I managed to get a comprehensive test (using existing flash programming functionality with the walking 1s, walking 0s technique) into 33 bytes.

I always look at 'frameworks' with a very sceptical eye.

Fujitsu finance chief says sorry for IT giant's role in Post Office Horizon scandal

Investigator suggested dropping wording...

From a witness statement as it could 'damage' the Horizon brand (might have been a lot less than it is now).

https://www.bbc.co.uk/news/uk-wales-68172203

Return to Office mandates boost company profits? Nope

Just reduced my office time

When I joined $COMPANY it was agreed that any office time would be on the basis of adding value.

A couple of months later this was changed to 2 days a week at one of the company offices (notably my direct team lead does not work at that office - his office is over 200 miles away); the reason was that people in the office I use when required were perhaps a bit jealous. That was recently cut to one day.

Three things are interesting here.

1. As my handle implies, I design electronics (among a number of other things such as low level software and process improvement - by which I mean making it work for engineers), which requires large amounts of peace and quiet for concentration (something not found in an open plan office).and my actual productivity went down. I can access the CAD licence and network from my home office (over a VPN) just as easily as I can in the company office.

2. In general, to talk to the team I work with I use Teams (I know...) even when I am in my local company office except for the occasion I go to that other office for a week.

3. The one day I go in actually does have value (an all site engineering update with the other sites on Teams).

Unsurprisingly, my overall productivity has gone up, I don't have the commute and the company can burnish its green credentials as my vehicle time is reduced (there is no public transport in the village).

The Post Office systems scandal demands a critical response

Re: It's still happening

The issue here is that a numerical overflow becomes possible.

'p' (the input variable) is probably ok, but when the compiler expands 2*p then it may not be ok for whatever type is actually being used.

Given that the function appears to be at the heart of the 'reverse transaction' process (which has been implicated in doubling the size of the transaction or otherwise not functioning correctly) such 'cleverness' is (as usual) not a good thing.

Re: It's still happening

Have you seen some of the code?

Take a look.

Computer weekly has an article about it, naturally.

From that article: “Whoever wrote this code clearly has no understanding of elementary mathematics or the most basic rules of programming."

WTF? Potty-mouthed intern's obscene error message mostly amused manager

Code comments

Some years back (about 10) I was doing the base code for a microcontroller with GCC in an eclipse based IDE. Think bare metal and device drivers.

The default for that was live warnings (equivalent to -wall) which can be useful, but when setting up DMA channels among other things (where pointers abound in the hardware) it can prove to be slightly annoying.

The DMA descriptors are in RAM, but the addresses of source and destination are in dedicated registers (as they are for many things - check out the memory map for any ARM Cortex device).

I would get implicit conversion warnings so the code was littered with <someregisteraddress> = (uint32_t*) myvariable where myvariable had been declared as a uint32_t (because what was loaded was a literal value).

At some point I got frustrated and added a comment along the lines of

// cast to suppress GCCs verbal diarrhoea

My then boss was not impressed so I had to remove it.

In all fairness, this code was going with our hardware to a university for their algorithm development so he had a point.

Epic decision sees jury find Google's Play store is illegal monopoly

Re: This is the mother of all Google trials

Many years ago when I lived in the USA, I had a girlfriend who was a lawyer.

When it comes to jury trials, she explained to me that the jury is the finder of fact and the judge is the finder of law.

For an appeal to overturn the jury finding, Google will have a very uphill battle as it will have to show that the evidence, as presented was incorrect or lacking. She told me that overturning a jury decision is very difficult.

To appeal otherwise they must show that the findings of law were incorrect.

Interesting times indeed.

FTX crypto-villain Sam Bankman-Fried convicted on all charges

Gold finish on PCBs

We have been using ENIG plating for decades.

The issue is the coplanarity of the finish. The previously popular HASL (hot air solder levelling) just isn't good enough for the vast majority of surface mount parts.

Some manufacturers (notably TI) have also provided parts where the lead finish is NiPdAu (Nickel Palladium Gold) for a long time; one major advantage is that this is compatible with both Tin Lead (SnPb) and lead free reflow profiles.

So it wasn't RoHS that pushed us into using gold on PCB surface finish - it was device geometry.

'Corrupt' cop jailed for tipping off pal to EncroChat dragnet

Official Sensitive

The security classification Official Sensitive is rather low and roughly equates to the old Restricted.

Access to that simply requires a BPSS (Baseline Personnel Security Standard) background check which is a very basic criminal records check. It does not require a SC (Security Check) which is required for those who handle Secret. Just what checks had been made on this person remains something of a mystery (there is a counter terrorism check which is not much different from BPSS).

That said, divulging anything protectively marked is a violation of the official secrets act which everyone is covered by. The typical 'they signed the official secrets act' statement means very little. Signing the form means you have had your responsibilities under the act highlighted. Even if you don't sign it, you are still (within the UK) subject to it.

I remember when newspapers would come onto a base and get stamped 'Restricted' Hilarious in a way.

Still, it is pretty dumb to take a selfie with a classified document (albeit of probably little intelligence value) clearly in view.

SpaceX accused of paying less to women and minority engineers

Pay in startups

Some 22 years ago I was in a startup near Philly (with all the attendant issues - long hours and the like but unusually we got 3 weeks of annual holiday and other benefits not normally associated with the US employment scene).

I was making over $120k then so it really is not an unusually large amount for an engineer.

Big Brother is coming to a workplace near you, and the privacy regulator wants a word

Re: "or offsite [...]outside work hours.

Even though I am back in full time employment, I typically work from home 4 days a week. There are times I need to go to the main office (a few hundred miles away) for a complete week. New hardware needs some TLC, usually.

At home, when it gets to 5PM, all the employer supplied equipment gets turned off and my office is not revisited unless I need to use one of my own devices to do something (print labels perhaps).

I am fortunate that my immediate boss understands the concept of 'I need to solve a problem and I need to think about it' so my 'productivity' is measured by results

I have heard tales from elsewhere (from those seeking to find employment elsewhere, usually) where things are not as civilised.

Dell reneges on remote work promise, tells staff to wear pants at least 3 days a week

No added value

The arbitrary 'n' days per week diktat is one of the most specious argument out there.

I will be starting with a new company soon.

There will be times I need to go to an office (one reasonably close, the other one a couple of hundred miles away - expenses paid for that one); we have agreed that there has to be value in actually going to the office (hardware commissioning, for instance).

Where there is no added value (teams call, perhaps) then there is no point in doing a 40 mile round trip. Bonus: they get to boast of their environmentally friendly policies.

II have no problem at all going to the office when it makes sense. That might be every day for a week and then not for a month.

The 'people work better in person' argument is also silly; people work better when they collaborate might be true but it does not always require physical presence. It also ignores the fact that introverts don't actually like being in a roomful of loud people such as marketing 'rah rah' types.

This seems to be a pure micromanager issue.

Microsoft cries foul over UK gaming deal blocker but it's hard to feel sorry for them

Not just crocodile tears

I think there are some real tears being shed (and much gnashing of teeth) at the missed bonus that several of the main people were expecting which were probably measured in 10s of $M and quite possibly far more.

Icon for Microsoft and Activision reaction.

Microsoft is busy rewriting core Windows code in memory-safe Rust

Re: "Oh no, not again!" said the potted petunia

"If you're building an application today that's either performance critical or low-level, then Rust is a no-brainer at that point."

Here we go again.

Depends on the definition of low level. I do a lot of bare metal stuff where dynamic memory allocation are very much a no-no [1] and for clarity C is still best for those applications. For arrays, it really doesn't matter if they are dynamically allocated or not as an overrun is an overrun regardless of how an array was declared.

I highly commend everyone to number 5 of the The Ten Commandments for C Programmers reproduced here:

5. Thou shalt check the array bounds of all strings (indeed, all arrays), for surely where thou typest ``foo'' someone someday shall type ``supercalifragilisticexpialidocious''.

I have picked up code directly driving hardware where no error checks were performed (see number 6 at the above) written by supposedly excellent software engineers; ion one case I added a status word to the code and as each part of the initialisation was done successfully, I cleared the appropriate bit. If everything went ok I got a '0' at the end; more importantly, I could tell exactly where in the sequence things had gone wrong by finding the first non cleared bit.

Rust makes sense in certain things (courses for horses and all that) but in many applications, C will remain on top. It has its issues but they can all be dealt with.

1. On some occasions it is necessary to use malloc() to get the alignment correct. I had that requirement for a DMA descriptor setup.

How Arm aims to squeeze device makers for cash rather than pocket pennies for cores

Device agreements?

How that would work looks something like this, we're told: Arm would still license its processor designs to chipmakers, but under so-called development licenses that require the chips to only be used by manufacturers that have device agreements with Arm.

Many ARM microcontrollers are sold (as parts) via distribution (as are a lot of other architectures); if the above is true, that would no longer be permitted.

Example: I design test interface equipment, among other things. Some of them are using a Cortex M4 device and I have perhaps 4 of them made once the design is solid. Would I (or $COMPANY) need a device agreement? It certainly seems so. How much would be the royalties? This stuff is for internal company use (never sold).

I realise Cortex M series might not be affected but the reputational damage from this would see me drop them.

What about the cores in Xilinx (now AMD) FPGAs? The Ultrascale parts have quite a few (some M series, some A series). I can't see them paying this rent.

Quite apart from the engineering nightmare, the administration of it becomes terrible.

There are plenty of alternatives that will do the task; one reason I currently use ARM is that free toolchains are widely available and parts are reasonably easy to come by.

When either of those is not true, when the interfaces get replaced a different family will be chosen as the codebase is abstract enough that only low level drivers would need to change.

I don't know how many controllers as a percentage of sales are sold this way but as a revenue stream it would get stopped in its tracks.

John Deere urged to surrender source code under GPL

Electronic Serial Numbers

The way JD (and others) lock 'customers' in is very probably by the use of ESNs.

These do have many legitimate uses [1] but what these companies do is a consequence of technology not being able to prevent bad actors from turning it to their advantage.

Here is how it can be done (which might give a clue on how to reverse engineer it).

Each sensor has an ID and an ESN (ESNs are usually 64 bit numbers)

The entire list of sensors (IDs) and corresponding ESNs are written to non-volatile memory (probably an EEPROM).

The list is read at powerup and compared with the enumerated devices - mismatch == you have top pay through the nose for someone to come out and...

Connect to onboard computer.

Get current list of enumerated devices (ensuring that only the ones that have been newly installed are permitted to be updated)

Write new list of devices to EEPROM.

Parasites.

If a kernel driver needed to be linked for the device enumeration process then there might be interesting things that could be done. For at least the last 20 years, manufacturers of the devices have provided equipment vendors to define their own IDs, ESNs and so on.

1. In safety critical avionics (as an example) the serial number of every board is tied to traceability information so that if faults occur more often than they should, the batch of components (and manufacturer) can be easily determined. There are many examples of reasonable use of the technology. These parts have been around in various forms for at least 30 years.

Most Londoners would quit before they give up working from home

WFH / Hybrid has advantages for more than just the WFH crowd

Some of the people I work with need to be at the office (production for example) and some prefer the office for practical reasons.

Young grads / apprentices who are either flat sharing or living with family come to mind.

One particular team I work with have a relatively limited space (it is a repair facility for electronics) and the WFH / collaboration space solution that $EMPLOYER has put in place means fewer people traverse their limited space which is a win for them.

I do go in occasionally for those times I really need to but the rest of the time I am working in my home office (which was designated as such when we moved here).

Right now I am in the middle of detailing the principles of operation of a piece of electronics I am designing; the process of writing that down typing that up without any interruptions or extraneous noise lets me concentrate better (especially when I discover a flaw that only appears once I type it up but I can fix it right away). The other piece (currently) is the actual layout which can really require multiple sessions of several hours uninterrupted work (typically 4 to 6 hours at a time) which is difficult, if not impossible, to achieve in an office (especially open plan).

I have a meeting with one of the electronics distributors FAE tomorrow for lunch so we are meeting at a nice little cafe about 10 minutes from here; that will probably be far more productive than sitting in the office.

I also do a lot of microcontroller development and the first parts of all those uses a development kit attached via USB; I can (again) get that done far more effectively here. Once I am ready to commit to a PCB I will go in to commission it, then bring one or two home (I have a nice little lab setup) to do much of the rest of the work. Final integration is at the office.

The flexibility is the key and employers who refuse to see that as an advantage (or perhaps are worried about their 'stature' as viewed by senior management) are already losing out.

I might go in 3 days or more in a row and then not see the office for 2 or 3 months. We have no problem with collaboration online (Teams, for all its faults, works quite well for that sort of thing). I just did a team call yesterday to explain the process of designing electronics (for a particular piece of kit) to 8 junior engineers / grads and that went extremely well.

Arbitrary 'you must be in the office coz reasons' is a good way to demotivate people when they get a lot more out of the job when they can just get on with it without a detailed plan for the day and just when they are mentally up to it.

New software sells new hardware – but a threat to that symbiosis is coming

Memory vs. Data size

There is no real hardware reason that a 32 bit processor have a 32 bit address bus. 8 bit devices almost universally had a 16 bit address bus. One practical reason is the size of the program counter (if it is 32 bit than the maximum size within a given space is 4GB) but that is an implementation detail.

There is no consensus on what determines whether a device is 16 bit, 32 bit or whatever. The two most commonly used definitions are internal register size [1] or ALU data width.

The PowerQuicc 3 series (from what was Motorola -> Freescale -> NXP) have an internal 36 / 40 bit address space (device dependent) and exposes a 32 bit address for non SDRAM memory but with multiple chip selects. Boot flash, static RAM and so forth come to mind. Peripheral mapping can also be done using that interface.

SDRAM of all flavours is a different type of beast as the address space can be much bigger than might be expected as there is a row and column value latched on different parts of a memory transaction. Thee data interface can be 72 bits (64 + ECC) for those devices (That has been true for some 32 bits devices for over 20 years). SDRAM (DDRx included) interfaces are always separate from the main address (physical) bus due to the hardware requirements.

It can also assign a 32 bit address space for PCI/PCI express interfaces and so on which can then map much larger spaces behind each device (depending on device).

So it wasn't just memory space.

1. Register access width, anyway.. Many internal special function registers, although accessed as a 32 bit device only expose a few bits with any meaning.

FTC floats rule to ban imposed non-compete agreements in US

I had a non compete

Several years ago - early to mid 90s, in fact. In Florida.

It was done properly (to start with). The parent company boss wanted to implement these, but I saw [1] the memo which clearly stated that existing employees (of which I was one) could not be bound unless something of value was offered. It was, in the form of a profit sharing arrangement.

When I left them (I got fired as it happens) I went to a small outfit designing the same type of equipment (smart payphones) and I got a snottygram from a lawyer [2[ which I asked a friendly labour law practitioner about.

It turned out that non competes were very limited (the judgment was in the Southern Law Review) and was limited to some very specific areas.

1. Specialised training or knowledge that the company provided that I would not have found elsewhere. No. I probably brought more knowledge in than they ever provided.

2. Customer lists / contacts that were made using company resources. Not applicable to me.

3. The use of internal IP. They had some but I wouldn't use it anyway for a number of reasons including my reputation.

They tried to intimate that a linear 2 wire to 4 wire conversion (necessary for remote updates using QAM modulation - FSK would work in an ordinary diode based circuit) was their trade secret.

So we wrote back explaining that the technology found in their products was (apart from an ASIC, the details of which I knew but was not using) decades old. At time, semiconductor databooks (remember those?) were brimming with linear 2 wire to 4 wire converter circuits (they wanted to sell their modem ICs), one of which I adapted. They had not provided anything special in the way of knowledge and I was not involved in direct customer contact and to basically back off.

Never heard from the lawyer again.

1. I don't think I was supposed to see it but it was left on the operations director desk with the door wide open. I had reasons to go in there regularly (mainly to leave memos that he often didn't read but I could say had been delivered)..

2. Said lawyer was a major investor in the company. I mentioned the conflict of interest in the response to them.

Man wrongly jailed by facial recognition, lawyer claims

Somme questions

Let me get this straight. He was driving on a highway in Georgia

A police department in Louisiana apparently had access to his driving licence photograph.

He claims he has never been to Louisiana, so it would not be a Louisiana driving licence .

From the article, it would seem that either Clearview AI or Morphotrak had this image (or the Georgia driving licence agency shared its database [1]).

So how could the Louisiana cops ID him from a different state's data?

1. Assuming it was a Georgia licence.

Corporate execs: Get back, get back, to the office where you once belonged

Re: Collaboration

I completely agree that the review process has to be more explicit. I do quite a lot of electronics for interfacing which can be 'interesting' on occasion.

Because of that we (the team) always ensure we have real requirements (which may include software) and there is an 'originator - checker - approver process.

When onsite everybody is spread out over a large area so the remote method is not much different from what happened prior to the mass WFH.

Collaboration

"How do we get them working on things together? I mean, remote is great, but when you have new and difficult problems, putting people inside of rooms is absolutely critical," he added.

I call BS on this

I work as part of a team of various functions and we have found that a remote virtual meeting for problem solving is every bit as effective as a physical meeting and without the commute.

All the 'offices' have been converted to (bookable) hot desk collaboration areas and although we are 'encouraged' to go on site regularly what matters more is getting the issues sorted. There are times when it makes sense to go onsite (commissioning new hardware for example) but I can do design work far more effectively here.

If I need someone's opinion on a part of the design I can just call them.

Some people need to be onsite (production teams for example) but those who can work more effectively from home are encouraged to do so.