Why do people not like Microchip?

[Simon]

Microchip seem to like shooting themselves in the foot. They abandoned their IDE and are trying to make people abandon the atmel IDE. MPLABX is a massive nightmare. It feels like no two installations will ever be identical and it constantly demands to have you update little packs. It does not ooze confidence. Their code configurator in MPLABX is awful, I won't even touch it, atmel studio has a code configurator that looks much better but clearly they intend to just abandon it, links to documentation don't work. So I am using a basic ARM chip because it is all I could get hold of and writing my own initialization code. I'm only using a microchip/atmel chip because it's what I could get hold of at the time, I did not choose it because I like their idea of not support.

Their code configurator indeed isn't particularly great but it is an absolute dream compared to setting up a simple GPIO pin in Zephyr

believe it or not not all of us are pathetic teenagers that cannot tell the difference between their arduino or an OS application and embedded programming where you have to actually use brains, skill and some experience to produce something. I won't use configurators, the day they fuck it up I am fucked! Why hang my career on a bunch of assholes like those that mismanage at microchip?

[AaronD]

believe it or not not all of us are pathetic teenagers that cannot tell the difference between their arduino or an OS application and embedded programming where you have to actually use brains, skill and some experience to produce something. I won't use configurators, the day they fuck it up I am fucked! Why hang my career on a bunch of assholes like those that mismanage at microchip?

It's a good way to get started when you don't have a clue how any of it works, and you have to understand most of it all at once before you can be sure that something isn't waiting to bite you......provided that the tool actually matches the product and doesn't have bugs in it.   

But yes, as you gain understanding, I think you should get away from the automatic tools and write the init code yourself, manually.  The better you are at doing that, the better you can make the chip work for you, instead of the other way around.

I write this as a decades-long 8-bit guy trying to figure out a Raspberry Pi Pico (RP2040).  Everything looks to be spelled out in the datasheets (several of them, which itself is new), but there's enough going on that I think I'll be back to the library functions for a while.

[Simon]

The code configurators setup the peripherals, if you don't understand the peripheral then either you won't be able to use the configurator at which point part of redocumenting the train crash that the datasheet is is to write that init code. Some peripherals are rammed with subtle options that may not make it into the configurator or without thorough understanding of what you are doing are wasted on you, or your money was wasted on the chip. Sometimes you need to do something a bit special that no configurator can account for.

Because no manufacturer gives a shit about helping you and just wants your money they suck you in with the peripheral setup, confuse you with higher level functionality that need not be tied to the hardware and before you know it you are using just another arduino.

I've been working away at a samc, yes it's a pain, but the code configurator will not teach you the pitfalls, you either do it one way or the other, they have little to do with each other unless you go digging in the code to find out how it works but then that is not why you are using a configurator.....

[T3sl4co1l]

believe it or not not all of us are pathetic teenagers...

Y'all right there pal?


For my part, code generation is fine.  As with anything, it's how you use it, don't assume it knows what it's doing -- or that you know what you're doing, for that matter.   If in doubt, come up with a test to figure it out.

And never lose sight of what code is.  It's for humans.  (As unbelievable as that may sometimes seem for C...)  Any random language will do for the machine, with whatever degree of translation might be necessary (whether trans/compiled, interpreted, emulated..).  We choose to work in higher level languages, because it's easier, more powerful.  So, what's the point?  Use whatever is most accessible and familiar for anyone who may be reading your code.  So if that means using mfg recommended tools, code generation and etc., probably use that.  If a GNU or other free toolchain is reasonable enough, probably use that.  The more you cook from scratch, the less familiar it will be to others, and the more important it is to be well written.

Which, mind my audience, here -- if you're just getting started, or inexperienced, or tend to be a bit too clever for your own good -- pace yourself, tone it down, let the tools work for you.  Make your intentions known, try to leave insightful comments (AND KEEP THEM SYNCED WITH THE CODE), don't prematurely optimize.  CPU cycles are cheap, stick to easy-to-use, well-trodden platforms.  If you're experienced, you write clear code, you're prepared to deal with less common, less well documented platforms -- by all means, rock on.

If you aren't writing for anyone -- just pet projects and such, maybe you don't even intend to publish it online or anything -- sure, anything goes, get experimental, no problem.  Just don't burden others with spaghetti code and lacking comments.

Except when that's the point, in which case send it off to ioccc or whatever.

Tim

[Simon]

I don't have a problem with a functioning code generators but what microchip offer is several different versions of a mess. It's not even clear what to use for new projects. I hear lots of good things about the STM32 system. naturally things it offers are nothing to  do with the hardware at all like screen/gui creations which could probably be used on any mcu but for the fact that you have now been locked to that manufacturer, hm.

September last year I had been in my new job for 3 months. Various future projects had been mentioned and I was getting anxious. I had no MCU's. I had spent 3 months trying to find replacements for parts that the design subcontractor had been unable to find or that build contractors could not get. I had been giving my opinion on the limited and now expensive options for buying the STM32 parts used by the current projects: I don't have one, it's 50/50 that they are genuine and work just because I don't want you to blame me when several £1000's tits up wit the products.

I explained to my boss that given the supply situation we needed to purchase 2 years worth of stork of ANY MCU just so that we had some knocking around for potential projects with enough leeway to buy more before we run out with a years warning. So he agreed and I went hunting, I did consider the STM32 parts, including a good look at what we were already using but none available or the prices were quite high. For what I am doing I only need an M0/+, I had been looking at the samc parts because of the CAN availability and good price and they are safety rated which in my previous job was a good start but not needed here but still at $1.47 each they trumped similar spec chips from STM on price given the demand and no availability.

Now my experience was highly limited with the samc and I had set myself the task of finding anything and bare metal programming it, but at $1.47 each in 1000 and with some basic exposure to it the samc was ideal as a basic start and we could get them in November. Early December a "big" order of less than 20 units landed for something I had been working on that i inherited using the arduino/samd, we had the chips in stock and delivered. Now we are doing a new batch.

the samd of course was also unobtainium as everyone is probably using it in their arduino based projects - yuk.

i would love to find a source of higher level libraries wiuth proper instructions that are not architecture locked but it's hard. Now if you look at the price of a PCB with an SNTM32 based on an M3 or M4 you may as well get a PI-0 board for the same price and leapfrog into GUI projects with all that support operating systems give you. Currently we - me and my new apprentice who can none the loss teach me a thing or two  - are developing a project that uses a PI 3 for the heavy graphical stuff accompanied by a samc to do the real time stuff. Perfect solution. Yes he insists on spending a week to get the microchip configurator to work, but in less time I wrote start up code for the same peripherals he uses and he had to give up on the ADC DMA only to end up after 2 days of messing about with the same result in the configurator that I had with less than a day writing code.

[Rudolph Riedel]

I like to use ATSAMC21 and ATSAME51 but it looks like the next time I can get some fresh ATSAMC21J will be 10/2023, maybe. :-(
At least I have a few more ATSAME51 and the ATSAMC21J and the ATSAME51J electrically only have 1 pin different, so I already have designs on which I could populate either with the help of two configuration resistors.

In the future I probably will only use ATSAME51J19A-AU and buy a few more.

STM32H7 looks interesting but is unobtainium as well.
All the manufacturers would do themselves a favour if they would feed some stock to catalog distributors like Mouser and Digikey.
Right now there is a good chance that I give up on hoping to get more ATSAM when I find a controller family from some other manufacturer to replace these with - and the bar is getting lower.


 

[Simon]

For microchip buy direct, for STM32 go screw, TSM won't let you pre-order, you can sign up for a notification when they are in stock, I expect by the time you get to them they will be sold out to the dealers. So STM are not looking very hopeful for me.

[chickenHeadKnob]

STM32H7 looks interesting but is unobtainium as well.
All the manufacturers would do themselves a favour if they would feed some stock to catalog distributors like Mouser and Digikey.

My intuition on STM32H7 and other similar desirable higher end parts is that many folks have designs ready or nearly ready to go and are simply waiting for them to come back. Thus prolonging the shortage for those parts in particular. You  need to apply game theory and guess which expedient parts are lowest in demand and will be first in surplus.

[NorthGuy]

My intuition on STM32H7 and other similar desirable higher end parts is that many folks have designs ready or nearly ready to go and are simply waiting for them to come back. Thus prolonging the shortage for those parts in particular. You  need to apply game theory and guess which expedient parts are lowest in demand and will be first in surplus.

Shortages affect everything. It is impossible to predict the future. The best strategy is to buy what you can and re-design.

[SiliconWizard]

Yep. The times are better either for the very large corporations that can hoard large stocks and manage to have priority anyway, or for the companies selling low volumes of very high-margin products (then you can handle relatively modest stocks and still make a profit.)

[Simon]

STM32H7 looks interesting but is unobtainium as well.
All the manufacturers would do themselves a favour if they would feed some stock to catalog distributors like Mouser and Digikey.

My intuition on STM32H7 and other similar desirable higher end parts is that many folks have designs ready or nearly ready to go and are simply waiting for them to come back. Thus prolonging the shortage for those parts in particular. You  need to apply game theory and guess which expedient parts are lowest in demand and will be first in surplus.

Which is what I did and picked SAMC, a part that I have to actually program, pitiful code configurator no one will want to use and not used in arduino or other stupid open source or IoT crap.

With high end chips there will be people hoarding just in case like I bought 1'000 parts that I knew were more than capable enough so that I could do anything with them. My new colleague wanted to buy STM32's but to get any worth using given that we have an M0+ solution already would be so expensive that we won't.

[westfw]

picked SAMC, a part that I have to actually program ... and not used in arduino

MattairTech has a core for Arduino that supports SAMC21.

Even if you really hate Arduino, it can be useful to look at their code (MattairTech's code, I guess) to see how they have initialized things (for example.)

[Rudolph Riedel]

For microchip buy direct,

Good one, most of what I am looking at has an estimate shipping date of 31-May-2023, much as the rolling dates on Mouser.
And most orders need to be in multiples of 160 if not 1500.
Then I found one item with incoming stock of 81 in june but could not pre-order since it could only be ordered in multiples of 160.
Order now and you may receive some day.

At least I ordered enough CAN-Transceivers directly from TI before things got ugly.
Now I ran out of digital isolators.

[Simon]

picked SAMC, a part that I have to actually program ... and not used in arduino

MattairTech has a core for Arduino that supports SAMC21.

Even if you really hate Arduino, it can be useful to look at their code (MattairTech's code, I guess) to see how they have initialized things (for example.)

I can use the code configurator as an example, it will be less crap than anything arduino like. I just wont use the arduino it would be a ridiculous business decision.

[Brianf]

I just wont use the arduino it would be a ridiculous business decision.

Why would it be a ridiculous business decision?

Surely good business decisions are about delivering the right product at the right time, and at the right price. If using Arduino achieves that then what is the problem?

[Simon]

To rely on something like that is not a good idea at all. My main requirement is libraries for higher level stuff, anything that is about setting up peripherals needs doing properly. It's a bad business decision to use a system that hides the majority of the hardware from you. I want to use the hardware, not ignore it.

People seem to be obsessed with programming microcontrollers like they are running an OS or are a larger computer. Microchip confused the hell out of me by calling interrupt routines callbacks, a callback is a software interrupt in an OS, nothing to do with a microcontroller hardware interrupt, but clearly they target the software developers.

Arduino is all about the software and totally ignores the hardware. So people will be running around mimicking what the hardware will do for them using slower software constructs.

I like to know that when I open my code tomorrow it is exactly as I left it and that I am not trouble shooting someone elses problem as I have in the past with arduino.

Unsurprisingly Arduino are now targeting the industrial market as monetising the arduino no longer works. Industrial stuff wants to be safe etc. my guess is that they will now be selling licenses in the same way freeRTOS does if you want to use it in something safety critical. Good business move.

[Simon]

I just wont use the arduino it would be a ridiculous business decision.

Why would it be a ridiculous business decision?

Surely good business decisions are about delivering the right product at the right time, and at the right price. If using Arduino achieves that then what is the problem?

Yes, and at a time when I could not get a SAMD because it is used in arduino and everyone wanted them I chose another chip and resolved to do it the hard way. As a consequence my employer is very happy with the fact we have delivered on time, my time (money) spent bare metal coding taught me a lot that will save a lot of time in the future and instead of some STM32 that was inflated to £14 I got perfectly adequate chips for £1.50. So yes at the right time certainly, price uh you can argue about the value of development time and the cost of debugging stuff you did not write.

[westfw]

I could not get a SAMD because it is used in arduino and everyone wanted them

Normally when the subject comes up, we conclude that "the hobby market" is too small to have much impact on sales of a manufacturer as large as Microchip...
(Right now, of course, everything is unobtainium, whether it's used in Arduino, or used elsewhere. :-( )

I can use the code configurator as an example, it will be less crap than anything arduino like.

If you say so.  I thought your string of recent complaints was based on the configurator NOT working on your target chip...

[Simon]

I'm sure plenty of companies use arduino too, I remember selling some atmega328's through ebay to a company, I believe they did safety stuff....... People may also be using the SAMD for non arduino stuff, at the time SAMD or SAMC were the same to me, I would have used either but SAMC was about to be available so I woent with that, they are similar chips.

Yes I do hate the code configurators, but my colleague uses them as he does not see the value in spending time on registers and as much as he hate microchips particular sluttering he uses it anyway. So I don't mind letting him set the peripheral up for me in a project with it so that I can then go in and look at the code to see that I have understood it correctly.

one day people might realize that their projects are broadly similar and they may be relying on their own standard boards so actually setting up the whole thing with a code configurator becomes more time consuming than taking a copy of your own template code that has already got all the settings you plan to use. If I didn't want to do registers I'd get something running a proper OS and learn to use that.

[voltsandjolts]

I have a lack of trust in configuration tools, but viewed as a reference for getting configuration done in your own code, they're quite useful.

[Simon]

I have a lack of trust in configuration tools, but viewed as a reference for getting configuration done in your own code, they're quite useful.

Quite. The little look I have had at microchips I need to understand the hardware so have to read the datasheet in detail, by the time I have done that I can simply write some code myself that is in english with copious comments and whatever I like to explain to my future self what I was thinking.

I looked at the code to set a counter up. having got a signal it confirmed that the debugger could not show what the counter value was so I wasted a day or more trying to strat the counter not realizing it was running.

I did learn that to check if a sync bit is set busy something I thought should work did not. I thought that

while ("sync reg" & "my mask of bits to test") {}

should work it turned out that this was not the case. The configurator code used:

while ("sync reg" & "my mask of bits to test" == "my mask of bits to test") {}

which when I replicated it worked, who knew, to me they look the same. but solved.

[T3sl4co1l]

Yeah if you need "mask of bits" plural, and need all of them, you need the equality.  Or if some should be low, equality to the desired states.  If it's just one bit, it's equivalent to truthiness (or falsity) of course.

Tim

[nigelwright7557]

 I have used PIC's since 1985.
Was a PIC consultant for 13 years.
So done hundreds of PIC projects.
The current situation is most SMD PIC's are not obtainable.
I have been resorting to using through hole PIC's as I can still get some.

MPLAB is huge, not easy to navigate and error prone.
My PICKIT 3 does fall over now and then.
I found the Snap to be more reliable but it needs the pcb powering to use it. Just a cut down and cheap pickit 4.
I found debugging is difficult with MPLAB if you use compiler optimisation. It gets hung up about optimised code and ends up with broken breakpoints.

The PIC's can be pretty horrible. i/o pins that have strange uses and mean you cant use a full port to pass data. i.e. VUSB
The smaller PIC's have none linear memory spaces which can sometimes cause problems.

I no longer use assembler and just use XC compilers. Most of my projects are converted to C now.