Great dev kit experience?

[the_hard_six]

Has anyone ever gotten a dev kit and had a great experience?  Well designed, well packaged, well documented, well supported?  Looking for examples.  Thanks.

[Sal Ammoniac]

You ask a tough question.

I have dozens of dev kits and all of them have their quirks and annoyances. Bad documentation is a common theme. Another pet peave is putting too many useless gadgets on the board so as to make it difficult to use some of the peripherals on the MCU.

Some of the best designed and documented dev kits I have are from Digilent. ST dev boards are not bad, considering their low cost. Microchip boards are good too.

[EverydayMuffin]

I think the best dev kit I've gotten for any MCU is the PSoC 4 BLE Pioneer Kit from Cypress.

https://www.cypress.com/documentation/development-kitsboards/cy8ckit-042-ble-bluetooth-low-energy-42-compliant-pioneer-kit

Cypress's documentation is top notch and they also have a great video series (about 6 videos) called PSoC  4 BLE 101.

[tmadness]

The PSOC series def has some good documentation and some really unique features, but by good lord the IDE sucks and it does not do the basics well. The PSoC is like Andy from the office (he graduated from Cornell), the whole experience is defined around the "cool" thing called analog blocks, if you want a Cornell graduate MCU with analog blocks there's literally nothing else like it, but if you want a good general purpous MCU look elsewhere. Double- oh god no -for the PSoC6 it looks capable but everything about is half baked, even the PSoC IDE did not support it properly.
There's STM32, its a steep initial hill to climb but once you are on top at least ST is consistent. Also use LL drivers HAL is waste of time.
Now, some people are going to skin me but consider the teensy series, aka Arduino for grownups. It is literally just Arduino in name. You want something to just work, hey you can do that fast and easy with a very well documented and written code base (SPI, UART, printf). You think that you can write better code or you want to eek out better performance go arount6 the provided libraries, hell just re-write them they are all just NXP microconrollers. NXP datasheets are okay, the peripherals don't have the kitchen sink included (like ST) but hey it will work for 90%+ use cases. Just check out PJRC forums the core team is very very helpful. They have 4 Microcontroller options(LC, 3.5,3.6and 4.1), if you want to migrate your project down the road to an actual product just use the same code and build a new PCB. Just a note DO NOT INSTALL THE TEENSYDUINO/ARDUINO IDE, its built for kids. Use platform IO + VSCode (FREE) endlessly customizable and supports some really advanced stuff with just add-ons

[westfw]

I have a 68HC05 development kit ("only $68.05 !") that came with a looseleaf binder full of tutorial/textbook like info.Pretty amazingly informative.   Unfortunately, the chip itself was from the "erase using UV" vintage.

[Kerlin]

Got to agree there that 6805 Dev kit was the best ever, I bought it for to do a project at work sure made it easy.

It was however an assembler based kit which made it very hard to mess up, unlike most kits now which have very buggy compilers done by college kids.
After battling away with such later kits I can see why the old guys prefer ASM, but you have to move away from it to stay employed.

But it was later followed by the worst I have come across the one for the 6808.
The problem there was you had to call to get a password for the software. Here in Australia had to call the HK office they wanted to know why we wanted it and wouldn't give out any password, didn't even know the kit included their free software.
I had recommend it to work, a Defense company and they bought a kit for everyone, which was instantly useless because of the password problem, boy didn't I look bad, good buy Moto.
 Later did large production domestic items in Taiwan and China and refused to use Moto due to this. See what happens when you produce crap kit.
I don't think they seem to know it but good kit is how a silicon manufacturer wins sales.

Because one is expected to hit the ground running (fast) in this profession I buy kits from my own pocket and learn them at home.
I find its best to do do this in this in stealth mode at home so you can try them out. This happens more than they realise.

For me Atmel/AVRs early success is very much due to the STK500 which was simple and neat. That situation is currently being utterly destroyed.

Arduino deserves and has got a lot of credit.

Like most people I am guessing its time to move onto STM, I have heard some real horror stories about their compilers, just look on here.
It seems the Blue/Black pill have been messed up by fake copy kits, know people who have bought them.
 

[westfw]

I don't think they seem to know it but good kit is how a silicon manufacturer wins sales.

Indeed.  Although it's difficult to judge between chips you are familiar with, and "new stuff."We ported a pretty massive project from 68000 to 68331, and made good use of the Motorola Eval Board for the 331.  But we were already pretty much 68k "experts", so that was a much different experience than if we had been starting from scratch.
I've had a couple of significant observations, over the years:

• The number of times that a "real" development engineer actually gets to "evaluate" a processor, as in "based on ordering several different development kits from different vendors", is pretty small.  More often, other criteria override such evaluation.  "Known Vendor", "new chip in known architecture", "supported/supports current development infrastructure", "compelling sales presentation", "the boss said this one", "reputable vendor", "second sources", "reputation", "it's what they taught in class", or "I used one of these at home"...
• It follows that the win the hearts and minds of engineers, it's more important to get chips and boards and things into their awareness when they're NOT specifically evaluating anything.  The (relatively recent) replacement of $400 Eval boards (purchasable with finance department approval) with $10 to $100 boards (purchasable on a "personal whim") is a great thing.  "If you want to lure engineers, treat them like hobbyists."

I have heard some real horror stories about their compilers

I haven't seen that many complaints about compilers.  Code Libraries seem to be well-hated, and everyone has pet peeves about essentially every IDE that has ever existed, but almost everything bigger than 8 bits uses some release of gcc, which is pretty good, usually.

[TK]

You should first select which MCU you want to use, then buy the dev kit that has the same MCU or the same family.  I tried AVR, arduino, TI launchpad, PSoC, STM32 nucleo.  For me the best is STM32 MCU and dev board (Nucleo, Discovery).  The ST nucleo boards offer a huge selection of STM32 MCUs to use.  STLINK debugger is so so, but you can reprogram it with a limited free version of Segger Jlink or buy the Segger Jlink edu mini for less than $20.  With the nucleo + Jlink mini edu you will spend something around $40-$60.

[westfw]

You should first select which MCU you want to use

But how do you know that, without having experimented with a development kit?

[tszaboo]

The old PIC16F384 devkit, that came with the PICKIT 2 was great. I could get that micro working with the IDE in no time. As far as I remember, it even had a CD with everything on it.

 The issue with todays devkit, it the time it takes to get it working. You need to download an IDE, the examples, you have 0 chance to get it working without the libraries. They ask you to set up GCC or KEIL or whatnot. The example is not written to free IDE, and you have to figure out how to set it up. And the links in the example dont work, because some weed smoking hippy website developer moves the links every 2 weeks. God forbid, you have to write makefiles.

If it takes more than 15 minutes from the opening of the box, to be able to compile for the board, just give up.

[brucehoult]

This is where both Teensy and standard Arduino, both with USB connections, are very good. When I got a Teensy 4.0 (600-1000 MHz Cortex M7) it took about 2 minutes from unboxing to get a blinky working. (I already had and knew how to use the much maligned Arduino IDE) The same also with the RISC-V based HiFive1 or SparkFun RED-V RedBoard.

In all cases, just connect the board to USB, enter a provided URL in "File/Preferences/Additional Board Manager URLs", give it a few seconds to download the BSP (including compiler and libraries), select the board and port in the Tools menu .. and you're done.

[TK]

You should first select which MCU you want to use

But how do you know that, without having experimented with a development kit?

In my case, I did my research starting with AVR, which is the MCU I worked with before... then purchased a dev kit for the micro I needed... then you find something you don't like or a roadblock due do a missing feature or peripheral, then you move to the next MCU in your list and so on... so in the end OP's original question is irrelevant.  You will end up buying and testing most of the dev kits available in the market

[Doctorandus_P]

You will end up buying and testing most of the dev kits available in the market

Have you ever really thought about how many development boards there are? My first guess would be multiple thousands of different boards, but it also depends on what you call a "development board". Do you consider a reference design for a 25kW solar inverter a development board?

The mbed website lists some 160+ boards. Not that I particularly like mbed, (dislike would be closer), but the website does give a nice overview of different boards from multiple vendors.
https://os.mbed.com/platforms/

I would also focus less on a particular development board, or what's bundled with it. That was nice in the '90ies, before Internet happened. Now you can get any info about chips through your TCP/IP cable.

One of the most important things to look out for is probably a broad spectum of chips, and a growing path to bigger chips if you need them, unless of course you have some very specific application and some obscure chip is an exact match for it.

I do recommend getting a book or 2. All info is floating on the 'net somewhere, but it can be fragmented over lots of different places. Information from books tends to be more to the point and coherent.