[Mike] is an avid PIC developer and replaced his ICD3 debugger for an ICD4. He made a video with his impressions and you can see it below. [Mike] found the heavy aluminum case with a sexy LED attractive, but wondered why he was paying for that in a development tool. He was also unhappy that they replaced the ICD3 cable connections with new connectors. Finally, he wished for the pin out to be printed on the case.
On the other hand, the ICD4 will also do JTAG and handle the Atmel parts (which Microchip acquired). [Mike] opens the box and shows the inside of the device before actually using it for the intended task.
As you would expect, the ICD4 integrates well with MPLAB and does the same kind of functions you expected from the ICD3 and other Microchip programmer/debuggers. [Mike] found the programming algorithm was a little different from previous devices, so he put the scope on it and also compares the speed of programming between a PICKit3, an ICD3, and the ICD4.
ICD4 is a pretty serious tool. If you want to just experiment, you can build your own PIC programmers or try the PICKit 3. You can also pick up a cheap board and go the online IDE route.
Really glad to see others use the “Code, Flash, Test” method of debugging code (plus shitting out of variables via UART) – I was lead to believe I was crazy for doing this.
Where exactly did you see it here?
14:00 ->
I see, thank you. Indeed, for small programs it’s possible, otherwise a debugger is gold.
for somethings it is gold, for much hard real time stuff it is useless
I use “Code, Emulate, Test” faster, but probably still in need of debugging.
You are crazy to do that. Print F is super slow.
Try a real processor like an STM32 or even an STM8 with live debugging in a watch window.
Who said anything about printf? Write directly to TXREG and use a terminal that displays values in hex. With FTDI you can easily get megabit speeds
We have loads of problems with our production programmers of Atmel. We upgraded to Atmel ICE and studio 7 (atbackend) because mk-II and stk500 couldn’t program SAM. The mk-II became obsolete and there is no suitable replacement as the Olimex is currently unstable at higher programming speeds (with atbackend, driver level).
The mk-II works fine with atbackend. The Atmel ICE is slightly faster but has a failure rate of 10% and is not acceptable. I’ll take a look at the ICD4 but it’s likely we would need to rewrite our programs/drivers and I’m not happy with the extra work.
Why don’t you use SAM-BA Boot for your SAM chip? No programmer needed.
Thanks I’ll give it a look!
Take a look at programmers from ASIX – Only used them for low-end PICs a while ago but very fast and configurable.
Thanks for the information. I’ll review the products. I’ve talked with a few vendors about in system programming, there are loads of solutions. I would go this route. The problem is development programmers are really cheap.
That’s funny, we had loads of problems with Microchip programmers but the Atmel ones have been really solid. I was hoping that the ICD4 would do away with having to download a different OS when you switch device families, but it seems not.
The Atmel ICE just works. You can use the command line tool they provide with it as well. We have a lot of automated test equipment that calls the command line tool. Of around 150 deployed we have had a single failure in over a year of hard use.
The ICE3 units are a real pain. We ended up having two per test system, because swapping the OS between PIC24 and PIC18 versions often fails. They often simply fail to program the target device too, and are very slow. With Atmel it’s easy to set the programming speed to whatever you want, as fast as your board can accept.
When marketing team gets their fingers on product design. It is nice looking and robust design, but in lab it is completely unnecessary. I come from Atmel ecosystem and I have get accustomed to plain PCB dev tools without casing. Usually I just put some kind of insulator to bottom of PCB and do not bother to make casing. For end user ready product I appreciate hefty aluminium case design, but during development easy access and no BS bling is main point of tools.
Round puck design is just bad, does not fit nicely in any box or drawer and looks cheap even with heavy aluminium case.
But hey, it looks nice in Microchip booths at conferences. And then some other marketing people will see it and must order it for looks. They even get nice stickers that they can put on the door of dev lab.
some kind of case is very valuable, to prevent accidental shorts but any more than that is just a waste. If programmers are supplied as bare boards for minimum cost, they should be designed to fit a standard Hammond box or similar for those that want it
A reminder that Atmel ecosystem IS microchip ecosystem now, the future of AVR tools will undoubtably be headed down the same tunnel
I’ve talked with Microchip/Atmel. They plan on integrating Atmel into the MPLAB environment. Atmel Studio will become obsolete. This tool is a step in that direction (Atmel support).
Jeebus, how much coffee did this guy drink before shooting the video? All that fidgeting in the beginning was driving me crazy, hold the damn thing still for 2 seconds, please…
I don’t get that 8 minute ranting part about case design. What they should have done, sell you 250 bucks device in cheap plastic case that feels like 2.5 bucks eBay clone?
No, use a cheap case and charge $200 instead of $250
They should sell them at cost. As soon as they took over Atmel they doubled the price of the cheap and better Atmel ICE debuggers, and of course they charge for non-crippled compilers, because why on earth would they want to lower the bar to entry to their platform?
They should focus on selling the parts and view all the development equipment and software as a way to get people to buy more MCUs, not a source of profit.
something is really wrong with microchips programming procedure, because the very cheap cc debugger can program a bigger hex file into the ti’s cc microcontroller, in less than 10sec, but for the pickit3 with little less hex file, it took more than a minute, so i’m really considering to buy icd4, but only if it can program a device around 10sec, anything more is a waste of your time
That must be some serious hex file. I can fully load a PIC24FJ32GA002 in under two seconds with an ICD3. We also use a variety of larger flash PICs that flash well under the 10s mark.
The PICKit3 in my experience is noticeably slower than the ICD3, but you get what you pay for. At less than $75 which is what I paid for mine, I don’t expect the performance of the higher level programmer/debuggers.
I went to DigiKey to purchase an ICD3 and was surprised to see they had 0 stock and suggested a substitute. Is the ICD3 being phased out ? I have the newer ICD4 and it works fine but production all uses the 3. Maybe I should call microchip ?