siy
Member
>I would like to change the 64LQFP to make a few things more convenient, but I don't have the
>time to hack on the bootloader. Based on threads in the forum, ala42 has a better bootloader
>approach than LeafLabs; a future project?
I've tried to find thread with relevant discussion, but haven't found one. Can you, please, describe the approach or point me where I can read about approach in more details?
gbulmer
Moderator
siy - welcome back. I hope you're well. It's always great to see what you've been making :-)
Let me look for the thread. I'm getting sleepy (3am here) so I'll do that much later today.
The basic approach is to use the existing manufactured-in STM32F4 bootloader instead of having to flash LeafLabs bootloader. This should work with STM32F3 too.
There is a 'thin shim' of code 'under' the USB library which detects the host PC trying to upload a program. The 'shim' jumps into the STM32's system-memory-resident bootloader.
So, "ordinary" programs, which start at the beginning of flash can be loaded, without LeafLabs bootloader. They run as normal, but are capable of detecting the host upload, and loading over themselves.
siy
Member
By the way: I've compared USB library used in bootloader (I guess this is one of the old versions of USB device library provided by ST for STM32F103) with USB device library provided by ST for STM32F3. Beside type naming and some other minor differences libraries seems are identical. The difference should be in USB peripheral initialization. If I remember correctly, it was moved to other bus in F2/F4 and most likely in F3 too, but it is trivial change. The code for writing to flash in newer MCU's should already exist somewhere (IIRC ala42 mentioned that it has such code for F2/F4). Overall it looks like porting Maple bootloader to new families should not be very complicated task.
raymadigan
Member
Just another perspective, I started using arduino's several years ago. I build real time running applications that I use around my property. I got to where I would build my own arduino style boards because you can't put an arduino on a circuit board. Making a project that has to run outside all year long where you connect with the microprocessor by sticking wires in headers didn't make sense to me.
The problem then becomes, how much code can you put on the Atmega328. It is so limited in memory that having more then one external device ran it out of memory before the application could be developed.
Then I found the Maple Mini. It is an awesome chip, the problem is there isn't any wy to get any eyes to assist one at getting through their issues. If this is going to be a successful project, there has to be a way get to the bottom of issues using it.
I think the platform is in a good spot. One can use these boards to build real application, that can do real things. I have one I am making the second pass at that has an LCD display, a set of switches to allow the user to adjust running parameters, and an SD card to log the state of the project when it is running so faults can be found and repaired. I could never build this application with an arduino and I am not sure what I will do if this project doesn't continue.
siy
Member
@raymadigan, I think there is no need to worry. Project definitely continues although now it is more driven by community than Leaflabs. Having Leaflabs more involved would be great, but we have what we have. Year ago situation looked much worse. Now we have several different designs of Maple compatible boards, much better libmaple with support for new MCU families. And more things to come. Wish I had more time for my hobby :)
gbulmer
Moderator
raymadigan - "If this is going to be a successful project, there has to be a way get to the bottom of issues using it."
I agree.
LeafLabs seem to be okay supporting the forum site. They have recently moved parts of the web site to new software software, so I think they plan to keep it alive. Several of us can sweep away spam, so it is maintained. Folks drop by often to try to help answer questions, and I think answers are high quality. However it isn't at anything like the scale of Arduino.
While it is only a small part of the answer, I sometimes pop over to ST's STM32 forum to look for threads on peripherals. I hope that expands my knowledge base to help here.
We are fortunate that ST produce STM32xxDiscovery boards at very low costs, sometimes under $10. These give us a low-cost way to try out the latest parts, and develop for new microcontrollers, or make new boards using them.
An important feature of STM32 Discovery boards is the ST-LINK/V2 hardware "SWD" debugger which can be used to debug any STM32 by connecting 6 (5 really) wires. There are several open source gdb servers (I've used https://github.com/texane/stlink), which give 'proper' hardware debugging (single step, instruction breakpoint, data watchpoint, etc). So, from a hardware perspective, Maple is in a good place.
ala42 has found that the the STM32F30xxx runs USB code with one change. ala42 has also developed a bootloader-less technique for using the newer STM32Fs which contain a manufactured-in USB bootloader. So based on ala42's approach, a new board 'just works' (TM:-).
I hope the community can encourage folks to participate by creating boards which fit a variety of uses. For example, several people are making or working on more DIY friendly boards than Maple or Mape-mini, while being software compatible.
Several folks have improved the IDE. For example by replacing the ageing version of gcc with a newer ARM sponsored version https://launchpad.net/gcc-arm-embedded which includes new features including fixed point arithmetic.
Some of this is quite technical, not integrated or not documented (maybe wiki pages gathering some of this stuff up might help?).
However, I hope it shows that the community is alive, and actively progressing.
raymadigan
Member
Thanks for your response. It is not my intention to say here that there isn't anything going on, just that it is the very top of the technical spectrum.
I don't like the arduino at all, I just use it as an example. I think what they have is way inferior to the Maple, they even are building an ARM board. The issue with the Leaf Labs is at this point, it is only the highly technical people solving highly technical issues for themselves.
If there is a real issue someone has they have to be willing to solve the problem alone they move on. There is no middle technical community of people working with other people helping them get their problems addressed.
I don't think it is anyone's problem here, there just doesn't seem to be enough to gather any momentum.
For example, one of the things you mentioned in your post might help me, I thought of trying to do something like it myself. I have a problem with an SD Card board that when I build my application in a flat model, a bunch of IDE files with everything global my application works. If I build it with libraries, just classes really, then I get my problem. I think it has something to do with how it is put together. So, a new compiler would be potentially useful. I will spend the weekend building my application classes again from base classes to see if i can narrow down my issue.
My issue is not why I am writing here though. Somehow there has to be more discussions for solving issues with people trying to build applications on this project, or there won't be people building application on this project.
You must log in to post.