search

grumat / glossy-msp430

Glossy MSP430 - Embedded GDB Debugger for MSP430
GNU General Public License v2.0
4 stars 0 forks source link

Testing #1

Closed codehero closed 1 year ago

codehero commented 2 years ago

Can I buy an assembled probe PCB from you to test with MSP430FR2422 and MSP430FR5739?

grumat commented 2 years ago

I am glad you are interested. Actually the project is at a very early stage. The best stand is based on the Bluepill prototyping board. I am working on a revision 2 of that board. The physical board V1 that I have here requires at least one important patch for use. Also firmware state is still not usable. Since I've added unit tests, and will improve it all 3 MSP430 variants (at the protocol level) are now evolving more consistently. If you are not in a hurry I can send you one for free. I am not a hardware GUI, so my soldering iron skill helps me build usable units. If you have a BluePill around then it will be good to start.

codehero commented 2 years ago

No worries I am not in a hurry, but I am very interested in this effort as the TI FET is very expensive, slow and has failed on the factory floor. When you are finished with V2, I could order some PCBs here in the US. Shipping from Austria would be very expensive

grumat commented 2 years ago

For every hardware that I am testing I am using JLCPCB and the fabrication files are on the repository. usually shipping is more expensive than fabrication. I am making some efforts to change the STM32 variant, since the F103 simply disappeared and I don't want to play around with clones, which could add more unnecessary distraction. But yes, the first working prototypes will still be based on F103. Although I am trying to optimize peripheral use to avoid bit banging at some moment I will end up with a ST-Link pin-compatible variant, exactly like the Black Magic Probe, so people will have more chance to test the environment. Those that are searching for top optimization, can always use the fine tuned and official version. I will consider this option, so it may help everyone who wants to test; just invest in a cheap ST-Link clone to have a start.

codehero commented 2 years ago

If I understand correctly, you are developing FW that can be installed over an ST-link clone, so no custom PCB necessary? that would be awesome.

grumat commented 2 years ago

Yes it is planned, but because I am researching all possible ways to improve speed using SPI and DMA wherever possible I am using a fine tuned pin layout. Now that the project has some audience, so I am motivated to keep a parallel configuration compatible with the ST-Link. Take a look at Converting a Clone ST-Link to BMP on the wiki. This is my Black Magic Probe conversion that I am using for Debug sessions, which includes a Serial port like usual Nucleo boards. Although a VCP is currently not implemented it is still on my scope. I am buying these clones, since I am interested on the plastic case for the main project. Last year I've made a "scaled" prototype to validate the mechanical feasability and it was a success. Some photos are already on the Wiki. They are actually working, but as I said, now I am crunching bits and timings. It is my spare time hobby, no boss, no customers to worry about, so I can try to optimize every single feature, but still conforming to datasheets.

codehero commented 1 year ago

Hi, thank you for work on this project so far! I am planning on a Blue Pill V2 board build soon, but I saw images of V1 on the Hardware/BlackPill-BMP/README.md file. I just want to confirm that V2 PCB is ready to get fabricated.

grumat commented 1 year ago

Hi, yes, I received the PCB last week, together with a batch of other PCBs. The V1 is deprecated, although firmware is still configured for it. I will start to solder components today and it will probably ready just today. On the next days I will upgrade firmware, when V1 boards will be completely useless. Besides, voltage translators for the V1 concept are causing serious stability issues. They are automatic ones, but because of parasitic charges on cables fails badly. The V2 uses voltage translators that aren't automatic, which means some new control lines are required to handle data flow direction. I will maintain a CHANGES.md file for this project too, which may have patch notices as it happened with other PCBs of the repo. Very important: use a BluePill as these boards are easier to handle. For the BlackPill, I've got a single original unit and all other samples I've got, are just weird clones that mixes BluePill peripherals with Cortex M4 cores, a kind of a merge of both generations. So I gave up with this line-up and in the future, upgrades to Cortex M4 will use more modern variants, like STM32L432KBU6 or STM32G431KBT6 which I've already have samples here (but this is another story).

codehero commented 1 year ago

I see, you have added enable lines for JTAG etc. I will be primarily focused on the JTAG aspects, so this looks good to me. I will go ahead and order some PCBs. Looking forward to trying it out. I bought blue pills on amazon before, so I should be good there.

codehero commented 1 year ago

What manufacturer and exact part numbers did you use for U4 and U5?

Also which part numbers for RN1, RN2, RN4, RN3, RN5?

grumat commented 1 year ago

U4 and U5 are the 74LVC125. important is the TSSOP-14 package and the working voltage is in the range of 1.8 up to 3.6V. The 74LVC125APW-Q100J should work. The others are 0402-sized resistor arrays of 100 ohms and 47Kohms. Part numbers could be EXB-28V473JX and EXB-28V101JX. Start to solder them first. I noticed today that footprints are very small and not suited for solder iron. A hot air gun will help fix two solder points and then you can continue with the soldering iron. Because no other component are soldered you can check if connections are ok using a multimeter. Having other parts on the board may interfere readings. All other parts are kind easy to mount. For the LED use a green one, because red is already mounted on the BluePill.

grumat commented 1 year ago

The part numbers above are on stock on mouser.com and I also checked for low prices.

codehero commented 1 year ago

Thanks for the quick response. I ordered a stencil and will reflow them in a toaster oven. I attached my intended BOM blackpill-bmp-v2-bom.pdf

grumat commented 1 year ago

Sorry for this one! I saw that the CSV and XML provided on the "output" folder are quite inaccurate. Points are:

I will fix the schematics and BOM outputs.

codehero commented 1 year ago

Thanks for the clarifications. I use breakaway header for the connectors. I also tend to toward digikey because their qr codes work well with my inventory system

grumat commented 1 year ago

It is Ok. Most parts have lots of equivalencies and with a max. 10 MHz bus, the range of options increase well. If you intend to use a debugger like shown here, I also recommend to have an adapter like this. Alternatives usually does not provide a SWO trace connection.

By the way, I sent updates regarding your issues.

codehero commented 1 year ago

Initially my application is to preprogram MSP430FR2422 before populating on a board. I currently use mspdebug FET430 and SBW and it is slow and unreliable. GPIO JTAG is not realtime on Raspberry Pi so I look forward to trying your device out

grumat commented 1 year ago

News: V2 is already working with the firmware. By reviewing connections I noticed that BluePill uses the green LED for PC13 pin. So please solder a red LED for D1, instead of green. Now it is consistent.

codehero commented 1 year ago

PXL_20230324_161311582 MP

I have a means of testing the hardware now, so I am going to close out this issue. When I get to using the actual software I will open a new issue if needed. Thank you for your help so far!

grumat commented 1 year ago

Nice job! I am currently at a slow pace because of undergoing projects at my job, but hope in the next weeks I can continue with this project.