Thu, 10 Nov 2022 20:55:51 +0100
Enable image_process pugins again
137 | 1 | ========================================== |
2 | HP 34970A Data Acquisition Unit - part 6 | |
3 | ========================================== | |
4 | ||
5 | ||
6 | :Author: David Douard | |
7 | :Category: Electronics | |
8 | :Tags: HP, 34970A, HP34970A, DMM, repair, test equipment | |
9 | :series: HP 34970A repair | |
10 | :series_index: 6 | |
11 | ||
12 | ||
13 | Quick follow up; I recently got back on this project to try to complete it and fix | |
14 | remaining problems. | |
15 | ||
16 | The source code for the firmware is available here: | |
17 | ||
18 | https://hg.sdfa3.org/hp34970-firmware | |
19 | ||
20 | The Kicad project is available here: | |
21 | ||
22 | https://hg.sdfa3.org/hp34970-pcb | |
23 | ||
24 | ||
25 | Making the USBDevice work | |
26 | ========================= | |
27 | ||
28 | As described in the `previous episode <{filename}/hp34970a_5.rst>`_, I could not make | |
29 | the USBDevice of the STM32F303RD work with `mbed-os <https://os.mbed.com/>`_ framework I | |
30 | use to write the firmware. | |
31 | ||
32 | The `STM32F303RD | |
33 | <https://www.st.com/en/microcontrollers-microprocessors/stm32f303rd.html>`_ is not yet | |
34 | officially supported by mbed-os (mainl because there is no `NUCLEO | |
35 | <https://www.st.com/en/evaluation-tools/stm32-nucleo-boards.html>`_ board with | |
36 | this CPU. | |
37 | ||
38 | However the STM32F303RE is supported, at least partially. For example, the USBDevice is | |
39 | not working at all in the latest version of mbed-os (6.15.0). But some `work has been | |
40 | done recently <https://github.com/ARMmbed/mbed-os/pull/15116>`_ to enable USB_DEVICE on | |
41 | CPUs of the STM32F3 family. But this last PR was not enoght to make the USB device work; | |
42 | I have submitted a `PR <https://github.com/ARMmbed/mbed-os/pull/15132>`_ that needs to | |
43 | be merged to make the USB device actually work. | |
44 | ||
45 | Also, the difference between the 2 being (as far as I know) only the size of the flash size | |
46 | (384KB vs. 512 KB). | |
47 | ||
48 | So I had 3 problems to solve: | |
49 | ||
50 | - compile the code for a custom board using the mbed framework in platformio, | |
51 | - use a custom CPU decription, | |
52 | - use a patched version of the latest version of mbed-os in platformio (while the | |
53 | officially supported version at this moment is only the 6.9.0). | |
54 | ||
55 | ||
56 | Custom version of mbed in platformio | |
57 | ------------------------------------ | |
58 | ||
59 | Using a custom version of a framework (here mbed) in platformio is not properly | |
60 | documented. But thanks to `the response | |
61 | <https://community.platformio.org/t/support-for-mbed-os-6-stable-and-mature-apis-cloud-services-support-enhancements-to-the-bare-metal-profile/15079/9>`_ | |
62 | to my question on the `platformio forum <https://community.platformio.org/>`_, I could | |
63 | make it work. Obviously, I submitted `a PR | |
64 | <https://github.com/platformio/platformio-docs/pull/214>`_ to `platformio-docs | |
65 | <https://github.com/platformio/platformio-docs>`_. | |
66 | ||
67 | So now I have my (more or less) `up to date repository | |
68 | <https://github.com/douardda/platformio-framework-mbed>`_ usable as `platform_packages` | |
69 | config argument. | |
70 | ||
71 | Custom board and CPU description | |
72 | -------------------------------- | |
73 | ||
74 | This part I am not completely sure how it should be done. There are actually 2 parts: | |
75 | - tell mbed about the CPU description, | |
76 | - tell platformio about is and use it. | |
77 | ||
78 | The platformio part is `documented here | |
79 | <https://docs.platformio.org/en/latest/platforms/creating_board.html>`_ | |
80 | ||
81 | It mostly consists in creating a json file in the `boards/` directory of the platformio | |
82 | project directory. | |
83 | ||
84 | For this project, I used the description of the NUCLEO_F303RE as starting point. The | |
85 | file is `boards/hp34970_fp_f303rd.json | |
86 | <https://hg.sdfa3.org/hp34970-firmware/file/tip/boards/hp34970_fp_f303rd.json>`_ in the | |
87 | `firmware source code repository <https://hg.sdfa3.org/hp34970-firmware/>`_. | |
88 | ||
89 | Once the board is described, it can be used in the `platformio.ini` file: | |
90 | ||
91 | .. code-block:: bash | |
92 | ||
93 | ~/e/hp34970-firmware$ pio boards | grep hp34 | |
94 | hp34970_fp_f303rd STM32F303RDT7 72MHz 384KB 64KB HP34970 Replace Front Panel Board (STM32F303RD, 64k RAM, 384k Flash) | |
95 | ||
96 | ||
97 | ||
98 | Then we need to tell mbed about this board. This is `documented there | |
99 | <https://os.mbed.com/docs/mbed-os/v6.15/porting/porting-custom-boards.html>`_ | |
100 | ||
101 | It is a matter of creating a `custom_targets.json` file with the board description. | |
102 | Again, I used the existing NUCLEO_F303RE as starting point, then added/removed what I | |
103 | needed for this project: | |
104 | ||
105 | .. code-block:: json | |
106 | ||
107 | { | |
108 | "HP34970_FP_F303RD": { | |
109 | "inherits": [ | |
110 | "MCU_STM32F303xE" | |
111 | ], | |
112 | "overrides": { | |
113 | "clock_source": "USE_PLL_HSE_XTAL" | |
114 | }, | |
115 | "device_has_add": [ | |
116 | "USBDEVICE" | |
117 | ], | |
118 | "mbed_rom_size": "0x60000", | |
119 | "bootloader_supported": true, | |
120 | "device_name": "STM32F303RE" | |
121 | } | |
122 | } | |
123 | ||
124 | Then the actual description (pins, memory config, etc.) are C files put in a | |
125 | `TARGET_HP34970_FP_F303RD` directory. Here again, started from files copied from the | |
126 | `STM32F303xE target | |
127 | <https://github.com/ARMmbed/mbed-os/tree/master/targets/TARGET_STM/TARGET_STM32F3/TARGET_STM32F303xE>`_. | |
128 | ||
129 | ||
130 | End result | |
131 | ========== | |
132 | ||
133 | It took me some time, but in the end, I finally got a firmware that works OK, including | |
134 | a working USB serial port. For now, this serial connection is still used only for | |
135 | debugging purpose (and firmware upload in DFU mode), but it may be improved in the | |
136 | future, if needs be. | |
137 | ||
138 | So I could put the HP34970A back together (at last). I drilled a couple of small holes | |
139 | for the Reset and DFU buttons and a (not so) square hole for the USB plug in the blue | |
140 | glass. | |
141 | ||
142 | The result looks very decent to me: | |
143 | ||
144 | .. image:: {static}/images/hp34970a/finished.jpg | |
145 | :alt: finished look of the device with the replacement front panel | |
146 | :class: image-process-large-photo | |
147 | ||
148 | .. peertube:: 14d0e7e2-74d5-44fe-90e9-d6b102ae01a1 |