Update on VisionFive 2 Compatibility Test 2022.9.15

SD Card

Brand Specification Startup
Biaze 32GB U1 Normal
DM 16GB U1 Normal
SUMSUNG 64GB U3 Normal
aigo 16GB U1 Normal
Lexar 32GB U1 Normal
kingston 16GB C10 Normal
SanDisk 16GB U1 Normal
Hikvision 64GB U3 Normal
Netac 64GB U3 Normal
Netac 32GB U1 Normal
kodak 32GB U1 C10 Normal
HP 32GB U4 Normal
Lenovo 64GB U1 Normal
banq 64GB U3 Normal

debug UART

Type Model Serial Print display
USB to serial port module CH340 normal output printing
USB to serial port module CH341 normal output printing
USB to serial port module CP2102 normal output printing

USB Interface

Type Brand Specification Test Results
USB flash disk Lenovo 64GB for USB 2.0 Normal identification
USB flash disk Aigo 32GB for USB 2.0 Normal identification
USB flash disk HP 32GB for USB 2.0 Normal identification
USB flash disk Toshiba 32GB for USB 2.0 Normal identification
USB flash disk Kingston 32GB for USB 2.0, 64GB for USB 3.0 and 128GB for USB 3.0 Normal identification
USB flash disk SanDisk 32GB for USB 2.0, 64GB for USB 3.0 and 128GB for USB 3.1 Normal identification
USB flash disk KIOXIA 32GB for USB 2.0, 64GB for USB 3.0 and 128GB for USB 3.2 Normal identification
USB flash disk Netac 32GB for USB 2.0, 64GB for USB 3.0 and 128GB for USB 3.3 Normal identification
USB flash disk Lexar 32GB for USB 2.0, 64GB for USB 3.0 and 128GB for USB 3.4 Normal identification


Type Brand Model Test Results
Television Xiaomi L75M5-4S Display normal but no sound
Display Lenovo T27q-20 Display normal and sound normal for 6 six minutes
Display SAMSUNG S27A350H Display normal and sound normal for 7 six minutes
Television SAMSUNG S32AM700UC Device is recognized but not displayed
Television SONY SONY/KDL-46EX720 Display normal but no sound

Would be great if you could add NVMe test results

I’ll keep updating when our integration group comes out more results.

1 Like

Please clarify if your UART tests are using the devices as a UART on this board, e.g. testing if Ubuntu is able to recognize these uarts as /dev/ttyWHATEVER and successfully transmit and receive when connected to a known device, or if the device is being tested the other way, controlled by some other computer and testing the TX and RX pins to use a serial console. Either BeagleV or V5v1 had an issue being cranky about 3.3v line swings. Thile technically, 3.3V isn’t part of the EIA-232 standard (neither is 5V or pretty much any TTL-like thing that doesn’t drive both signals strongly in opposite directions, ground-relative) it was a real grind to figure out. So please clarify if this is connecting the TX and RX pins to the board’s RX and TX pins or if this is connecting these chips, probably via USB, to connect a downstream device.

That HDMI category, as I read it, is pretty disturbing. Of five tested devices, only two will have a picture for more than 7 minutes and even they will have no sound. Or are we to understand that the vast majority works in high resolution in full 3D glory, with sound, for extended periods of time and that these are the only known exceptions at this time.

Hopefully, both can be addressed in happy/easy ways with “just” doc/spec corrections surrounding the testing performed.

Thank you for the update. Mine’s booked and I look forward to ti!

We use the usb-to-serial cable (picture below) to test debug UART. The dupont cable is cross-connected to the board, and used USB to connect to the PC (Windows). This debug UART supports 3.3V TTL voltage.


The HDMI testing are only for display function test, and we are still optimizing the sound.

1 Like

how many contacts do you connect board - 3 or 4 ?

3, connecting to TX、RX、GND. The Power (the red one) does not need to connect.

Thank you. That clarifies sufficiently.

Are the main JTAG and PORT console ports now sufficiently buffered and tolerant that 5v are likely to be damage and 3.3V’s may or may now work depending on the lenght of the cable (in CM, not meters and meters long) and the depth of the red zone between how high a low could still be and how low a might be in order to infer a proper stpread of voltages. This is a big problem for us on BeagleV when people started plugging random lab drawer cables and two “identical” cables (one with PL2303 and one with FTDI or CH330 (I think) would work completely differently. on the ports on the I/O bus (buffered pins!) and on the planar where they went almost directly into the 7110 at 1.8V.

Thank you!

Balbes150, I don’t directly conflict the staff of StarFive, but this sets off alarms for me. This cable isn’t passive It needs more that connecting wires in a rectangular shell to dupont-style connector. That chip needs to do do USB Stuff (enumaration, pipe multiplexion, error detection/ retransmission, hub power notifications, tx pipe, rx pipe, etc) and plain old UART-y things on the other, marking and detecting parity error, handling 5-8 data bits, stop bits, stop-and-a-half, break, and all that jass. You can argue that it’s not a super powerful chip and you wouldn’t be wrong, but I don’t want to build one out of 7400 series parts, either. There’s no charge pump inthese things; if you need more swing on the 232 side, you us serial BUS drivers and receivers to boost the signals UP to 12V (or more) but coming right out of these. I’m pretty sure the VOUT on TX is never going to exceed V unless is a traction of a volt coming from an external cap that they can get recharged in time… That’s just not something serial port chips do any more, is it?

Looking at that blue case, it looks a lot like an Adafruit 954. (it’s entirely possible that cas isn’t unique) depending on teh date of manufacturing.

Blockquote from Adfruit’s Product Manual, shows two posssible USB/Serial chips:
1 Prior to July 1 2014 this cable would come with a PL2303HX chipset and a black USB body. After July 1, 2014 we changed to a violet body and the PL2303TA chipset. The new 'TA cables are Win8 compatible, and are more reliable especially when dealing with random USB-disconnects.
2 As of Dec. 21, 2016 we will be shipping cables with SiLabs CP2012 chipset instead of Prolific. SiLabs chips seem to have better driver stability, and can handle very high speeds and ‘unusual’ non-standard baud rates. Other than the driver installation, the cables perform identically.

This body isn’t black,so we can disregard the earliest run, leaving us with PL2303HX and PL2303TA

I know from my own experience with chips that they sometimes CAN leak enough resistance from the R&D (I/O, bank of channel switches, whatever) up enough to look like the simplest cases worked. Debugging the hard stuff because you didn’t actually plug in (or fix the power supply to…) The Thing can make you crazy. Because of the diversity of serial speeds, a mark and a break can keeep the line almost in a steady state for the length of a character time, repeated long time.

Section 9.0 of the book shows us that older cables will have PL2303HX to provide energy and it has mixed messages.VDD_5 and VO_33. These are biased. Anything on the I/O bus from in the rang e can be converted rom that. The exercise is thinkingm, though, no faking, so I don’t get who looks good and who looses badly there,

One point of open source races, gamges, cond contestants id that anyone (qualiffied) should be able to yell O “whoa? WTH?” and get a restart after an evaluation of mechanics . That didn’t seem to happen here.

Collectively, the SiFive/IceFive/JacksonFive/MaroonFive teams could have had enouh hint from early though in the game to simply wirthdraw, pool their worksfor a year, probably won with the second most fast and kept Top Secrtet for starting next year, It’s OK to hold a bit back. It can’t stay there forever. Those jnw jockey boots fo spiking the dars of drivers gone past may best be forgotten about. :stuck_out_tongue: