Has anyone here tried the revB? Are there any actual differences or improvements vs. the original?
1 Like
I doubt that anyone outside SiFive without an NDA has one yet. Mouser says they are expected to arrive Jan 22, 2024. But you can learn some info by looking at the Getting Started Guide and schematic changes.
There is a new auto-restart DIP switch and associated circuitry. That is useful for anyone that wants to run it as a server, and doesn’t want wire up a remote control relay to do the auto-restart that the original board needs. There are two headers for chassis fan control. That avoids the USB powered fan that the original board needs. There is a jumper for RTC frequency control that may not be useful to anyone outside SiFive. There is a power regulator that has changed to a different part. There are some misc minor resistor, capacitor, diode, transistor, and jumper changes. Otherwise, it looks like the same board. Some of the ICs on the board might be newer revisions of the same part. That kind of info isn’t on the schematic. But it looks like the exact same SiFive CPU as the original board.
2 Likes
It doesn’t appear to me that the issue of software reset has been sorted though, so presumably a remotely controlled relay is still required for that unless one is happy to power-cycle the system via a remotely controlled PDU every time instead.
I’m curious as to the extra jumper pin headers for fan connector isolation though, it seems to me that they add no value. IMO they’d better be used to optionally isolate the UART0_RX and UART0_TX lines from the FTDI USB serial adapter chip (or to strap the OE input of the U5 voltage converter chip so as to place it in the high-impedance state) so that the serial console could be driven directly via an external RS-232 transceiver chip wired to the relevant pins of J8.
What’s the matter with a USB powered fan, what is it needed for? I have two 12cm chassis fans in my Rev A system operated from the two onboard fan connectors each and the setup appears to run as expected.
1 Like
I’d be interested to know whether memory map is the same, specifically peripherals and their default/reset states; and if any of the Mask ROM and/or OTP code has changed?
The announcement says that the new board is software compatible with the original one, so there should be no memory map change or peripheral state changes, or at least none that affect the linux kernel. The mask rom is on the CPU isn’t it? In that case, it didn’t change as that would be far too expensive. As for OTP changes, I can’t guess, but seems unlikely.
1 Like
The notice says the fan connectors are new. I didn’t double check. One of the chips gets rather warm under load. I think it was the pci bridge (switch?). I put a fan on it to feel better. But maybe I just don’t have a good enough setup for my board.
1 Like
Ah, I can see what the matter is.
Updated GSG misleadingly says these are Chassis Fan Enable Header 1 and 2, however upon checking the schematics (sheet 6, BTW) I can see they’re really fan PWM enable headers. By positioning the jumper in either pin header position suitably you can select between driving fan’s VDD from the output of the associated PWM generator or directly from VDDX_12V0. Of course you can leave the jumper out entirely, causing any connected fan not to operate, but I don’t think it’s the primary function.
I guess it makes some sense, though I find it quite unusual: none of the mainboards I have ever seen had such a feature, as to drive a fan directly from 12V you can just use a suitable adapter cable, readily available to purchase from various sources, to get the fan powered from one of the ATX PSU’s standard storage power plugs rather than placing extra components on a potentially expensive mainboard.