Did you mean RP2350 ?
At first I was confusing these with the āPIO state machineā cores, but they are accessible via ARM MCUs too.
Canāt wait to see a mini-version of the board alĆ” waveshare pico zero.
I can live without current sensing if the board is smaller.
How much do we gain by real current sensing compared to voltage controlled current (kV and Ri based)?
96 cycles??? This is so long, itās uncomparable even to ATMega328p (15 cycles single conversion). Why is it so long? Most STM32 are also in this region, 10-20 cycles.
Well, to be fair it does also depend on the clock speed, a cycle isnāt a constant unit of time you can compare across MCUs. ATMega has 16MHz clock, but the ADC shouldnāt be clocked faster than 250kHz or so - while the RP2350 ADC is clocked at 48Mhzā¦ so even though it takes 96 cycles to convert compared to 15 on the ATMega, that is still 30x faster, if I got that math rightā¦
I thought that cycles are more fair to compare than clock speed, because it should be related to the efficiency of the chip design? The G4 can also do 12.5 cycles (main clock 170MHz but ADC clk / 4, so 42MHz)
Indeed youāre right that is certainly an indication of more (performance)-efficient design.
I guess it comes at a tradeoff for space on the silicon.
RP has features like the PIO, Dual Core, alternative RISC-V etc which STM and ATMega donātā¦ those were given priority over ADC functions it seems.
Iirc the double fpu things behave/interface more like the stm32 cordic unit rather than a normal FPU.
Also, I think the riscv cores donāt have any hard floating point support.
Anyhow, I ordered a couple of them, but they were out of stock so Iām not sure when Iāll get them.
1 Like