What are you doing with Padauk chips you bought?

[KaeTo]

Just finished another little padauk projects. It uses the PMS150C. It's a battery protection circuit (li ion) which cuts of the battery if voltage is under 3V. Futhermore a led indicates this state (blinking every 6s) and shows the actual state if a device is connected (blinking 1x: <3,6V, 2x: <4,0V, 3x: >4,0V). Only 5 components are required (PMS150C, n-ch Mosfet, p-ch Mosfet, schottky diode, LED). I think with this a price under the standard protections boards could be achived.

[c64]

I noticed  LCSC have these parts in stock

https://lcsc.com/search?q=Nyquest

https://www.nyquest.com.tw/en/product/HomeAppliances/NY8_Series/download

They are in a similar slot, to Padauk parts, 14b opcodes and 8 level stack,  and OTP with HV Vpp in SO8 and SO14 and SO16  - from Taiwan, so the English support looks good.
They also look to have a "C compiler " and some MTP ones are coming.

Some have 12b ADC and can support crystals - which is rare for bottom-feeder MCUs, so they seem well suited to simple peripheral slaves and timer tasks, around a larger FLASH MCU.

Are where any cheap programmers for those?

[danymogh]

I noticed  LCSC have these parts in stock

https://lcsc.com/search?q=Nyquest

https://www.nyquest.com.tw/en/product/HomeAppliances/NY8_Series/download

They are in a similar slot, to Padauk parts, 14b opcodes and 8 level stack,  and OTP with HV Vpp in SO8 and SO14 and SO16  - from Taiwan, so the English support looks good.
They also look to have a "C compiler " and some MTP ones are coming.

Some have 12b ADC and can support crystals - which is rare for bottom-feeder MCUs, so they seem well suited to simple peripheral slaves and timer tasks, around a larger FLASH MCU.

Are where any cheap programmers for those?

I've been eyeing those "Nyquest" parts as well. they are even cheaper than padauk now. the problem is the programmer as you mentioned. I think i saw them selling off at around 100$ at aliexpress https://www.aliexpress.com/item/1005004560890408.html which is a lot. however looking at the datasheet and the modified SDCC compiler, I have a strong hunch that these are other FMD like MCUs with PIC-like architecture, but can't confirm because you'll have to look at the final assembly.

the problem with FMD and PIC alike MCU is that they didn't fully clone the PIC architecture so no compatible compiler exists for them other than the one the company is providing. HOWEVER! , I've seen origin-gd (which is just a rebrand of FMD) that uses Hi-Tech compiler for PIC with modifications to the final assembly! they later got rid of the Hi-Tech compiler and rebranded it as some kind of self-made gnu compiler packed with a modified UPX packer to hide its presence! I guess because of copyright reasons? and FMD stopped shipping its IDE with the Hi-Tech compiler at some point.

anyway long story short, I think it's possible to reverse the programmer (it's much simpler than padauk) and make it as cheap as possible.

[c64]

I noticed  LCSC have these parts in stock

https://lcsc.com/search?q=Nyquest

https://www.nyquest.com.tw/en/product/HomeAppliances/NY8_Series/download

They are in a similar slot, to Padauk parts, 14b opcodes and 8 level stack,  and OTP with HV Vpp in SO8 and SO14 and SO16  - from Taiwan, so the English support looks good.
They also look to have a "C compiler " and some MTP ones are coming.

Some have 12b ADC and can support crystals - which is rare for bottom-feeder MCUs, so they seem well suited to simple peripheral slaves and timer tasks, around a larger FLASH MCU.

Are where any cheap programmers for those?

I've been eyeing those "Nyquest" parts as well. they are even cheaper than padauk now. the problem is the programmer as you mentioned. I think i saw them selling off at around 100$ at aliexpress https://www.aliexpress.com/item/1005004560890408.html which is a lot.

Yes this sucks. Price is OK if you already use them. But if you just want to buy few of these chips to test, it's hard to justify $100 for programmer and $1 for chips

[brucehoult]

I noticed  LCSC have these parts in stock

https://lcsc.com/search?q=Nyquest

https://www.nyquest.com.tw/en/product/HomeAppliances/NY8_Series/download

They are in a similar slot, to Padauk parts, 14b opcodes and 8 level stack,  and OTP with HV Vpp in SO8 and SO14 and SO16  - from Taiwan, so the English support looks good.
They also look to have a "C compiler " and some MTP ones are coming.

Some have 12b ADC and can support crystals - which is rare for bottom-feeder MCUs, so they seem well suited to simple peripheral slaves and timer tasks, around a larger FLASH MCU.

Are where any cheap programmers for those?

I've been eyeing those "Nyquest" parts as well. they are even cheaper than padauk now. the problem is the programmer as you mentioned. I think i saw them selling off at around 100$ at aliexpress https://www.aliexpress.com/item/1005004560890408.html which is a lot.

Yes this sucks. Price is OK if you already use them. But if you just want to buy few of these chips to test, it's hard to justify $100 for programmer and $1 for chips

That's where the $5 CH32V003 kit is much better value. You get 5 bare chips, a USB programmer/debugger, a dev board (with 6th chip on it).

https://www.aliexpress.us/item/3256804709476544.html

You can buy another 50 chips for $5:

https://www.aliexpress.us/item/3256804850399956.html

- modern 32 bit RISC ISA, better code size than PIC, executing fewer instructions to do the same task. Supported by upstream gcc and llvm, not customised gcc or limited pseudo-C
[1]

- flash, not OTP

- 2k RAM not 32 bytes

- 16k flash (5000-6000 instructions), not 1000 instructions. Probably equivalent to 12k-15k PIC instructions.

See Dave's recent video

Note: both items are out of stock right now, as they have been multiple times before. They were in stock a few days ago, and no doubt will be again soon.

[1] WCH do have a custom compiler, but it's not necessary, even if you want to use their non-standard features, with some simple use of attributes and inline asm.

[hans]

Not to mention that probably those RISC-V chips will still be very usable by the same GCC mainline tools (from your favorite linux package manager) in 10 years time. The Padauk and whatnot needs completely custom toolchain stuff, that on a bad day, only work with an old version of Windows (huray for rolling Windows 10/11 releases now huh), so you'd rather archive in a VM..

Hardware programmers are a fairly standard serial or SWD programmer it seems. Is there also debugging? That would make it even greater. The OTP aspect of the Padauk chips put them off for me. Sure ct-level MCU is great, but not if it requires soldering to fix a firmware mishap that sucks (which, let's face it: at some point; the firmware works, while at some point, the hardware used to work).

The only disappointment for RISC-V MCU is the power consumption. 0.4 - 2.7mA sleep you got to be kidding me. But to be honest not much worse than some other PICs that were marketed with XLP branding, or perhaps, all eastern "clones" that primarily focus on function compatibility over electrical compatibility (in particular bad analog behaviour)

[brucehoult]

Hardware programmers are a fairly standard serial or SWD programmer it seems. Is there also debugging? That would make it even greater.

There is "SWD" debugging, yes. It's not the same as ARMs, you need WCH's customised OpenOCD, but the source code for their version is published (but not upstreamed). You kind of have to ask WCH for it and they easily give it to you, but (multiple by the looks) people have put it on github. The license certainly allows that. Someone needs to organise that, but yeah...

https://github.com/kprasadvnsi/riscv-openocd-wch
 

The only disappointment for RISC-V MCU is the power consumption. 0.4 - 2.7mA sleep you got to be kidding me.

Maybe some of WCH's more pricey MCUs have better sleep. I haven't looked.

I know the GigaDevice GD32VF103 that has been shipping since 2019 (I bought a dozen Longan Nano boards for $4.90 each back then, and some bare chips from tme.eu for $1.98 each in October 2020) has "sleep", "deep sleep", "standby" modes:

sleep: 108 MHz clock: 10 mA peripherals off, 25 mA peripherals on; 2 MHz clock: 1.0 mA peripherals on, 0.7 mA peripherals off. Wake time 4.5 µs.

deep-sleep: 460 µA, wake time 6.0 µs

standby: 22 µA, wake time 118.8 ms

I believe those are reasonable compared to STM MCUs, except the wake from standby (and from power on reset is the same) is long because the program is copied from in-package (2nd die) flash into 0 wait-state SRAM for execution.

So I dunno. The sleep figures you quote for the WCH are pretty comparable. I don't know what you expect. I've never really had to deal with really conserving power like that.

[SiliconWizard]

Ultra low power modes are not the forte of low-cost chinese MCUs in general, with few exceptions. And RISC-V ones tend to be worse for now, probably due to market reasons (vendors are probably still looking to adjust their positioning) and maturity of the RISC-V cores they use.

But again, not limited to RISC-V MCUs. And even the RP2040 suffers from that.

What we take now for granted, standby modes with current draws in the µA range or lower, is actually a pretty new trend. A few years back, it was still niche and you'd need to select specific vendors / product lines for that.

[tim_]

Ultra low power modes are not the forte of low-cost chinese MCUs in general, with few exceptions. And RISC-V ones tend to be worse for now, probably due to market reasons (vendors are probably still looking to adjust their positioning) and maturity of the RISC-V cores they use.

A bit ironic to post this in the Padauk thread. The Padauk MCUs are actually quite capable low power devices, see here:

https://cpldcpu.wordpress.com/2021/02/07/ultra-low-power-led-flasher/

Most of the low-cost chinese ARM / RISC-V are manufactured in smaller and lower voltage CMOS which are inherently more leaky. Allowing for a very low current standby mode would probably require adding low leakage transistors for power gating. This does usually increase wafer prices - hence we end up with fairly mediocre MCUs. I would also assume that they do not spend too much time on optimizing the analog aspectes of their MCUs in general, as this requires a different skillset and is quite time consuming.

[brucehoult]

I think we can all agree a 32 bit CPU is overkill if you only want to flash a LED.

Which reminds me -- I have to change the battery in my "Pulse" CD box set again.

[Nominal Animal]

PMC251 might be suitable for HID buttons and rotary encoders, using SPI (chain) as a shift register for accessing their state, plus a separate latch signal for sampling all in parallel.

Essentially, PMC251 can be made to run two "cores", interleaved, on every other system clock cycle; each "core" has their own program counter, stack pointer, accumulator, and flags.  One "core" would monitor the encoder state(s) and latch signal, and the other "core" handles the SPI shift register.  (The latch transition simply preloads the SPI shift register with the encoder state.)  Running at 8 MHz, this should be fast enough so that no transitions are lost and debouncing can be done in software.

At €0.17 or less apiece (with each capable of handling two encoders or seven buttons), it'd make a nice way to make human interface buttons and encoders for a number of gadgets, if one designs the data stream sensibly.  Changes since last latch might be most useful (with buttons reporting both press and release events so that a button press is not lost even if it happened entirely between two consecutive reads), because it'd allow encoder acceleration profiles also, and the per-encoder state wouldn't need too many bits.