AVR Freaks

El Tangas, I think you should have bite of the Cake $$$ ;)

EDIT: typos

well, 7$ for a chinese kit is quite a lot also...not to say its bad quality...but...

@El Tangas,

Yes,  it would be nice if the "Nano Every" gets cloned by the Chinese.

They just replaced the Atmel USB chip with a CH340 in the traditional Nano.

The ATmega328P bootloader remains the same.

The punter gets exactly the same operational behaviour with genuine or clone Nano.

They could replace the SAMD10 with a similar Chinese MCU to give exactly the same operational behaviour

A genuine ATmega4809 is always used.    This would be their main BOM item.

I would guess that the Chinese would prefer to use a CH340 instead of the genuine Atmel SAMD10 chip for USB.

This would mean a different Bootloader and different Arduino IDE "board" description i.e. fail as a 100% clone.

David.

and then the OP say they dont use AVR. well, O_o

I think thats why people should stay away from these companies....

Interesting...what should it be ?

I don't care what USB chip they use.    As long as it gives exactly the same operational behaviour

I don't like the idea of scrubbed chips.    They can keep the USB firmware private to their own company.    The 328P Nano clones do not publish the CH340 source code (even if trivial)

And yes,   of course the innovative Chinese company deserve to be profitable.

The SAMD10 seems to have the same price as a mega4808.  (which is slightly cheaper than the 4809)

David.

My apologies.   I wrote SAMD10 from memory.   The SAMD11 seems to be a similar cost.

Yes,   I am sure that there are several ways to get USB-Serial with 5V tolerance.    And if a Chinese company can produce a cheaper product it seems a good idea (tm).

The EVERY has a full 48kB of Flash for user applications.

If it had been a regular AVR + USB-Serial design,   you lose 1kB for an Optiboot-style bootloader.   e.g. ATmega4808 + CH340.

As far as I can see,   the mega4809 + SAMD11 design has "UPDI programming" but no "debugging".

The world would be a better place if all electronics was 3.3V.

David.

Edit.  Just ordered one from Thinary (AliExpress).   I will see how well it compares with the genuine EVERY   (that arrived promptly from Italy/Holland)

In an ideal world it will work exactly the same.

In the real world it might take 30 days to arrive.   (if not longer)

AliExpress is a mystery.    Recently orders have arrived in 8-9 days.    Others take weeks.   AliExpress pages change to random language even if you Sign In as a known English speaker/country customer.

David.

The Thinary 4808 EVERY arrived this afternoon.

The Megaavr Arduino Core expects a 4809 but the "upload" finds a 4808.

So I added an extra section to boards.txt which adds -pATmega4808 to upload.extra_params

##############################################################

nona4808.name=Arduino Nano Every 4808

nona4808.vid.0=0x2341
nona4808.pid.0=0x0058

nona4808.upload.tool=avrdude
nona4808.upload.protocol=jtag2updi
nona4808.upload.maximum_size=49152
nona4808.upload.maximum_data_size=6144
nona4808.upload.speed=115200
nona4808.upload.use_1200bps_touch=true
nona4808.upload.extra_params=-P{serial.port} -pATmega4808

nona4808.build.mcu=atmega4809
nona4808.build.f_cpu=16000000L
nona4808.build.board=AVR_NANO_EVERY
nona4808.build.core=arduino
nona4808.build.variant=nona4809
nona4808.build.text_section_start=.text=0x0
nona4808.build.extra_flags={build.328emulation} -DMILLIS_USE_TIMERB3 -DNO_EXTERNAL_I2C_PULLUP
#nona4808.build.extra_flags=-B{runtime.tools.atpack.path}/gcc/dev/{build.mcu}

nona4808.bootloader.tool=avrdude
nona4808.bootloader.file=atmega4809_uart_bl.hex
nona4808.bootloader.SYSCFG0=0xC9
nona4808.bootloader.BOOTEND=0x00
nona4808.bootloader.OSCCFG=0x01
nona4808.fuses.file=fuses_4809.bin

menu.mode=Registers emulation
nona4808.menu.mode.on=ATMEGA328
nona4808.menu.mode.on.build.328emulation=-DAVR_NANO_4809_328MODE
nona4808.menu.mode.off=None (ATMEGA4808)
nona4808.menu.mode.off.build.328emulation=

##############################################################

Sketches obviously build ok since they think it is a 4809

It seems to upload ok via -c jtag2updi as a 4808

However the sketch does not run.

@El Tangas,

Any ideas?

I will try an AS7 project later.

David.

Edit.   I built a binary for ATmega4809 in AS7.   It will upload via avrdude -c jtag2updi -p ATmega4808.   It does not run after a successful upload and verify.

I built a binary for ATmega4808 in AS7.   It does not run after a successful upload and verify.

I will try tomorrow to debug with ATMEL-ICE and UPDI pin.

IDK. If the verify checks out, the program should run, I suppose. But Thinary doesn't use my jtag2updi version, so I can't be sure.

Also, I never tested jtag2updi with the Mega AVR-0 series, a bug is always possible.

edit: Ok, I opened the Arduino IDE and loaded the "blink" test sketch on a bare mega4809 (PDIP version), using my current version of jtag2updi (it's not released yet, but the only change is support for LGT8F328P boards). The programmer is a LGT8F328P board. I told the Arduino IDE this setup was a Nano Every.

It's working fine, but it's just a small program, a bigger one might fail especially one bigger than 32kB if there is some numeric bug, we never know...

Anyway, the Arduino IDE keeps writing fuses and erasing the whole chip at every upload cycle. I don't know if this is the best option... I wrote jtag2updi to erase only the pages that are written to speed up things and not stress the flash so much, but whatever...

Recently, the manufacturer of the classic CH340G has introduced new models that don't need a crystal, the CH330N and CH340E, this means less space and probably less money.

For the UPDI interface, the clones could use a LGT8F328P running jtag2updi (upcoming version).

This 2 chip interface solution would cost about $0.50 ($0.25 each chip, TaoBao prices), if we add the cost of the mega4809 that would be a total of $1.50 in silicon. I'm sure the rest would cost less than $1 judging from the other stuff that comes from China. Therefore, manufacturing cost: $2.50.

Selling the whole thing (Nano Every drop-in clone) for $4 or so should still turn a reasonable profit, if it becomes very popular.

edit: problem is, the Nano Every is already quite cheap at €8, even if the clones cost 1/2 I'm not sure people would go for it.

Yeah, there's no way they follow that rule in China.

The Nano Every apparently implements the USART bridge like this (I'm speculating since I don't own one, just observing what the IDE does):

Normally, the SAMD11 acts as a USB/serial bridge connected to the USART0 of the Mega4809. Before programming starts, the IDE sends a DTR pulse of 20 seconds (why so long?), that the SAMD11 interprets as a signal to enter programming mode, that is, instead of USB/USART0 bride, it becomes a USB/UPDI bridge.

I think I should try to implement this behaviour in future versions of jtag2updi.

Thanks for the info. The manufacturing cost is a bit higher than my estimate.

Yeah, the manufacturing cost of the Thinary board at least makes sense. But this for example: https://www.aliexpress.com/item/...

A LGT nano clone for $1.30 incl. delivery, how can it be possible?

No such things around here, probably the size of the market doesn't justify it. Normally I buy "big ticket" items (well, actually anything over €20) from such warehouses, usually located in the UK, Germany or Italy.

For me the main advantage is that those sellers take care of customs taxes and VAT, I don't care how, it's their problem. What matters is that I buy from an EU store so that customs don't bother me.

According to this https://www.avrfreaks.net/commen... you are supposed to use their core.

Problem is, they don't tell you where to get it... I'm guessing it's here:  https://github.com/Thinary/ThinaryArduino/releases

Well, maybe this board does not emulate the Every behaviour correctly. The programmer chip is supposed to pass the UART signals by default. Only when a DTR pulse is received, is it supposed to enter programming mode.

Silicon CPU "cores" have run at low voltage for at least 40 years.    The outside world accesses the core via the GPIO drivers.

The core will run at whatever voltage suits the manufacture, power, speed, ... requirements.

I suspect that the outside world will remain at 2.8V - 3.3V or at least be tolerant.

A lot of electronics is NOT tolerant of 5V.

David.

Since you have an ATMEL-ICE everything should work quite nicely.    UPDI is much nicer than debugWIRE.

There is a trough-hole terminal for UPDI on the Thinary.   The Nano-Every has a tiny SMD pad.

Mind you,   debugWIRE works fine on genuine Uno  (or clone with RESET-EN solder-bridge)

But only works if you hack a Nano (328P type).   You must remove both 1k0 pullup and 100nF

I will dig out my Thinary.  And see if I can remember what I did.

Incidentally,  when did you buy your Thinary?

It looks like they sold a handful in Aug-Sep 2019 and then quietly dropped it.

David.

Thanks for the link.   

 

9898 pieces available

It looks as if they manufactured a batch of 10000 in August 2019.   And have sold 192 so far.   Hence the attractive $4.89 price.

to United Kingdom via Cainiao Super Economy Global

Estimated Delivery: 30-50 days

Horrifies me.    Mind you,   Chinese packets generally take 8-18 days to get to the UK.   You might have to wait 50 days for a "no delivery".

If you are buying several boards it is worth paying for tracked shipping.

David.

It would be nice to have a feature in JTAG2UPDI that caused it to switch to some sort of Serial Passthough mode after you were done programming the chip, until some sort of signal were received that put it back into UPDI mode.

This is what the SAMD11 on the genuine "Every" does, except that it has USB on one side, whereas it'd be nice to have a defined method with serial on both sides.  (heh. Rx/Tx/DTR on one side, Rx/Tx/UPDI on the other.)  I assume that this is essentially what the Thinary board is doing with its mystery chip...

I try to draw pinout of this board

Could you check it and tell me what's wrong?

I would only label pins with functions provided by your boards.txt file.

For example, you have Ain8-Ain12 in the diagram, but your boards.txt only maps A0-A7.

Please read #53 and other messages in this thread.

Yes,  the Thinary board works.   The Arduino "Thinary" core has a few problems.   It has never been updated.   Look at the GitHub repository.

Thinary do not seem to respond to Issues or Pull Requests.

Note that the pin usage is very different to the genuine Nano-Every.

You can see the usage from the pins_arduino.h and variant.cpp files

It is probably worth the current $4.89 sale price if you know what you are going to do with it.

And it would help Thinary to recoup some of the cost of assembling their first batch of boards.

It is unlikely that they will ever be able to shift the whole batch unless the "Thinary Core" worked 100%.

So there is little point in making a nice graphic for the pinout.

David.

Hi All, will share my experience with this board. 

The price is attractive, I agree here, but other than that I'm frustrated about it, it is a waste. Making "blink" work with Arduino IDE is mostly all you can do in a way you used to. For a more complex solution, you will mostly be struggling with various issues. The Serial thing one of them. Another major pain point is the lack Platformio integration. Also do not expect any support from a manufacturer, they're github repo is dead but people seem to be still posting issues there.

The upload works (after importing package index). Fully agree, the board looks really promising, but in the current state is a total waste.

Some update here. So I tried to give it another chance, but it's just very hard to work with, so I give up on this board. 

I got STM32 (STM32F103C8T6) a.k bluepill to experiment with and it looks really promising so far. It is 2x cheaper (around 2$ on aliexpress) however it has much more RAM which (as well as other resources) is a key for me (how come I didn't discover that board earlier....? :)). It was easy to start working with ArduinoIDE and most importantly a Platformio. It has failed to compile my project due to some Arduino framework incompatibilities, but that is not a big issue. Another issue is it's 3.3v, but I have my other hardware on a 5v ATM, so the signal converter will have to be used.

>  if I were to buy regular nano clones, I'd probably get them from these guys, at least they have a brand to protect

Personally, I have added their aliexpress store onto my blacklist. I'm not the only one who is struggling with this board, there are other posts on the internet that are complaining of the same set of issues. If they were to protect the brand, they would've tried to mitigate such a poor user experience. They claim it's Arduino compatible, but it's not, this is virtually a fraud. 

Yes,   they designed the board and wrote the Thinary Core to support it.

I suspect that they manufactured an enormous batch before even announcing it.    Just look at the "number available".

It would have been far safer to just copy the NANO-EVERY4809 design but with CH340 instead of SAMD10.   i.e. just like the Uno and Mega2560 clones.

Creating a NANO-4808 board with completely different pinout was "brave".    It could have been successful if it worked 100%.

Or even if it just came with Optiboot bootloader.  e.g. like a NANO-328 but with 4808/4809

There is still a chance that another Chinese company creates a NANO format board.   Perhaps with AVR128DA32 and CH340

However entering a race to the bottom (like Nano-328 or USBASP) is hardly a good business model.

You need a big market.    (and make sure it works 100% first)

David.