Numbers on the ATINY1606

[Rollpopptery]

The numbering on the ATtiny does not match the documentation.
Some ICS have TINY1606F
Some ICS have TINY1606N

The meaning of the 'N' and 'F' are not described
It does not match the part numbering in the datasheet.

[pcfreak1201]

N means only Industrial-Grade (40C to +105C)
F means only Extended-Grade (-40C to +125C)
They are only preselected during manufacturing process / QC. Have a look:
https://www.microchip.com/en-us/product/attiny1606#Purchase

[Rollpopptery]

Thanks.
Specifically I had issue with serial UART at 115200 baud having errors on a number of micros.

Fixed by using 16 MHz CPU clock.
Our previous testing show that 16MHz clock is always accurate on these chips.
But 20MHz derived clocks (20, 10, 5) can be inaccurate, (slightly slow)

[pcfreak1201]

Did you read this?
https://github.com/SpenceKonde/megaTinyCore/blob/master/megaavr/extras/Ref_Clocks.md

[Rollpopptery]

Yes.
The 16MHz clock is accurate. never need to fiddle with registers

[SpenceKonde]

When having issues with serial that indicates a slight mismatch (particularly, when 1 character comes through okay, but but things go to hell if it receives a block of characters in immediate succession), the side that isn't receiving gibberish (that's the fast side) should change their serial settings from SERIAL_8N1 to SERIAL_8N2. If the slow side does as well, no harm other than a reduction of 1/11th of the data transfer rate results - the second stop bit (except in extraordinary repurposings of the USART) doesn't have the risk of breaking transfers that would otherwise work I recommend using 2 stop bits whenever possible.

I'm sorely tempted to make that the default. I bet nobody would notice it was different from stock

[SpenceKonde]

I will also note that, based on extensive measurements, the 16 and 20 MHz oscillators have similar error. It's very nearly 50:50 which one is closer.