AVR Freaks

to gchapman: Are you on some sort of mailing list that you get announcements for all these new documents that you post?  Is there some forum or web page at Microchip where they are announced?

(In any case, thank you for reposting them here, attached to relevant threads!)

I looking forward to -DD because of the package size. 4mm QFN vs 5mm makes *a lot* of difference to me. Then I can drop SAMD11 which is too memory-constrained (although it does have USB, if you can afford the memory use).

Coupon codes are nice, but can anyone actually buy anything from them?

I have been trying to buy the DB48 eval kit for the last hour, but I can't get past the checkout. It says there is something wrong with one or more fields in my address, but all is fine when I look. The address is the same as last time I ordered from Microchip Direct, and back then there was no problem.

Sigh.  Emotionally, I really "want" an 8pin replacement for the ATtiny85.  At least 8k of program space, DIP8/SOIC8 and cheaper than the t85.  AVRxxDy8 or xTiny, I don't care.

(logically, I realize that many of the existing new chips already replace t85 WRT board space (in those tiny QFN packages, etc.), but it doesn't stop the "want.")

Better a wireless interface.

Has anyone experimented with overclocking an AVR-Dx? https://github.com/SpenceKonde/D... mentions that while the max rated speed is officially 24 MHz, speeds up to 32 MHz are "an overclock that is likely to work on all parts as long as it is run at room temperature" and even 40 MHz should work on "most parts" if you're making a one-off project. Somewhere else, I saw speculation that AVR-Dx was originally conceived as a 32 MHz part, but it couldn't quite meet that across its entire temperature and voltage range, so the official max speed was reduced.

I'm curious if anyone else has played with this, and which peripherals or behaviors tend to fail first when overclocked.

This is not for serious projects. In the vast majority of cases, frequencies below 10 mhz are really sufficient...

Thanks, I'm really interested in hearing any hands-on experience with overclocking the AVR-Dx series to 28 or 32 MHz. Sure, in an ideal world this would never be necessary, and I would just use a 120 MHz ARM or something, but it's helpful to have first-hand info about how AVR-Dx behaves above 24 MHz. If that DxCore document is reliable, 32 MHz is readily achievable under favorable voltage and temperature conditions.

Why? I'm revising an ATMEGA1284 design due to lack of available parts. AVR128DB makes the existing code easy to port, which is the most important factor to me. It's actually available right now in decent volumes, unlike so many other MCUs. 24 MHz operation will work to duplicate the existing design's capabilities, but 28 or 32 MHz would help it work even better under certain situations where it sometimes struggles now. So I'm considering an optional overclock setting. It wouldn't need to work 100 percent of the time, but it would be extra gravy if it does work most of the time.

I have a board with a 25MHz clock available (at 3.3V) because of an ethernet onboard. Thinking of overclocking an AVRxxxDx at 3.3V too, with the external 25MHz wave.

Real life temperature corners are 0C...50C, lets say -10C...60C.

The main question in OC is whether flash access will work or not. But I doubt it will not work at 25MHz given it will at 24MHz :-)

Still, I'm not very comfortable getting out of specs...

I asked SpenceKonde (the DxCore guy) about his overclocking tests. He said they were all run with a supply voltage close to 5V. But he believes that the supply voltage doesn't really matter, because the Dx AVRs have an internal voltage regulator for the core, so the core will run at the same voltage no matter what the supply voltage is. He said he runs at 32 MHz very often, and has neopixel controllers with LCD screens for UI that run at 40 MHz. In his experience, when increasing the overclock, the ALU fails first. So that's a very bullish overclocking report.

Do what you will with this information... personally I don't think I'd build a commercial product that requires running beyond the datasheet's maximum clock speed spec. There are some AVR projects that do this, though, based on ATMEGA chips. Two I can think of are JeeNode and Uzebox. But as an optional feature, I find the idea of a 32 MHz Dx pretty interesting and exciting. I'll definitely be experimenting with this more.