Skip to content

Latest commit

 

History

History
543 lines (483 loc) · 41.5 KB

Ref_Defines.md

File metadata and controls

543 lines (483 loc) · 41.5 KB

Identifying menu options and part identity within sketch or library

It is often useful to identify what options are selected on the menus from within the sketch - one of the most critical is the millis timer the millis timer selection. You may know that you require a certain configuration and want to prevent accidentally compiling with the wrong option and wasting time chasing that instead of a more meaningful issue. Submenu options are not saved on a sketch or board basis, but an IDE-wide basis so you can easily get them mixed up.

You cannot check for options that depend on the fuses at compile-time, because we don't know what they will be - sketch can be uploaded separately from setting them.

Note: Most of this document is the same for DxCore and megaTinyCore. This both makes my life easier, and likely yours as well, since these are most useful when writing code that works across a wide range of parts; there's no reason we need to send you looking at two versions of this file. Only the "deprecated" constants are not universal.

Timer identification from within code

You may need to know programmatically what millis source is in use. There are several defines to help with this. There are the legacy ones:

  • MilLIS_USE_TIMERA0
  • MILLIS_USE_TIMERA1
  • MILLIS_USE_TIMERB0
  • MILLIS_USE_TIMERB1
  • MILLIS_USE_TIMERB2
  • MILLIS_USE_TIMERB3
  • MILLIS_USE_TIMERB4
  • MILLIS_USE_TIMERD0
  • MILLIS_USE_TIMERE0 - for the unreleased AVR EB-series. This may be removed upon datasheet release if it is apparent that TCE is not suitable for millis.
  • MILLIS_USE_TIMERF0 - for the unreleased AVR EB-series. This may be removed upon datasheet release if it is apparent that TCF is not suitable for millis.
  • MILLIS_USE_TIMERRTC
  • MILLIS_USE_TIMERNONE

It is often more convenient to use these newer macros, which identify instead testing the millis time,.

Millis timer and vector

  • MILLIS_TIMER
  • MILLIS_VECTOR

Millis timer codes

These are the values that the MILLIS_TIMER may be defined as:

Name Numeric value Meaning
NOT_ON_TIMER 0x00 Millis is disabled
TIMERA0 0x10 Millis is generated from TCA0 (All non-EB-series)
TIMERA1 0x08 Millis is generated from TCA1 (DA*, DB*, EA only)
TIMERB0 0x20 Millis is generated from TCB0 (all parts)
TIMERB1 0x21 Millis is generated from TCB1 (No tinyAVR 0-series or 1-series with 4k or 8k of flash)
TIMERB2 0x22 Millis is generated from TCB2 (DA, DB, DD*, EA only)
TIMERB3 0x23 Millis is generated from TCB3 (DA*, DB*, EA only)
TIMERB4 0x24 Millis is generated from TCB4 (DA*, DB*, only)
TIMERD0 0x40 Millis is generated from TCD0 (megaTinyCore only)
TIMERE0 TBD - undefined Millis is generated from TCE0 (EB-series only).
TIMERF0 TBD - undefined Millis is generated from TCF0 (EB-series only).
TIMERRTC 0x80 Millis is generated from the RTC from internal low frequency osc. (megaTinyCore only currently)
TIMERRTC_XTAL 0x81 Millis is generated from the RTC from external crystal oscillator. (megaTinyCore only currently)
TIMERRTC_XOSC 0x82 Millis is generated from the RTC from external active oscillator. (megaTinyCore only currently)

To to find the millis timer:

if (MILLIS_TIMER & 0x40) {
  //timer D (not available on DxCore)
} else if (MILLIS_TIMER & 0x10) {
  //TCA0
} else if (MILLIS_TIMER & 0x08){
  //TCA1
} else if (MILLIS_TIMER & 0x20) {
  //timer B, look to othr nybble to find which one
} else if (MILLIS_TIMER & 0x80) {
  //RTC
}

Timer identifier interpretation (digitalPinToTimer() return values)

These are 8-bit values. They are fully enumerated and much more detail provided near the end of this document.

0x10-0x15 are TCA0. The three low bits are the mux option (DxCore only - other.
0x08-0x0D is TCA1. The three low bits are the mux option (DxCore only).
0x20-0x2F is a TCB
0x30-0x3F is a TCB alt pin (not yet implemented for PWM)
0x40-0x77 is a TCD pin (high nybble is the bit within the group, low nybble is the mux. On megaTinyCore, this is always 0x40, as there is no mux for TCD.)
0x80  is the DAC output pin
TCE and TCF details pending release of more information

Using to check that correct menu option is selected

If your sketch requires a certain timer NOT be used for millis, or that a certain timer nust be the timekeeping timer, you should check these #defines and be sure to either #error out, or at least emit a #warning.

#ifndef MILLIS_USE_TIMERB2
  #error "This sketch is written for use with TCB2 as the millis timing source"
#endif

Testing if millis enabled

The names of these two functions are #defined as themselves if the function can be used. On DxCore, millis is never defined if micros is not, since there are no timekeeping options which support millis but not micros. On megaTinyCore, if the RTC is used, millis, but not micros, will be available, so defined(micros) is false but defined(millis) is true.

Identifying part family within sketch

When writing code that may be compiled for a variety of target chips, it is often useful to detect which chip it is running on.Defines of the form __AVR_partname__ are provided by the toolchain. However, it is often MORE useful to not have to have a giant block of #ifdefs, and have the part numbers distilled down to something more that can be checked more intuitively, since what you care about is the family, number of pins, or occasionally size of the flash.

This core provides an additional define depending on the number of pins on the part and it's family (which identifies it's peripheral selection) and the core version:

  • For megaTinyCore:
    • __AVR_ATtinyxyz__ - where z, and optionally x and y are substituted with those parts of the part number, ex: __AVR_ATtinyxy4__ or __AVR_ATtinyx02__. The fully substituted form is provided by the compiler.
  • For DxCore:
    • The toolchain provides __AVR_AVRxx(D|E)(A|B|D)yy where x is flash size in kb, and y is pincount. These are what is used to deduce flash and pincount parameters below.

The below parameter defines should be used in preference to the above.

Part basic parameter determination

  • _AVR_FAMILY - String - "DA", "DB", "DD, "DU", "EA", "EB" or "T0", "T1", "T2" depending on what kind of part it is.
  • _AVR_PINCOUNT - The number of physical pins
  • _AVR_FLASH - Flash size, in KB
  • __AVR_DA__ (official, defined by avrlibc))
  • __AVR_DB__ (official, defined by avrlibc))
  • __AVR_DD__ (official, defined by avrlibc))
  • __AVR_DU__ (unreleased - official, defined by avrlibc)
  • __AVR_EA__ (official, defined by avrlibc)
  • __AVR_EB__ (unreleased - official, defined by avrlibc)
  • __AVR_TINY_0__ (unofficial - defined by the core)
  • __AVR_TINY_1__ (unofficial - defined by the core)
  • __AVR_TINY_2__ (unofficial - defined by the core)
  • MEGATINYCORE_MCU - The number after "ATtiny" in the part name.
  • MEGATINYCORE_SERIES - the part series, 0, 1, or 2.

Additional, deprecated, macros

This core provides an additional set of defines depending on the number of pins on the part and it's family. These are deprecated; the above definitions are preferred. The __AVR_xx__ and _AVR_PINCOUNT macros give the same information, are less ugly, and work equally well on megaTinyCore and DxCore.

  • DA_28_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DA_32_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DA_48_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DA_64_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DB_28_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DB_32_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DB_48_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DB_64_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DD_14_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DD_20_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DD_28_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DD_32_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DX_14_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DX_20_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DX_28_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DX_32_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DX_48_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • DX_64_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • EA_28_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • EA_32_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • EA_48_PINS (Deprecated - use _AVR_FAMILY and _AVR_PINCOUNT)
  • HAS_64_PINS (Deprecated - use _AVR_PINCOUNT)
  • HAS_32_PINS (Deprecated - use _AVR_PINCOUNT)
  • HAS_28_PINS (Deprecated - use _AVR_PINCOUNT)
  • HAS_24_PINS (Deprecated - use _AVR_PINCOUNT)
  • HAS_20_PINS (Deprecated - use _AVR_PINCOUNT)
  • HAS_14_PINS (Deprecated - use _AVR_PINCOUNT)
  • HAS_8_PINS (Deprecated - use _AVR_PINCOUNT)

Peripheral detection

Most of these need no explanation - they usually just give you the number of instances of a peripheral. Those that don't need an explanation don't have one.

  • _AVR_AC_COUNT
  • _AVR_ADC_COUNT - 1, or 2 on the tinyAVR 1+series.
  • _AVR_DAC_COUNT
  • _AVR_DAC_VERSION - 0, 1, or 2. 0 is 8-bit (tinyAVR), 1 is 10-bit (Dx-series), 2 is 10-bit with OUTRANGE (Ex-series), 1.5.5+/2.6.6+ only.
  • _AVR_OPAMP_COUNT
  • _AVR_EVSYS_COUNT - The number of total EVSYS channels. 1.5.5+/2.6.6+. 3 on 0-series, 8 on high pincount Dx, 6 everywhere else
  • _AVR_EVSYS_SYNC - The number of synchronous EVSYS channels. 1 or 2. Only defined on tinyAVR 0/1. 2.6.6+ only
  • _AVR_EVSYS_ASYNC - the number of async EVSYS channels. 2 or 4. Only defined on tinyAVR 0/1. 2.6.6+ only
  • _AVR_EVSYS_VERSION - 0, 1, 2, or 3 - 0 is the tinyAVR 0/1 shitshow, 1 is mega0, 2 is Dx-series 3 is the AVR Ex-series. Version 0 has two kinds of generator channels. Version 3 eliminates all differences between channels. Version 1 and 2 give different RTC channels to odd and even generator channels and limit access to pin events to to 2 or 4 channels per port. 1.5.5+/2.6.6+ only
  • _AVR_LUT_COUNT
  • _AVR_TCA_COUNT
  • _AVR_TCB_COUNT
  • _AVR_TCD_COUNT
  • _AVR_TWI_COUNT
  • _AVR_USART_COUNT

The EVSYS defines are very useful if you are using EVSYS without event, but even if you are using event, you want to, for example, select channels to minimize the functionality you're blocking off with your configuration, unless you're using an Ex-series, where all channels are created equal.

Identifying other attributes (added 1.5.8)

_AVR_FLASHMODE describes which method of writing to the NVM is used.

  • There are three basic modes, described here as 0, 1 and 2.
    • 0 is the traditional paged mode where you fill a page buffer then do an erase+write typically.
    • 1 is much like 0, except it has RWW support. Unfortunately, the current EA implementation is a disaster.
    • 2 is the AVR DA/DB/DD method of writing flash. Pages are much larger, but only used for erase.
    • The high bit is special - if set, it indicates that severe errata impacts the chip compromising the breadth of the applications to which it is applicable
      • 0x81 is mode 1 as it first shipped (on 64EA) - erase-write broken, no multipage erase over UPDI, etc.
      • 0x82 is the AVR128DA family where the byte write mode improperly applies protection. Only word writes should be used on those parts.

_AVR_CLOCKMODE is a 1-byte descriptor of the clocking options.

  • The pattern is 0bTTTAUESW
    • TTT indicates the number of tuning "notches", 0 = 32, 1 = 64, 2 = 128, 3 = 256 (not seen on modern AVR).
    • A, if set, indicates that en external watch crystal can be used to autotune.
    • U indicates auto-tuning from USB frames is supported (native USB parts only).
    • E indicates that external HF crystals can be used.
    • S indicates that the speed is selected at runtime, not by fuses.
    • W watch crystal for RTC supported. For example an AVR-DB would be 0x17, while a tiny 2-series would be 0x41
  • This is meant to be tested by libraries and similar.

Identifying core version

These define the version of the core:

  • For megaTinyCore:
    • MEGATINYCORE - megaTinyCore version number, as string. If it's "Unknown 2.5.0+" or similar, see below.
    • MEGATINYCORE_MAJOR - megaTinyCore major version
    • MEGATINYCORE_MINOR - megaTinyCore minor version
    • MEGATINYCORE_PATCH - megaTinyCore patch version
    • MEGATINYCORE_RELEASED - 1 if a released version, 0 if unreleased (ie, installed from github between releases).
    • MEGATINYCORE_NUM - megaTinyCore version, as unsigned long.
  • For DxCore:
    • DXCORE - DxCore version number, as string. If it's "Unknown 1.5.0+" or similar, see below.
    • DXCORE_MAJOR - DxCore major version
    • DXCORE_MINOR - DxCore minor version
    • DXCORE_PATCH - DxCore patch version
    • DXCORE_RELEASED - 1 if a released version, 0 if unreleased (ie, installed from github between releases).
    • DXCORE_NUM - DxCore version, as unsigned long.

DxCore

Serial.println(DXCORE);
Serial.print(DXCORE_MAJOR);
Serial.print(' '); // using ' ' instead of " " saves a tiny amount of flash!
Serial.print(DXCORE_MINOR);
Serial.print(' ');
Serial.print(DXCORE_PATCH);
Serial.print(' ');
Serial.println(DXCORE_RELEASED);
Serial.println(DXCORE_NUM,HEX);

Will produce output like:

1.3.7
1 3 7 1
01030401

or - if your non-Arduino IDE is not able to handle the escaping and you happened to be using a 1.3.8 github build:

Unknown 1.3.7+
1 3 8 0
01030800

megaTinyCore

Serial.println(MEGATINYCORE);
Serial.print(MEGATINYCORE_MAJOR);
Serial.print(' '); // using ' ' instead of " " saves a tiny amount of flash!
Serial.print(MEGATINYCORE_MINOR);
Serial.print(' ');
Serial.print(MEGATINYCORE_PATCH);
Serial.print(' ');
Serial.println(MEGATINYCORE_RELEASED);
Serial.println(MEGATINYCORE_NUM,HEX);

Will produce output like:

2.5.0
2 5 0 1
02050001

or - if your non-Arduino IDE is not able to handle the escaping and you happened to be using a 2.5.1 github build:

Unknown 2.5.0+
2 5 1 0
02050100

The various numeric representations of version are of interest to those writing libraries or code meant to be run by others. They are meant to ensure that there is always an easy way to test against specific versions (for when a release contains a regression), as well as an "X or better" type test. Warning: In some third party development environments, the DXCORE define is lost. Since that's how we tell people to detect if DxCore is in use (#if defined(DXCORE)) we now detect that and put define it as "Unknown 1.3.7+" (where the number changes whenever I happen to notice it while editing the file for a new version). DXCORE_NUM is by far the easiest way to detect if a certain fix or patch is present.

__AVR_ARCH__

This can be set to 102, 103, or 104 depending on flash size - the compiler sets this, and it works for ALL AVR parts. Anything with it set to 103 has fully mapped flash. Anything with 102 has 64k, and 104 indicates a part with 128k of flash - on megaTinyCore, it's always 103, because no parts have too much flash to map it all.

  • __AVR_ARCH__ == 103 - All parts where all of the flash is mapped in the data space. This means Dx and Ex-series parts with 32k or less of flash, tinyAVR 0/1/2-series, and megaAVR 0-series.
  • __AVR_ARCH__ == 104 - Parts with 128Kb of flash, mapped flash is split into 4 sections (AVR128DA, AVR128DB).
  • __AVR_ARCH__ == 102 - Parts with 64Kb of flash, mapped flash is split into 2 sections (AVR64DA, AVR64DB).

Core feature detection

  • DEVICE_PORTMUX_TCA = 1 1 = each wave output channel can be moved individually, like tinyAVRs. 2 = Whole thing moved at once There are a number of macros for determining what (if any) features the core supports (these are shared with megaTinyCore and ATTinyCore (2.0.0+)

  • CORE_HAS_FASTIO = 2 - If defined as 1 or higher. indicates that digitalWriteFast() and digitalReadFast() is available. If undefined or defined as 0, these are not available. If 2 or higher, as it is for all parts on megaTinyCore, pinModeFast() is also available. If defined as -1, there are no digital____Fast() functions, but with constant pins, these functions are optimized aggressively for speed and flash usage (though not all the way down to 1 instruction). If not defined at all, the core may not be aggressively optimized.

  • CORE_HAS_OPENDRAIN = 1 - If defined as 1, indicates that openDrain() and (assuming CORE_HAS_FASTIO is >= 1) openDrainFast() are available. If undefined or 0, these are not available.

  • CORE_HAS_PINCONFIG = 1 - If defined as Indicates that pinConfigure() is available. If not defined or defined as 0, it is not available.

  • CORE_HAS_TIMER_TAKEOVER = 1 - if defined as 1 or more, takeOverTCxn() functions are available to let user code take full control of TCA0, TCA1 and/or TCD0. If a part, for which support has been provided by one of my cores, in a release, without a takeOverTCxn() function for a timer it has, and later gets a takeOverTCxn function for that timer, there would need to be a CORE_HAS_TIMER_TAKEOVER = 2 for versions of the core that have that new function. I hope, however, to give all new timers a takeOver function if one is appropriate with the first release of support for parts having those timers.

  • CORE_HAS_TIMER_RESUME = 0- if defined as 1, the corresponding resumeTCAn() functions, which reinitialize them and return them to their normal core-integrated functions, are available. Different values in the future might be used to indicate that resume timers other than TCAn, or which might have an option to resume TCD0. Not available on megaTinyCore.

  • ADC_NATIVE_RESOLUTION = 10 or 12- This is the maximum resolution, in bits, of the ADC without using oversampling. 2-series get 12-bits, other tinies get 10, and all Dx/Ex get 12, though the Ex has a much nicer ADC.

  • ADC_NATIVE_RESOLUTION_LOW = 8 or 10 - All the ADCs, in addition to converting at ADC_NATIVE_RESOLUTION, have a lower resolution, slightly faster option. On the tinyAVRs this is 8-bit for all parts, while for Dx, it's 10 bits.

  • ADC_DIFFERENTIAL = 0, 1, or 2 - This is defined as 1 if the part has a basic differential ADC (no gain, and Vanalog_in constrained to between Gnd and VRef, as on the Dx-series), and 2 if it has a full-featured one. It does not indicate whether said differential capability is exposed by the core. This is 0 for 0/1-series and 2 for 2-series. Classic AVRs on ATTinyCore 2.0.0+ will define this as -1 if they have a differential ADC, as their ADC takes differential channels in a totally different way, and will not define it at all if they do not. See also CORE_HAS_ANALOG_DIFF

  • SUPPORT_LONG_TONES = 0 or 1 - On some modern AVR cores, an intermediate value in the tone duration calculation can overflow (which is timed by counting times the pin is flipped) leading to a maximum duration of 4.294 million millisecond. This is worst at high frequencies, and can manifest at durations as short as 65 seconds worst case. Working around this, however, costs some flash, so megaTinyCore only supports long tones on parts with more than 8k of flash. If SUPPORT_LONG_TONES is defined as 1, as long as (duration * frequency)/500 < 4.294 billion, the duration will not be truncated. If it is defined as 0, the bug was known to the core maintainer and they chose not to fully correct it (eg, to save flash) but took the obvious step to reduce the impact, it will be truncated if (duration * frequency) exceeds 4.294 billion. If SUPPORT_LONG_TONES is not defined at all, the bug may be present in its original form, in which case the duration will be truncated if (duration * frequency) exceeds 2.14 billion.

  • CORE_HAS_ANALOG_ENH = 1 - If defined as 1, analogReadEnh() (enhanced analogRead) is available. Otherwise, it is not; it is 1 for all parts on recent versions of megaTinyCore.

  • CORE_HAS_ANALOG_DIFF = 0 or 1 - If defined as 1, analogReadDiff() (differential enhanced analogRead) is available. Otherwise, it is not. 1 for 2-series (which has a differential ADC) and for the Ex/Dx-series, which have at least a mediocre , 0 for others (as they do not). It has same features as enhanced, except that it takes a differential measurement. If this is -128, (128 unsigned), it is a classic AVR, not a modern one with a differential ADC, and the core's analogRead implementation accepts the constants listed in the core documentation to make an analogRead of a differential pair.

  • CORE_HAS_ENH_SERIAL = 1 - if defined as 1 the core has the new implementation of Serial. See the serial reference

  • CORE_HAS_ERRORFUNS = 1 - Newly added define - it's become apparent that reliance on the core's badArg and badCall functions was impacting portability of library code; this can be tested for by libraries which use badArg() and badCall() so that if it's not present, the library can include a copy of the definitions. Mostly meant for internal use, though some other people might be using badArg() or badCall() in their libraries?

Hardware feature detection

  • ADC_MAX_OVERSAMPLED_RESOLUTION = 13 (tinyAVR0/1), 15 (Dx-series) or 17 (tinyAVR 2 or Ex-series) - If either CORE_HAS_ANALOG_ENH or CORE_HAS_ANALOG_DIFF is 1, this will be defined as the maximum resolution obtainable automatically via oversampling and decimation using those functions.
  • ADC_MAXIMUM_GAIN is 0 (DA, DD, DU, tinyAVR0/1), -1 (DB), or 16 (EA, EB, tinyAVR 2) - Some parts' ADC is blessed with a Programmable Gain Amplifier (PGA) amplifier, often used for differential readings (though modern AVRs can use it for single-ended readings, if they have it - not that it's terribly useful). The Dx-series are not among them, though the EA will have it. If this is defined as a positive number, it is the maximum gain available (16 on the EA). If this is defined as 0 that means there is no way to amplify an analog signal we might want to measure (if this is a problem you need external opamps). If this is defined as -1 (or 255 if stuffed into a uint8), there are one or more OPAMP peripherals available which could be directed towards the same purpose, though more deliberation and part specific work would be needed; they may be more flexible in some ways, but they are very rigid in others (for example, in that there is only one pin option for them. If it is defined as -128 (128 in a uint8) there is a gain stage on the differential ADC, but it is specified along with the pair of pins, not with separate configuration options. This also means that it's NOT a modern AVR! This is generally what is done on all classic AVRs with a differential ADC) so the available gain options depend on which pins are being measured, and there is a different procedure to use it, as detailed in the core documentation (ex, ATTinyCore 2.0.0 and later). If it is undefined, there is definitley no support exposed through the core, and it may not be a feature of the hardware at all.
  • _AVR_PORT_INLVL is 1 if the INLVL bit is present. If undefined or 0, that feature is not available.

Hardware pin determination

Timers (DxCore only)

For type A and type D timers, defines are provided for each pin included by a PORTMUX option for a type A or type D timer. This is NOT_A_PIN if the pin in question is not available on the device. This includes PA0/PA1 when they are used for the clock source or PD0 when that pin was replaced by the VDDIO2 pin. The type B timers are not yet covered.

  • PIN_TCA1_WO0_DEFAULT
  • PIN_TCA1_WO5_ALT1
  • PIN_TCD0_WOA_DEFAULT While Microchip has thus far been quite kind to us regarding keeping mappings consistent between part families, we can't count on that to always be the case. While these values can be deduced from datasheets, the point is to make it more straightforward to detect whether the pin is present and usable.

Other peripherals (DxCore only)

If present, any OPAMP peripherals have each of their pins defined as shown (where n is 0, 1, or 2) pins are defined as:

  • PIN_OPAMPn_INP - Positive Input to OPAMPn
  • PIN_OPAMPn_OUT - Output of OPAMPn
  • PIN_OPAMPn_INN - Negative Input of OPAMPn

Compatibility macros

Occasionally Microchip has not kept the names of registers or bitfields consistent between families of parts, even when the function was not changed. In some cases these have even been changed between versions of the ATpack! The places where we've papered over identical register functionality with different names are:

  • The GPIO/GPIOR/GPR registers - on the most recent parts, these are officially GPR.GPRn. We provide whichever of the following are not present: GPIOn, GPIORn, and GPR_GPRn. GPIORn is recommended for maximum compatibility. (GPIOR compatibility note: If writing code that may be ported to classic AVRs, GPIOR3 does not exist at all, and only GPIOR0 is guaranteed to be in the low I/O space - check the register map.
  • The TCA_SINGLE_EVACTA bitfield - formerly known as TCA_SINGLE_EVACT and it's group codes - the old names will work too, permitting code portability from tinyAVR to Dx.
  • The RTC_CLKSEL group codes have different spellings of some abbreviations on tinyAVR vs Dx/Ex. The old names will work too, permitting code portability from tinyAVR to Dx.
  • The CLKCTRL_SELHF_CRYSTAL_gc option on DB-series parts which was renamed to CLKCTRL_SELHF_XTAL_gc. Both names work.
  • All things CLKCTRL_FREQSEL related, which had the E dropped in some ATPACK versions. Both FRQSEL and the old FREQSEL spellings work. I dunno, there must be a shortage of capital E's at Microchip. Maybe it's the new EA and upcoming EB parts hogging all the E's?
  • All multi-bit bitfields - the _bp and _bm defines now have an underscore before the bit number. There are about a thousand impacted defines. At least they had the decency to bump the major version.

Errata

There are ah, a lot of errata on these parts, sad to say. See the errata summary for details on them. These constants are defined as either:

  • ERRATA_IRREL (-128) - This errata does not apply to a part family because the conditions that would make it manifest can never exist there (ex: ERRATA_PORT_PD0 is irrelevant to the DA, because there are no DA parts that don't have a PD0, only smaller DBs have that issue (the nominal PD0 pin is where they put the VDDIO2 pin instead, but the input buffer for PD0 was still present, and with no pin, would always be floating, and needs to have it's input buffer turned off to avoid wasting power).
  • -1 - This applies to all specimens of this part.
  • 0 - This does not apply to any parts in this family
  • Any other value - This is the REVID that corrected this problem.
  • use checkErrata() for best results, ex, if (checkErrata)

checkErrata(errata name) will return 1/true if the errata applies to this part, or 0/false if the errata does not apply or is irrelevant.

Note that checkErrata does not get compiled to a known constant unless either all specimens of the part in question or none of them have the issue. Right now, only the A4 AVR DB-series parts have bugs that were fixed by a subsequent die rev - but very few of them made it out of Microchip - at least to the general public. I ordered them basically as soon as they were for sale, and received them before they even posted the datasheet; they were all A5s. Therefore, to maximize performance, the problems corrected by A5 in the DB128 and DB64 are are reported by the core to not impact those devices, that is, on a DB-series, ERRATA_PLL_RUNSTBY will be defined as 0, not 0xA5. A similar thing happened on the DD-series, except that the DD-series didn't have any particularly bad errata. The 64k parts with 28+ pins and the smaller parts with less have two minor issues related to HV override of UPDI pin function (no define is provided for them; I couldn't think of a plausible use case where you would want different software options compiled in based on that - it impacts the upload process, not the code itself.

When future die revs fix some of these problems, checkErrata() will no longer compile to a known constant. Thus, do not use it in #if statements. Do not use it in functions that your application relies upon the compiler constant-folding and optimizing away.

Errata name 0 1 2 Quick description
ERRATA_CCL_PROTECTION All All All The whole CCL peripheral must be disabled in order to disable any one LUT to reconfigure it.
ERRATA_CCL_OUTEN Some Some No Some 0 and 1-series parts can't use the LINK input if the other LUT doesn't have OUTEN set.
ERRATA_CCL_DLATCH All Most No Except on Rev. C 3216 and 3217, the D-type Latch sequencer option is broken.
ERRATA_TCA_RESTART All All All TCA restart resets direction, like DA/DB datasheet says. Both datasheet and silicon were "wrong".
ERRATA_TCB_CCMP All All All In 8-bit PWM, CCMP treated as a 16-bit register not 2 8-bit ones, so both must be written, never just 1. Universal on modern AVRs.
ERRATA_TCD_ASYNC_COUNTPSC - All - Async event's are missed when TCD tries to use them if count prescaler is engaged.
ERRATA_TCD_AUTOUPDATE - Some - One of the three errata fixed in rev B of the 16k parts, out of the 34 errata present.
ERRATA_TCD_PORTMUX - All - TCD PORTMUX unusable. Only default portmux pins work.
ERRATA_TCD_HALTANDRESTART - All - Halt and Wait for SW restart fault mode does not work in dual slope more or if CMPASET = 0
ERRATA_TWI_PINS All All No The OUT register for SCL and SDA must be low, otherwise TWI will try to drive the pins high! (Wire.h makes sure this is done correctly)
ERRATA_TWI_FLUSH All All All TWI_FLUSH command leaves the bus in unknown state. That is exactly what it was supposed to fix.
ERRATA_USART_ISFIF All All All After an inconsistent sync field (ISFIF), you must turn off RXC and turn it back on
ERRATA_USART_ONEWIRE_PINS All All No The DIR register for TX must be set INPUT when ODME bit is set, or it can drive pin high. This is handled internally by the core, and you likely don't need to know.
ERRATA_USART_FRAMING All Most No Framing errors would corrupt future received bytes. The fix for this, present on Rev. C of the tiny 3216/3217, introduced the USART SFDEN wake bug below, and the ISFIF bug.
ERRATA_USART_WAKE No Few All You must clear SFDEN when waking on start of frame detection before you clear the RXCIF. This is handled internally by the core, and you likely don't need to know. Present on any parts that don't have the framing error bug.
ERRATA_USART_ISFIF No Few All Likely a consequence of the revised USART silicon's fix for "Full range duty cycle not supported when validating LIN sync" issue. That issue does not have a define, as it is related to one of the very few serial features we don't support
ERRATA_RTC_PITANDRTC All Most No One of the nastiest modern AVR errata. The PIT and RTC are inappropriately entangled, and neither will work correctly if the other is not enabled, and the ways in which they don't-work are perverse. Additionally, writing RTC.CTRLA resets the PIT interval timer on impacted parts. Both of these were fixed in 3216/3217 Rev C, but nowhere else.
ERRATA_TCB_SYNCUPD All All No SYNCUPD mode does not restart the TCB from a TCA restart command.
ERRATA_TCB_PULSELEN All All No The pulse length must be longer than the selected TCB clock period to be detected when in Time-Out Check and Input Capture Frequency and Pulse-Width Measurement.
ERRATA_TCB_ICFPWM Some Some No Unevenly scattered across the 0/1-series. Input Capture Frequency and Pulse-Width Measurement on impacted parts only works when using CLKSEL = 0.
ERRATA_AC_FALSETRIGGER Some Some No The analog comparators on some tiny 0/1-series parts can give false triggers, see the errata for details.
ERRATA_AC_INTFLAG Most? Most? No ACs didn't raise the INTFLAG unless the interrupt is enabled.
ERRATA_FUSE_OSCLOCK Most? Most? No OSCLOCK from fuses was applied before the factory cal was loaded, preventing factory cal from being loaded.
ERRATA_ADC_FLAGSRESH Some Some No Only an issue in 8-bit mode - the interrupt flags are only cleared automatically by reading the high byte of the result.
ERRATA_ADC_FREERUN Some Some No Certain control registers don't properly get applied if changed while freerunning mode in use, and an extra sample is taken after it is disabled.
ERRATA_ADC_ASDVSAMPLEN Some Some No SAMPLEN and INITDLY/ADSV do not play well together
ERRATA_ADC_WCMP Some Some No WCMP needs another INITDLY after waking the part from standby sleep mode.

Identifying Timers

Each timer has a number associated with it, as shown below. This may be used by preprocessor macros (#if et. al.) or if() statements to check what MILLIS_TIMER is, or to identify which timer (if any) is associated with a pin using the digitalPinToTimer(pin) macro. Defines are available on all parts that the core supports, whether or not the timer in question is present on the part (ie, it is safe to use them in tests/code without making sure that the part has that timer). There are two very closely related macros for determining pin timers:

  • digitalPinToTimer() tells you what timer (if any) the pin is associated with by default. This is a constant, when the argument is constant, the optimizer will optimize it away.
  • digitalPinToTimerNow() tells you what timer (if any) the pin is associated with currently. On megaTinyCore, this is either the result of digitalPinToTimer() unless that timer has been "taken over" by user code with takeOverTCA0() or takeOverTCD0(). On modern AVR cores like Dx-core which support use of analogWrite() even when the PORTMUX.TCxROUTEA register (where x is A or D) has been changed, this will return the timer currently associated with that pin. megaTinyCore does NOT support non-default timer pin mappings with analogWrite()- so if PORTMUX.TCAROUTEA (2-series) or PORTMUX.CTRLC (0/1-series) has been altered, this will not not reflect that.

These are the "timers" that can be associated with a pin:

NOT_ON_TIMER      0x00
TIMERA0           0x10
TIMERA1           0x08
TIMERB0           0x20
TIMERB1           0x21
TIMERB2           0x22
TIMERB3           0x23
TIMERB4           0x24
TIMERD0           0x40 (see notes)
DACOUT            0x80 (not a timer - but treated like it was for purposes of pin-info.)

TCD0 will be 0x40 when it's being used as the millis timer (not supported on DxCore), but when a pin is interrogated and reveals that it can be used by TCD0, the value it returns will be of the form 0b01cc_0mmm. Here cc is the channel (0 = WOA, 1 = WOB, 2 = WOC, 3 = WOD), and mmm is the value of PORTMUX.TCDROUTEA | PORTMUX_TCD0_gm. In the event that a future part is released with a second type D timer, there are two bits that could be used to signify that, leaving a second bit available if they added a WOE and WOF (to replicate the functionality of the venerable ATtiny861, which is still the only game in town if you want to control a three-phase BLDC motor), without having to change the numbering system otherwise. Regardless, unless madness breaks out at Microchip and they release a part with 3 or more type D timers and a third comparison channel, or a part with 5 or more type D timers, you can count on timer & 0x40 == 0x40 for a type D timer. We'll burn that bridge when we come to it.

This is a stark contrast to classic AVRs the timers were just numbered sequentially - here we can still uniquely specify any combination of timer and mux setting

Additionally, other contexts involving timers may return these values (obviously the RTC won't generate PWM for you).

TIMERRTC          0x90
TIMERRTC_XTAL     0x91
TIMERRTC_CLK      0x92

When digitalPinToTimer is used to get the timer associated with a pin, and that timer is a TCD the following constants may be returned.

TIMERD0_0WOA      0x40
TIMERD0_0WOB      0x50
TIMERD0_0WOC      0x60
TIMERD0_0WOD      0x70
TIMERD0_1WOA      0x41
TIMERD0_1WOB      0x51
TIMERD0_1WOC      0x61
TIMERD0_1WOD      0x71
TIMERD0_2WOA      0x42
TIMERD0_2WOB      0x52
TIMERD0_2WOC      0x62
TIMERD0_2WOD      0x72
TIMERD0_3WOA      0x43
TIMERD0_3WOB      0x53
TIMERD0_3WOC      0x63
TIMERD0_3WOD      0x73
TIMERD0_4WOA      0x44
TIMERD0_4WOB      0x54
TIMERD0_4WOC      0x64
TIMERD0_4WOD      0x74

Similarly, future parts with more TCD0 mux options may use:

TIMERD0_5WOA      0x45
TIMERD0_5WOB      0x55
TIMERD0_5WOC      0x65
TIMERD0_5WOD      0x75
TIMERD0_6WOA      0x46
TIMERD0_6WOB      0x56
TIMERD0_6WOC      0x66
TIMERD0_6WOD      0x76
TIMERD0_7WOA      0x47
TIMERD0_7WOB      0x57
TIMERD0_7WOC      0x67
TIMERD0_7WOD      0x77

Hence, if a pin returns a value with bit 6 set, and (PORTMUX.TCDROUTEA & PORTMUX_TCD0_gm) is equal to the value of the three low bits, then the pin is on TCD, and TCD is pointed at that set of pins, and you could proceed to get PWM out of it from TCD0.

A future version of the core may use these constants to specify that a pin has access to TCA0 or TCA1.

TIMERA0_MUX0      0x10
TIMERA0_MUX1      0x11
TIMERA0_MUX2      0x12
TIMERA0_MUX3      0x13
TIMERA0_MUX4      0x14
TIMERA0_MUX5      0x15
TIMERA0_MUX6      0x16
TIMERA0_MUX7      0x17
TIMERA1_MUX0      0x08
TIMERA1_MUX1      0x09
TIMERA1_MUX2      0x0A
TIMERA1_MUX3      0x0B
TIMERA1_MUX4      0x0C
TIMERA1_MUX5      0x0D

Similar to above, the mux value is stored in the 3 low bits. Bit 4 signifies TCA0, and bit 3 TCA1 (TCA2 would get both, in the event that we get a part with a third TCA).

And finally there are these. These are currently not used in any way by the core but may be used in a future version.

TIMERB0_ALT       0x30
TIMERB1_ALT       0x31
TIMERB2_ALT       0x32
TIMERB3_ALT       0x33
TIMERB4_ALT       0x34

Timer Redux

This means a specific order must be used when testing an 8-bit timer identifier (the format that the core uses internally refers to timers) to avoid incorrect results; assuming we take it as a given that we are using hardware that has been announced or released as of late 2022:

  1. If the high bit is set, it's the DAC or RTC, not a PWM timer.
  2. If 0x40 is set, TCD0. If any other bits are set, it's specifically a reference to a pin - bits 4 and 5 give the channel (A-D), and the low 3 bits are the mux.
  3. If 0x20 is set, it's a TCB, and the number is in the low nybble. If it's a pin reference, 0x10 indicates that it's on the alternate pin.
  4. If 0x10 is set, it's TCA0, if 0x08 is set it's TCA1, the mux option is in the low 7 bits, and if neither 0x10 or 0x08 are set, but it's not zero, you're trying to look up something that does not indicate any timer at all.

Timer identification examples

Two strange examples cam be found below

if (timer & TIMERD0 == TIMERD0) {
  // it's a TCD.
  // if it's a pin, the mux option that points to it is:
  uint8_t pinmux = timer & 0x07;
  // and we find the WO channel in bits 4 and 5:
  uint8_t wochan = (timer & 0x30) >> 4; // compiler ought to render this as mov andi swap
  TCD_t* tcd = &TCD0;
  // to get the bit corresponding to this pin in TCD0.FAULTCTRL
  // 0x40 << wochan;
  // looking up port and bit would take a lookup table of 1 byte per mux option, containing the port number and a number indicating pin layout.
  // 0b00000000, 0b00000001, 0b00001101, 0b00000110, 0b101100000
  // 0 = 4567
  // 1 = 0123
  uint pin = wochan;
  uint8_t tcdport = pgm_read_byte_near(&tcdtable[pinmux]
  if !(tcdport & 0x08) {
    pin += 4;
  }
  if (tcdport & 0x80) {
    if(wochan & 0x02) {
      _SWAP(tcdport);
      pin -= 2;
      // this is the pin within the port;
    }
    PORT_t *port = &PORTA + (tcdport & 0x07);
    // and now we have the port!
  } else if (!(timer & 0x80)) {
  if ((timer & (TIMERB0) == TIMERB0) {
    // TCB
    uint8_t timernbr = timer & 7;
    TCB_t* tcb=&TCB0;
    tcb += timernbr;
    // tcb now points to the TCB identified.
    bool altpins = timer & 0x10
    // This section left an an exercise for the reader.
    /* looking up pins would take a lookup table of 2 bytes per timer
     * hint: store each as 0bppp00bbb where ppp is the port number and bbb is the bit position
     * then it's easy to fish out the PORT and pin. But there's a trick to that format -
     * tcb_tbl & 0xE0 is the low byte of the address of the PORTx register! just 'or' with 0x0400
     */
  }
  uint8_t timernbr
  if(timer &(TIMERA0 | TIMERA1)) {
    uint8_t timernbr
    timernbr = (timer & TIMERA1 ? 1 : 0);
    if(timer & TIMERA0){
      timernbr + timernbr; // 0 -> 0,  1 -> 2
    }
    TCA_t *tca = &TCA0 + timernbr;
    if (timernbr == 0) {
      port = &PORTA + (PORTMUX.TCAROUTEA & PORTMUX_TCD0_gm);
    }
    // pin can't be worked back from the identity of the TCA.
    // and doing this for TCA1 is much messier because most of it's mappings are only 3 pin ones, and the mux to port mapping is completely random, but some aren't. Takes another lookup table.
  }
void setup() {
  Serial.begin(115200);
  if (checkErrata(ERRATA_TCA1_PORTMUX)) {
    Serial.println("ERROR: This chip cannot direct output from TCA1 to PG0:5 or PE4:7 due to silicon errata");
    while (1) {
      runInCirclesScreamAndShout();
    }
  } else {
    doSomethingWithTCA1PinMapping();
    partyDown();  // defined elsewhere.
  }
}