This shows you the differences between two versions of the page.
Next revision | Previous revision | ||
resources:tools-software:sigmastudio:usingsigmastudio:numericformats [18 Jun 2012 23:19] – created William Jahn | resources:tools-software:sigmastudio:usingsigmastudio:numericformats [27 Jan 2021 22:35] (current) – use wp> interwiki links Robin Getz | ||
---|---|---|---|
Line 1: | Line 1: | ||
- | ~~NOTOC~~ | ||
======Numeric Formats====== | ======Numeric Formats====== | ||
[[resources/ | [[resources/ | ||
\\ | \\ | ||
- | DSP systems use a standardized numeric format. Fixed-point numbers are formatted A.B, where A is the number of bits to the left of the decimal point (the integer part) and B is the number of bits to the right of the decimal point (the fractional part). | ||
- | |||
- | The AD1940/ | ||
+ | Audio data at the serial inputs and outputs is formatted as 24 bit, signed, integer values. SigmaDSP cores, however, use numeric formats with more than 24 bits for intermediate values, coefficients, | ||
+ | To represent non-integer values, all of the SigmaDSP architectures use fixed-point numeric formats (rather than the more common floating point) to maximize the precision needed for audio signal processing. Fixed-point numbers are simply formatted as **A.B**, where A is the number of bits to the left of the decimal point (the integer part) and B is the number of bits to the right of the decimal point (the fractional part). | ||
+ | The combined value A.B is in 2's complement format enabling both positive and negative values. For more information on two's complement representations, | ||
- | |||
- | Inputs to the SigmaDSP cores are 24 bits. In the core, the DSP adds 4 additional zeros for additional headroom. The result is a 28 bit number representation. | ||
- | |||
- | So, in the case of most SigmaDSPs, audio representation is represented with A = 5 and B = 23. In other words, the number format is 5.23. Control signals and index table values generally require integer representation, and are therefore represented with A = 28 and B = 0. In other words, | + | < |
- | |||
- | =====28.0 (Integer) Format===== | ||
- | + | ===== 32 bit Architectures ====== | |
+ | |||
+ | In 32 bit SigmaDSP architectures, | ||
+ | |||
+ | |||
+ | ==== 32.0 (Integer) Format ==== | ||
Signals that are in integer format follow standard binary rules for representation. | Signals that are in integer format follow standard binary rules for representation. | ||
- | 0 = 0, 1 = 1, 10 = 2, 11 = 3, 100 = 4, et cetera. | ||
- | Including zero padding, positive integers in the DSP are represented as follows: | ||
- | + | 0b = 0, 1b = 1, 10b = 2, 11b = 3, 100b = 4, et cetera. | |
- | '' | + | |
- | 0000 0000 0000 0000 0000 0000 0000 = 0\\ | + | |
- | 0000 0000 0000 0000 0000 0000 0001 = 1\\ | + | |
- | 0000 0000 0000 0000 0000 0000 0010 = 2\\ | + | |
- | 0000 0000 0000 0000 0000 0000 0011 = 3\\ | + | |
- | 0000 0000 0000 0000 0000 0000 0100 = 4\\ | + | |
- | ...\\ | + | |
- | 0000 1000 0000 0000 0000 0000 0000 = 8388608 (0 dB full scale represented in 28.0 format)\\ | + | |
- | ...\\ | + | |
- | 0111 1111 1111 1111 1111 1111 1111 = 134217727 (2^27 - 1)\\ | + | |
- | '' | + | |
- | + | ||
- | Negative | + | 32.0 formatted |
- | + | 0000 0000 0000 0000 0000 0000 0000 0000 = 0 | |
+ | 0000 0000 0000 0000 0000 0000 0000 0001 = 1 | ||
+ | 0000 0000 0000 0000 0000 0000 0000 0010 = 2 | ||
+ | 0000 0000 0000 0000 0000 0000 0000 0011 = 3 | ||
+ | 0000 0000 0000 0000 0000 0000 0000 0100 = 4 | ||
+ | ... | ||
+ | 0000 0000 1000 0000 0000 0000 0000 0000 = 8,388,608 (0 dBFS for 24 bit audio represented in 32.0 format) | ||
+ | ... | ||
+ | 0111 1111 1111 1111 1111 1111 1111 1111 = 2, | ||
- | The corresponding 28-bit two's complement integers are represented as follows: | ||
- | + | 32-bit negative integers are represented in two's complement as follows: | |
- | '' | + | |
- | 1000 0000 0000 0000 0000 0000 0000 = -134217728 | + | 1111 1000 0000 0000 0000 0000 0000 0000 = -2, |
- | ...\\ | + | ... |
- | 1111 1111 1111 1111 1111 1111 1100 = -4\\ | + | |
- | 1111 1111 1111 1111 1111 1111 1101 = -3\\ | + | |
- | 1111 1111 1111 1111 1111 1111 1110 = -2\\ | + | |
- | 1111 1111 1111 1111 1111 1111 1111 = -1\\ | + | |
- | '' | + | |
- | + | ||
In general, negative integers are not used in SigmaStudio or SigmaDSP algorithms. | In general, negative integers are not used in SigmaStudio or SigmaDSP algorithms. | ||
+ | |||
+ | As mentioned, the serial ports and DACs are use 24 bit, signed integer values. Avoid mapping a 32.0 formatted value to an audio output. | ||
- | When outputting to the serial ports or DACs, signals will saturate at 0 dBFS. This means any signal exceeding 8388608 in 28.0 format will be limited to full-scale on the outputs. | + | ====8.24 (Decimal) Format==== |
- | =====5.23 | + | Audio is commonly formatted as a decimal value in the range of (1 - 1 LSB) to -1. A full scale 24-bit input signal has a positive peak value in 24 bit, 2's complement representation of: |
- | + | 0111 1111 1111 1111 1111 1111 = 1 - 1 LSB = largest positive 24-bit value | |
- | Audio, unlike control signals, is not represented as an integer, but rather as a decimal | + | Adding |
- | + | 0000 0000 1111 1111 1111 1111 1111 1111 = 1 - 1 LSB | |
- | A full scale 24-bit input signal would have a positive peak of '' | + | Note that moving |
+ | |||
+ | 0000 0000 0000 0000 0000 0000 0000 0000 = 0.0 | ||
+ | 0000 0000 0100 0000 0000 0000 0000 0000 = 0.25 | ||
+ | 0000 0000 1000 0000 0000 0000 0000 0000 = 0.5 | ||
+ | 0000 0001 0000 0000 0000 0000 0000 0000 = 1.0 | ||
+ | 0000 0010 0000 0000 0000 0000 0000 0000 = 2.0 | ||
+ | 0000 0100 0000 0000 0000 0000 0000 0000 = 4.0 | ||
+ | 0111 1111 1111 1111 1111 1111 1111 1111 = (128.0 - 1 LSB) = max value | ||
+ | |||
+ | Similarly, for negative numbers: | ||
+ | |||
+ | 1000 0000 0000 0000 0000 0000 0000 0000 = -128.0 = min value | ||
+ | 1111 1100 0000 0000 0000 0000 0000 0000 = -4.0 | ||
+ | 1111 1110 0000 0000 0000 0000 0000 0000 = -2.0 | ||
+ | 1111 1111 0000 0000 0000 0000 0000 0000 = -1.0 | ||
+ | 1111 1111 1000 0000 0000 0000 0000 0000 = -0.5 | ||
+ | 1111 1111 1100 0000 0000 0000 0000 0000 = -0.25 | ||
+ | 1111 1111 1111 1111 1111 1111 1111 1111 = (1 LSB below 0.0) | ||
+ | |||
+ | When outputting to the serial ports or DACs, signals will saturate to 24 bits by right shifting by one and ignoring the upper 8 bits. This means any signal with a peak outside the range (1.0 - LSB) to -1 in 8.24 format will be clipped to full-scale on the outputs. | ||
+ | ===== 28 bit architectures ====== | ||
+ | |||
+ | In 28 bit SigmaDSP architectures, | ||
+ | |||
+ | |||
+ | |||
+ | ==== 28.0 (Integer) Format ==== | ||
+ | |||
+ | Signals that are in integer format follow standard binary rules for representation. | ||
+ | |||
+ | 0b = 0, 1b = 1, 10b = 2, 11b = 3, 100b = 4, et cetera. | ||
+ | |||
+ | 28.0 formatted numbers range from (2^27)-1 to -(2^27). Including zero padding, positive integers in the DSP are represented as follows: | ||
+ | |||
+ | 0000 0000 0000 0000 0000 0000 0000 = 0 | ||
+ | 0000 0000 0000 0000 0000 0000 0001 = 1 | ||
+ | 0000 0000 0000 0000 0000 0000 0010 = 2 | ||
+ | 0000 0000 0000 0000 0000 0000 0011 = 3 | ||
+ | 0000 0000 0000 0000 0000 0000 0100 = 4 | ||
+ | ... | ||
+ | | ||
+ | ... | ||
+ | 0111 1111 1111 1111 1111 1111 1111 = 134217727 (2^27 - 1) | ||
+ | |||
+ | |||
+ | 28-bit negative integers are represented in two's complement | ||
+ | |||
+ | | ||
+ | ... | ||
+ | 1111 1111 1111 1111 1111 1111 1100 = -4 | ||
+ | 1111 1111 1111 1111 1111 1111 1101 = -3 | ||
+ | 1111 1111 1111 1111 1111 1111 1110 = -2 | ||
+ | 1111 1111 1111 1111 1111 1111 1111 = -1 | ||
+ | |||
+ | In general, negative integers are not used in SigmaStudio or SigmaDSP algorithms. | ||
+ | |||
+ | As mentioned, the serial ports and DACs are 24 bit integer values. In 28 bit architectures, | ||
+ | |||
+ | ====5.23 (Decimal) Format==== | ||
+ | |||
+ | Audio, unlike control signals, is formatted intuitively as a decimal value in the range of 1 to -1. If a full-scale audio signal with a peak amplitude of 1 has -3 dB of gain applied, it should have an amplitude of approximately 0.707. If -6 dB of gain is applied, the signal has an amplitude of 0.5. | ||
+ | |||
+ | A full scale 24-bit input signal has a positive peak value in common 24 bit, 2's complement representation of: | ||
+ | |||
+ | 1000 0000 0000 0000 0000 0000 = 2^23 - 1 = 0dBFS | ||
+ | |||
+ | If we add 4 bits of headroom (MSB zero padding), the 28-bit representation becomes: | ||
+ | |||
+ | 0000 1000 0000 0000 0000 0000 0000 = 2^23 - 1 = 0dBFS (A=0, B=2^23 - 1) | ||
+ | | ||
+ | The negative peak of a full-scale signal becomes: | ||
+ | |||
+ | 1111 1000 0000 0000 0000 0000 0000 (A=-1, B=2^23 - 1) | ||
Moving the leading ' | Moving the leading ' | ||
- | + | | |
- | '' | + | 0000 0010 0000 0000 0000 0000 0000 = 0.25 |
- | 0000 0000 0000 0000 0000 0000 0000 = 0.0\\ | + | 0000 0100 0000 0000 0000 0000 0000 = 0.5 |
- | 0000 0010 0000 0000 0000 0000 0000 = 0.25\\ | + | 0000 1000 0000 0000 0000 0000 0000 = 1.0 (0 dBFS) |
- | 0000 0100 0000 0000 0000 0000 0000 = 0.5\\ | + | 0001 0000 0000 0000 0000 0000 0000 = 2.0 |
- | 0000 1000 0000 0000 0000 0000 0000 = 1.0 (0 dB full scale)\\ | + | 0010 0000 0000 0000 0000 0000 0000 = 4.0 |
- | 0001 0000 0000 0000 0000 0000 0000 = 2.0\\ | + | 0111 1111 1111 1111 1111 1111 1111 = (16.0 - 1 LSB) |
- | 0010 0000 0000 0000 0000 0000 0000 = 4.0\\ | + | |
- | 0111 1111 1111 1111 1111 1111 1111 = (16.0 - 1 LSB)\\ | + | |
- | '' | + | |
- | + | ||
- | For negative numbers, signed two's-complement is used. | + | Similarly, for negative numbers: |
+ | |||
+ | 1000 0000 0000 0000 0000 0000 0000 = | ||
+ | 1110 0000 0000 0000 0000 0000 0000 = -4.0 | ||
+ | 1111 0000 0000 0000 0000 0000 0000 = -1.0 | ||
+ | 1111 1000 0000 0000 0000 0000 0000 = -1.0 | ||
+ | 1111 1100 0000 0000 0000 0000 0000 = -0.5 | ||
+ | 1111 1110 0000 0000 0000 0000 0000 = -0.25 | ||
+ | 1111 1111 1111 1111 1111 1111 1111 = (1 LSB below 0.0) | ||
- | |||
- | '' | ||
- | 1000 0000 0000 0000 0000 0000 0000 = -16.0\\ | ||
- | 1110 0000 0000 0000 0000 0000 0000 = -4.0\\ | ||
- | 1111 0000 0000 0000 0000 0000 0000 = -1.0\\ | ||
- | 1111 1000 0000 0000 0000 0000 0000 = -1.0\\ | ||
- | 1111 1100 0000 0000 0000 0000 0000 = -0.5\\ | ||
- | 1111 1110 0000 0000 0000 0000 0000 = -0.25\\ | ||
- | 1111 1111 1111 1111 1111 1111 1111 = (1 LSB below 0.0)\\ | ||
- | '' | ||
- | |||
When outputting to the serial ports or DACs, signals will saturate at 0 dBFS. This means any signal with a peak exceeding 1.0 in 5.23 format will be limited to full-scale on the outputs. | When outputting to the serial ports or DACs, signals will saturate at 0 dBFS. This means any signal with a peak exceeding 1.0 in 5.23 format will be limited to full-scale on the outputs. | ||
+ | As mentioned, the serial ports and DACs are 24 bit integer values. The 24 least significant bits (the fractional portion = B) map to 0 dBFS. The four integer bits (A) are ignored. | ||
- | =====5.19 (Hardware | + | =====5.19 (Hardware |
- | |||
Some cells in SigmaStudio may use slightly different number formats. For example, since the hardware-based DSP readback registers in the ADAU1701 only have 24 bits, the lower 4 bits from the 5.23 signal are truncated and the number is represented in 5.19 format. | Some cells in SigmaStudio may use slightly different number formats. For example, since the hardware-based DSP readback registers in the ADAU1701 only have 24 bits, the lower 4 bits from the 5.23 signal are truncated and the number is represented in 5.19 format. | ||
- | |||
- | So, a full-scale signal that was represented in 5.23 format as\\ | + | So, a full-scale signal that was represented in 5.23 format as |
- | '' | + | |
+ | | ||
+ | | ||
would have its lower 4 bits truncated for 5.19 representation: | would have its lower 4 bits truncated for 5.19 representation: | ||
- | '' | ||
- | + | 0000 1000 0000 0000 0000 0000 | |
The result is that very small amplitude signals will be truncated and therefore cannot be read back from the DSP on the older generation of SigmaDSP cores. | The result is that very small amplitude signals will be truncated and therefore cannot be read back from the DSP on the older generation of SigmaDSP cores. | ||
- | |||
- | |||
Newer cores, such as the ADAU1761 and ADAU144x, have full 5.23 readback capabilities implemented in software. | Newer cores, such as the ADAU1761 and ADAU144x, have full 5.23 readback capabilities implemented in software. | ||
+ | |||
+ | |||
+ | ===== Decibel Conversion ===== | ||
Concept for dB conversion: | Concept for dB conversion: | ||
Line 142: | Line 200: | ||
In order to get a linear value into a hex/binary number that can be written to the SigmaDSP, and vice versa, it is important to understand number conversions. | In order to get a linear value into a hex/binary number that can be written to the SigmaDSP, and vice versa, it is important to understand number conversions. | ||
- | |||
- | |||
The actual level parameters that you need to write are only the last 28 bits, and this value is simply a linear gain value. Using -40 dB as an example, convert this value to a linear value using the standard dB equation: | The actual level parameters that you need to write are only the last 28 bits, and this value is simply a linear gain value. Using -40 dB as an example, convert this value to a linear value using the standard dB equation: | ||
- | + | | |
- | 20log10 (x/1) = -40dB x = .01 | + | |
Take this linear value and multiply by 223 in order to get the decimal representation of the value in hex, because it is a 5.23 value. | Take this linear value and multiply by 223 in order to get the decimal representation of the value in hex, because it is a 5.23 value. | ||
- | .01 * 223 = 83886.08. | + | |
Now take the integer part of this result and convert 83886 to hex, and you will get 0x00147AE. | Now take the integer part of this result and convert 83886 to hex, and you will get 0x00147AE. | ||
Line 159: | Line 213: | ||
The output of some blocks is a 5.19 number. The formula to determine the dB-output value from the readback value is the following: | The output of some blocks is a 5.19 number. The formula to determine the dB-output value from the readback value is the following: | ||
- | dB_value = 96.32959861 * (readback_value / 219 - 1) | + | |
+ | |||
+ | |||
+ | |||
+ | |||
+ | |||
+ | |||