Intel HEX

The Intel HEX format is a data format for storing and transmitting binary data. Today it is mainly used to save programming data for microcontrollers or microprocessors , EPROMs and similar components. However, it can also be used to store charging modules. The HEX format is the oldest data format of its kind and has been in use since the 1970s. Later expansions specifically support the segmented addressing of the Intel 80x86 processors .

An Intel HEX file is in ASCII format. The bytes of the coded binary data are each represented as a hexadecimal number made up of two ASCII characters (0… 9 and A… F). HEX files can be opened and modified with a text editor. The HEX file is roughly twice as large as the binary data it contains, since one byte is represented by two bytes in hexadecimal notation. The data records are provided with a checksum so that transmission errors can be recognized.

history

The Intel-Hex Format (originally Intellec-Hex ) was designed by Intel in 1973 for the Intellec Development Systems ( MDS ) to load and start programs from punched tape. It should also simplify the transfer of data to ROM production. At the same time, it was used to program (E) PROM by means of a punched strip or punched card-controlled EPROM programming device. With the introduction of floppy disk drives with the MCS Series II under ISIS II (1975), files were also created in this format. Since then, HEX has been used as the file extension .

format

The format described here corresponds to the Hexadecimal Object File Format Specification from Intel.

Structure of a data record

The coding is (7 bit) ASCII. Each data record is introduced by a colon (":"), consists of an even number of characters and is terminated by an end of line. The structure of the line end is not defined and depends on the media. Intel tools for streaming media always generate a CR / LF (0D0A _HEX ).

Two characters each represent a data byte. The notation is hexadecimal, big-endian with the characters 0..9 and A..F, i. This means that the more significant nibble comes first. All entries are also made in the Big-Endian address fields . Lowercase letters (a..f) are not mentioned in the definition but are supported by most implementations.

Intel designation	content	use
RECORD MARK	Start of sentence	":" (Colon, ASCII coding 3A _HEX )
RECLEN	Data length	Length of the user data as two hexadecimal digits
LOAD OFFSET	Loading address	16-bit address (big endian)
RECTYP	Record type	Record type (00..05)
INFO or DATA	Data	User data (RECLEN x 2 characters)
CHKSUM	Checksum	Check sum over the data record (without the beginning of a record)

Record types

Overview

There are six record types :

Type	designation	use
00	Data record	Payload
01	End of file record	End of file (and start address for 8-bit data)
02	Extended segment address record	Extended segment address for subsequent user data
03	Start segment address record	Start segment address (CS: IP register)
04	Extended Linear Address Record	Extended linear address, more significant 16 bits of the address for subsequent user data
05	Start linear address record	Linear start address (EIP register)

The data records can appear in any order; an end record (type 01 ) ends processing.

Data record (type 00 )

The data record contains the 16-bit load address and the user data.

	Start code	Number of bytes	address	Type	Data field	Checksum
length	1 character	2 digits	4 digits	2 digits	2 n digits	2 digits
content	:	n	address	00	Data	Checksum

n : number of bytes in the data field
Address : 16-bit address for storing the data record
Data : data field, n bytes

End of File Record (type 01 )

The data record marks the end of the file. In the original (8-bit) definition, the start address of the program (PC) is specified in the address field for loadable formats. In the 16/32 bit formats this must be 0000 .

	Start code	Number of bytes	address	Type	Checksum
length	1 character	2 digits	4 digits	2 digits	2 digits
content	:	00	0000	01	FF

Extended segment address record (type 02 )

The data field of the Extended Segment Address Record (extended segment address) contains bits 4-19 of the address segment of the following data records (counting starting with 0) in cases in which the size of a 16-bit address space (i.e. 64 kbytes) is insufficient. The address contained in the data field is shifted by 4 bits to the left (corresponding to a multiplication by = 16) and added to the 16-bit addresses contained in the following data records (type 00 ). The Extended Segment Address Record remains in effect until it is changed by another Extended Segment Address Record . The address field of the data record of type 02 is always 0000, the length is 02 . ${\ displaystyle 2 ^ {4}}$

	Start code	Number of bytes	address	Type	Data field	Checksum
length	1 character	2 digits	4 digits	2 digits	4 digits	2 digits
content	:	02	0000	02	segment	Checksum

Start segment address record (type 03 )

The data record specifies the start address for load modules. For x86 processors this is the CS: IP content. The data record can appear in any position. The start address is calculated as segment * 16 + offset . The address field is always 0000 , the length is 04 .

	Start code	Number of bytes	address	Type	Data field		Checksum
length	1 character	2 digits	4 digits	2 digits	4 digits	4 digits	2 digits
content	:	04	0000	03	segment	Offset	Checksum

Extended Linear Address Record (type 04 )

The data field of the Extended Linear Address Record (extended linear address, also 32-bit address data set or HEX386 record) is used to support a 32-bit address space with a 4 GB limit and contains 16–31 bits, the more significant 16 bits (ULBA , Upper Linear Base Address, counting starting with 0) of a 32-bit address (LBA, Linear Base Address). The address data record applies to all subsequent type 00 data records until it is replaced by another extended address data record. The absolute memory address of a type 00 data record is obtained by placing the address data from the extended address data record in front of the address field of this data record. If a Type 00 data record within a 32-bit address space is not preceded by a Type 04 address data record, the upper 16 address bits are set to 0000 by default .

The address field of the extended address data record itself is always set as 0000 (with a length of 02 ) :

	Start code	Number of bytes	address	Type	Data field	Checksum
length	1 character	2 digits	4 digits	2 digits	4 digits	2 digits
content	:	02	0000	04	ULBA, address (high word)	Checksum

Start linear address record (type 05 )

The data record specifies the start address for load modules. For x86 processors, this is the content of the EIP register. The address field is always 0000 , the length is 04 .

	Start code	Number of bytes	address	Type	Data field	Checksum
length	1 character	2 digits	4 digits	2 digits	8 digits	2 digits
content	:	04	0000	05	EIP	Checksum

Calculation of the checksum

The checksum is calculated from the entire data set, excluding the start code and the checksum itself. The data record is summed byte by byte, the lower byte is taken from the sum and the two's complement is formed from this.

The two's complement is formed by inverting the bits of the low byte and then adding 1. This can be done e.g. This can be achieved, for example, by using the exclusive-or link with FF _HEX and adding 01 _HEX . So 00 _HEX remains unchanged, 01 _HEX becomes FF _HEX , etc.

The two's complement expresses a negative number in the binary system. Since the checksum represents the negative sum of the remaining bytes, checking a data record for errors is very easy. You simply add up the individual bytes of a data record including the checksum and receive 00 _HEX as the low-order byte if the data record is correct.

variants

Intel

In the course of processor development from Intel 4004 until today, different variants have been defined:

variant	commitment	Permitted sentence types
I08HEX	4/8-bit CPU (4004..8085)	00 (Data), 01 (End of File)
I16HEX	16-bit CPU (8086/186/286)	00 (Data), 01 (End of File), 02 (Extended Segment Address), 03 (Start Segment Address)
I32HEX	32-bit CPU (from 80386)	00 (Data), 01 (End of File), 02 (Extended Segment Address), 03 (Start Segment Address), 04 (Extended Linear Address), 05 (Start Linear Address)

other producers

The HEX format was widely used as a quasi-standard. The byte order in the data field was partially changed, i. H. the order does not match the address location. Manufacturers (e.g. Texas Instruments) have also changed the addressing. There the address does not correspond to a byte, but to the width of a register of the processor.

example

:020000021000EC
:10010000214601360121470136007EFE09D2190140
:100110002146017EB7C20001FF5F16002148011988
:10012000194E79234623965778239EDA3F01B2CAA7
:100130003F0156702B5E712B722B732146013421C7
:00000001FF

Start code

Byte count

address

Type

Data field

Checksum

The checksum for the first sample data is calculated as follows: . ${\ displaystyle {10+}}$ ${\ displaystyle {01 + 00 +}}$ ${\ displaystyle {00+}}$ ${\ displaystyle {21 + 46 + 01 + 36 + 01 + 21 + 47 + 01 + 36 + 00 + 7E + FE + 09 + D2 + 19 + 01}}$ ${\ displaystyle \ {=} \ 3C0 \ rightarrow C0 \ rightarrow \ neg C0 + 1 \ {=} \ 40}$

Related file formats

The Motorola S format (also S-Record , SREC or S19 for short ) is very similar . There are also other formats for this area of application, such as the simple binary code or the Jedec format.

Sources, web links

Intel HEX format (English)
SB-Projects: fileformats: intel hex - very clear presentation (English)
hexformat - Python library for HEX formats IntelHex and SRecord
Intel HEX to bin C program with source code

Individual evidence

↑ Hexadecimal Object File Format Specification , Revision A of 6 January 1988
↑ ^a ^b General: Intel Hex File Format. ARM Germany GmbH, accessed on September 6, 2017 (English).

[1] Hexadecimal Object File Format Specification , Revision A of 6 January 1988

[data_type-2] General: Intel Hex File Format. ARM Germany GmbH, accessed on September 6, 2017 (English).