Refer to Section 11 of this user guide to see how error counts may be queried for specific Modbus slave devices.
The reliability code displayed for each object on the Data List page (see section 12) will indicate Modbus errors on a point by point basis. The reliability code generally indicates the nature of the problem.
Refer to Appendix J, Hardware Details, for a detailed description of what the LEDs are indicating.
If the Modbus slave device being queried by the gateway returns an error code, or there is simply a problem in communicating with the slave, this will be indicated by the gateway object's reliability code. When reliability code is non-zero, the fault flag is also set. Therefore, the Status Flags indication will typically be "F,T,F,F" any time the reliability code is something other than zero (zero means no errors to report).
Error codes returned by a Modbus slave device are encoded into reliability codes presented by the respective object. Reliability codes pertaining to polling of Modbus slaves (read via the Reliability property) are as follows:
no response (64)
crc error (65)
exception, illegal function code (66)
exception, illegal data address (67)
exception, illegal data value (68)
exception, other code returned by device, code+65, (69..79)
configuration property fault (80)
exception, code not recognized (81)
Reliability codes will “latch” by default and require that you read the Reliability property in order to reset it to zero, assuming the problem has gone away. Once the non-zero reliability code has been read (by reading the Reliability property), it will reset to zero the next time the object is updated, provided the problem has been resolved.
Since many systems do not automatically read Reliability codes, but do automatically respond to the Fault Status Flag associated with the non-zero reliability code, an auto-reset option is available (on the Device page). When set, reliability codes will return to zero as soon as the problem has been resolved, regardless of whether the non-zero reliability code was ever acknowledged by reading it.
Many systems will report an object as "offline" because its fault status bit is set. It is not actually offline, and is in fact communicating just fine trying to tell you that there is a problem. But many front end systems don't recognize this and blindly claim "offline" as a result of the fault bit in the status property. If you are having this issue as the result of communication errors on the Modbus side, try setting the Auto-Reset Errors option here. This only applies if you have "Map Modbus Object" checked on one or more Object Map pages, and the gateway is configured to be Modbus master.
No-response errors are probably the toughest because it means no activity is being recognized. CRC errors are marginal progress because it says the devices are at least seeing some bits on the line, even if the bits don’t make sense yet. Exception errors are a good sign because it means you are successfully communication with the Modbus device. Receiving an exception error requires receiving a good packet with a good CRC check. This means communication is ok, but configuration is asking for something the Modbus device does not like.
No-response errors:
• Check to see that communication parameters are correct (baud rate, etc).
• Check to see that the slave address matches.
• Check to see that Pre-Delay is at least 50 mS.
• Check wiring and power.
• Check for reversed polarity on RS485 lines. If uncertain, just try swapping them.
• Check to see that slave device is enabled for Modbus communication (many devices default to disabled)
CRC errors:
• Check baud rate and character format.
• Check wiring – if everything else is correct, CRC errors mean noise on the line.
• Check for reversed polarity on RS485 lines. Reversed polarity often looks like just noise.
• Check to see that Pre-Delay is at least 50 mS.
Exception errors:
• Check configuration. You cannot receive an exception error report if you are not successfully communicating with the Modbus device. Wiring, etc, is not a problem. Configuration has something wrong (most often the register number requested is not available).
Modbus Register Types
The types of registers referenced in Modbus devices include the following:
• Coil (Discrete Output)
• Discrete Input
• Input Register
• Holding Register
Whether a particular device includes all of these register types is up to the manufacturer. It is very common to find all I/O mapped to holding registers only. Coils are 1-bit registers, are used to control discrete outputs, and may be read or written. Discrete Inputs are 1-bit registers used as inputs, and may only be read. Input registers are 16-bit registers used for input, and may only be read. Holding registers are the most universal 16-bit register, may be read or written, and may be used for a variety of things including inputs, outputs, configuration data, or any requirement for "holding" data.
Modbus Function Codes
Modbus protocol defines several function codes for accessing Modbus registers. There are four different data blocks defined by Modbus, and the addresses or register numbers in each of those overlap. Therefore, a complete definition of where to find a piece of data requires both the address (or register number) and function code (or register type).
The function codes most commonly recognized by Modbus devices are indicated in the table below. This is only a subset of the codes available - several of the codes have special applications that most often do not apply.
| Function Code | Register Type |
| 1 | Read Coil |
| 2 | Read Discrete Input |
| 3 | Read Holding Registers |
| 4 | Read Input Registers |
| 5 | Write Single Coil |
| 6 | Write Single Holding Register |
| 15 | Write Multiple Coils |
| 16 | Write Multiple Holding Registers |
Modbus Exception (error) Codes
When a Modbus slave recognizes a packet, but determines that there is an error in the request, it will return an exception code reply instead of a data reply. The exception reply consists of the slave address or unit number, a copy of the function code with the high bit set, and an exception code. If the function code was 3, for example, the function code in the exception reply will be 0x83. The exception codes will be one of the following:
|
1
|
Illegal Function | The function code received in the query is not recognized by the slave or is not allowed by the slave. |
|
2
|
Illegal Data Address | The data address (register number) received in the query is not an allowed address for the slave, i.e., the register does not exist. If multiple registers were requested, at least one was not permitted. |
|
3
|
Illegal Data Value | The value contained in the query's data field is not acceptable to the slave. |
|
4
|
Slave Device Failure | An unrecoverable error occurred. |
|
6
|
Slave Device Busy | The slave is engaged in processing a long-duration command. The master should try again later. |
|
10
(hex 0A) |
Gateway Path Unavailable | Gateway could not establish communication with target device. |
|
11
(hex 0B) |
Gateway Target Device Failed to Respond | Specialized use in conjunction with gateways, indicates no response was received from the target device. |
|
17
(hex 11) |
Gateway Target Device Failed to Respond | No response from slave, request timed out. |
Modicon convention notation for Modbus registers
Modbus was originally developed by Gould-Modicon, which is presently Schneider Electric. The notation originally used by Modicon is still often used today, even though considered obsolete by present Modbus standards. The advantage in using the Modicon notation is that two pieces of information are included in a single number: (a) The register type; (b) The register number. A register number offset defines the type.
The types of registers referenced in Modbus devices, and supported by Babel Buster gateways, include the following:
• Coil (Discrete Output)
• Discrete Input
• Input Register
• Holding Register
Valid address ranges as originally defined for Modbus were 0 to 9999 for each of the above register types. Valid ranges allowed in the current specification are 0 to 65,535. The address range applies to each type of register, and one needs to look at the function code in the Modbus message packet to determine what register type is being referenced. The Modicon convention uses the first digit of a register reference to identify the register type.
Register types and reference ranges recognized by Babel Buster gateways are as follows:
0x = Coil = 00001-09999
1x = Discrete Input = 10001-19999
3x = Input Register = 30001-39999
4x = Holding Register = 40001-49999
Translating references to addresses, reference 40001 selects the holding register at address 0000, most often referred to as holding register number 1. The reference 40001 will appear in documentation using Modicon notation, but Babel Buster gateways require specifying "holding register" and entering that register number as just "1".
On occasion, it was necessary to access more than 10,000 of a register type using Modicon notation. Based on the original convention, there is another defacto standard that looks very similar. Additional register types and reference ranges recognized by Babel Buster gateways are as follows:
0x = Coil = 000001-065535
1x = Discrete Input = 100001-165535
3x = Input Register = 300001-365535
4x = Holding Register = 400001-465535
If registers are 16-bits, how does one read Floating Point or 32-bit data?
Modbus protocol defines a holding register as 16 bits wide; however, there is a widely used defacto standard for reading and writing data wider than 16 bits. The most common are IEEE 754 floating point, and 32-bit integer. The convention may also be extended to double precision floating point and 64-bit integer data.
The wide data simply consists of two consecutive "registers" treated as a single wide register. Floating point in 32-bit IEEE 754 standard, and 32-bit integer data, are widely used. Although the convention of register pairs is widely recognized, agreement on whether the high order or low order register should come first is not standardized. For this reason, many devices, including all Control Solutions gateways, support register "swapping". This means you simply check the "swapped" option (aka "High reg first" in some devices) if the other device treats wide data in the opposite order relative to Control Solutions default order.
Control Solutions Modbus products all default to placing the high order register first, or in the lower numbered register. This is known as "big endian", and is consistent with Modbus protocol which is by definition big endian.
What does notation like 40001:7 mean?
This is a commonly used notation for referencing individual bits in a register. This particular example, 40001:7, references (Modicon) register 40001, bit 7. Bits are generally numbered starting at bit 0, which is the least significant or right most bit in the field of 16 bits found in a Modbus register.
How do I read individual bits in a register?
Documentation tends to be slightly different for every Modbus device. But if your device packs multiple bits into a single holding register, the documentation will note up to 16 different items found at the same register number or address. The bits may be identified with “Bn” or “Dn” or just “bit n”. Most of the time, the least significant bit will be called bit 0 and the most significant will be bit 15. It is possible you could find reference to bit 1 through bit 16, in which case just subtract one from the number to reference the table below.
You cannot read just one bit from a holding register. There is no way to do that - Modbus protocol simply does not provide that function. You must read all 16 bits, and then test the individual bit you are interested in for true or false (1 or 0). Babel Buster gateways provide an automatic way of doing that by including a "mask" in each register map or rule. Each time the register is read, the mask will be logically AND-ed with the data from the register, and the result will be right justified to yield a 1 or 0 based on whether the selected bit was 1 or 0. Babel Buster gateways provide optimization when successive read maps or rules are selecting different bits from the same register. The Modbus register will be read from the slave once, and the 16-bit data will be shared with successive maps or rules, with each map or rule selecting its bit of interest.
The bit mask shown in the expanded form of the Babel Buster RTU read map is a 4 digit hexadecimal (16 bit) value used to mask out one or more bits in a register. The selected bits will be right justified, so a single bit regardless of where positioned in the source register will be stored locally as 0 or 1. The hex bit mask values would be as follows:
B0/D0/bit 0 mask = 0001
B1/D1/bit 1 mask = 0002
B2/D2/bit 2 mask = 0004
B3/D3/bit 3 mask = 0008
B4/D4/bit 4 mask = 0010
B5/D5/bit 5 mask = 0020
B6/D6/bit 6 mask = 0040
B7/D7/bit 7 mask = 0080
B8/D8/bit 8 mask = 0100
B9/D9/bit 9 mask = 0200
B10/D10/bit 10 mask = 0400
B11/D11/bit 11 mask = 0800
B12/D12/bit 12 mask = 1000
B13/D13/bit 13 mask = 2000
B14/D14/bit 14 mask = 4000
B15/D15/bit 15 mask = 8000
Some Modbus devices also back two 8-bit values into a single 16-bit register. The two values will typically be documented as “high byte” and “low byte” or simply have “H” and “L” indicated. If you run into this scenario, the masking for bytes is as follows:
High byte mask = FF00
Low byte mask = 00FF
When the mask value in a Babel Buster gateway is more than just one bit, the mask is still logicalle AND-ed with the data from the Modbus slave, and the entire resulting value is right justified to produce an integer value of less than the original bit width of the original register.
There have been a few instances of documenting packed bits in a 32-bit register. Although Modbus protocol is strictly 16-bit registers, some implementations force you to read pairs of registers. If your device documents 32 packed bits, then you would insert 0000 in front of each mask above, and the remainder of the list would be as follows:
B16/D16/bit 16 mask = 00010000
B17/D17/bit 17 mask = 00020000
B18/D18/bit 18 mask = 00040000
B19/D19/bit 19 mask = 00080000
B20/D20/bit 20 mask = 00100000
B21/D21/bit 21 mask = 00200000
B22/D22/bit 22 mask = 00400000
B23/D23/bit 23 mask = 00800000
B24/D24/bit 24 mask = 01000000
B25/D25/bit 25 mask = 02000000
B26/D26/bit 26 mask = 04000000
B27/D27/bit 27 mask = 08000000
B28/D28/bit 28 mask = 10000000
B29/D29/bit 29 mask = 20000000
B30/D30/bit 30 mask = 40000000
B31/D31/bit 31 mask = 80000000
Deciphering Modbus Documentation
Documentation for Modbus is not well standardized. Actually there is a standard, but not well followed when it comes to documentation. You will have to do one or more of the following to decipher which register a manufacturer is really referring to:
a) Look for the register description, such as holding register, coil, etc. If the documentation says #1, and tells you they are holding registers, then you have holding register #1. You also have user friendly documentation.
b) Look at the numbers themselves. If you see the first register on the list having a number 40001, that really tells you register #1, and it is a holding register. This form of notation is often referred to as the old Modicon convention.
c) Look for a definition of function codes to be used. If you see a register #1, along with notation telling you to use function codes 3 and 16, that also tells you it is holding register #1.
IMPORTANT: Register 1 is address 0. Read on…
d) Do the numbers in your documentation refer to the register number or address? Register #1 is address zero. If it is not clear whether your documentation refers to register or address, and you are not getting the expected result, try plus or minus one for register number. All Control Solutions products refer to register numbers in configuration software or web pages. However, some manufacturers document their devices showing address, not register numbers. When you have addresses, you must add one when entering that register into configuration software from Control Solutions.
Can I put 2 gateways on the same Modbus network?
You can not have more than one Master on a Modbus RTU (RS-485) network. Therefore, if the gateway is to be configured as the Master, you can only have 1 gateway. You cannot use multiple gateways to read more points from the same Modbus slave device.Multiple gateways configured as slaves can reside on the same Modbus RS-485 network.
If you are using RS-232 devices, you can have only two devices total, regardless of how they are configured. RS-232 is not multi-drop.
How many devices can I have on a Modbus RTU network?
Logically you can address over 250 devices; however, the RS-485 transceivers are not capable of physically driving that many devices. Modbus protocol states that the limit is 32 devices, and most RS-485 transceivers will agree with this. Only if all devices on the network have low load transceivers can you have more than 32 devices.