Modbus books?

[12345678910]

Hi, anyone could please share some book where I can read how to commission and config modbus communications?

Thanks in advance.

[carl_e]

No such book that I know of. Probably because the Modbus masters vary so much in their implementation.

For the most part, slaves are slaves, the manufacturer defines what data is where (in which register) and in what format. Their documentation is either good or bad at telling you this. An exception is the HART wireless gateways, where one assigns specific tags to specific registers - a custom configured slave register map. I assume DCS is the same deal - custom assigned registers for any Modbus slave functions.

Modbus over ethernet is easier than Modbus over RS-485 because Ethernet is electrically isolated and can't be hooked up backwards. The EIA RS-485 standard does not define which is (+) and which is (-) for 2 wire 485, and some mfgs do it one way, others the other. Hooking it up backwards doesn't damage the 485 drivers, but it won't work. Termination, distance, baud rate, and grounding/common mode all affect the signal which can make it a hassle, but it's a 485 hassle, Modbus is just the communications protocol, not the transport layer.

Modbus does not define the data format, so data can be different number formats: integer, signed integer, one of four 32 bit IEEE floating point formats, 64 bit floating point (typical of flow meters) 32 bit 'long integer', bit packed coil data, plain old coil data.

Modbus addressing has the 'one offset' issue. Addresses start at 00000, but registers start at 00001. Sometimes it isn't clear whether the slave's data starts at zero or one, and whether the Master's query instruction starts at zero or one. The problem manifests itself as the 'wrong data', because it is, the data has been read from an adjacent/contiguous register.

The leading 1, 2, 3 or 4 used in Modbus addressing refers to the legacy 'memory area' but continues to be used because a 40013 register address tells me it's a holding register that uses the 03 function code. But the leading 4 is NOT part of the address used in the communication's data packet, it is for human beings to more easily define the data type. The leading numerals are like hexadecimal characters, hex characters are not transmitted in communications packets, ones and zeros are transmitted in comm packets, but humans use hex to avoid long strings of zeros and ones.

My technique is to use Modscan 32 (a generic Modbus master that runs under Windows) to query the slave to determine what it takes to establish communications with the slave, what number format is in use, and whether the slave addressing starts at zero or one. It's quick and easy to change register addressing and number formats to establish what the slave really is. Then I move to the master and configure it to look for the same data in the same location.

It's a good idea to never look for data that is a zero, because too many unused registers are initialized to zero. I always try to force the slave to an analog value to an oddball value, like 87.3, and then make sure I can read that oddball value.

Modbus is more of a chore with Siemens because Siemens invented Profibus and only reluctantly supports Modbus. Same with Allen Bradley who has only recently started to adopt Modbus natively because they had their own protocols (partner Prosoft supplied Modbus backplane cards). Understanding what hardware and software licenses are needed with Siemens is a project in itself.

[12345678910]

Thank you. In your post I've found some things that suspected and now are confirmed to me.