LM as a Modbus slave : number of slaves

[SigmaTec]

Hi,

dear Admin, did you confirm than it is possible to create only one slave at the save time on the same LM ?
No way to increase it ?

Have a good day.

[admin]

In TCP mode you can have multiple slave scripts but they must use different ports (e.g. 502, 503 etc). In RTU mode only one slave is supported at the moment.

[SigmaTec]

Thank's !

[SigmaTec]

In TCP mode you can have multiple slave scripts but they must use different ports (e.g. 502, 503 etc). In RTU mode only one slave is supported at the moment.

Hi Admin,

Ok, I understood the principle. But how to decompose the cyclic script which sets the port (eg. 502) and the slave id (eg. 2) into several slave scripts with different ports?

How will the mbetregisters() function find its way?

Code:

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475

-- modbus init cyclic script if not mb then require('luamodbus') -->>>>>>>>>Modbus slave setting<<<<<<<<<<<<<<<<<<<<<<<<   ------------------------------------------------------ -- Pour Modbus RTU ------------------------------------------------------ --mb = luamodbus.rtu() --mb:open('/dev/RS485', 19200, 'E', 8, 1, 'H') --mb:connect() ------------------------------------------------------ -- Pour Modbus TCP ------------------------------------------------------ mb = luamodbus.tcp() mb:open('0.0.0.0', 502) ------------------------------------------------------  -->>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<    -- slave id mb:setslave(2)   -- init slave storage for coils, discrete inputs, holding registers and input registers mb:setmapping(1, 1, 3300, 1)   --***************************************************** -- >>>>>>>>>>>Coil write callback<<<<<<<<<<<<<<<<<<<<<< mb:setwritecoilcb(function(coil, value) if coil == 0 then grp.write('0/0/12', value, dt.bool) else alert('coil: %d = %s', coil, tostring(value)) end end) --***************************************************** -->>>>>>>>>>>Register write callback<<<<<<<<<<<<<<<<<<< mb:setwriteregistercb(function(register, value) if register == 0 then -- send value limited to 0..100 grp.write('0/0/13', math.min(100, value), dt.scale) else alert('register: %d = %d', register, value) end end) --***************************************************** end -- server part init if not server then log("Modbus slave handler (cyclic) : IN...server init") require('rpc') -- incoming data handler local handler = function(request) local fn, res fn = tostring(request.fn) if not mb[ fn ] then return { nil, 'unknown function ' .. fn } end if type(request.params) == 'table' then table.insert(request.params, 1, mb) res = { mb[ fn ](unpack(request.params)) } else res = { mb[ fn ](mb) } end return res end server = rpc.server('127.0.0.1', 28002, 'mbproxy', handler, 0.01) log("Modbus slave handler (cyclic) : OUT...handleslave = ") end res, err = mb:handleslave() res, err = server:step()

Thank's for your time....

[admin]

If you only need read-only values on the modbus side you can simply periodically update the register values in the resident script itself. Modify register mapping in updateregisters() as needed. Each script must use a unique port (502, 503 etc).

Code:

1234567891011121314151617181920212223242526272829303132333435

if not mb then   require('luamodbus')   mb = luamodbus.tcp()   mb:open('0.0.0.0', 502)   mb:setslave(2)   function setint64(grpaddr, regaddr)     local value = grp.getvalue(grpaddr) or 0     local raw = knxdatatype.encode(value, dt.int64).dataraw     local r1 = raw:byte(2) * 0x100 + raw:byte(1)     local r2 = raw:byte(4) * 0x100 + raw:byte(3)     local r3 = raw:byte(6) * 0x100 + raw:byte(5)     local r4 = raw:byte(8) * 0x100 + raw:byte(7)     mb:setregisters(regaddr, r4, r3, r2, r1)   end   function updateregisters()     setint64('32/1/1', 3204)   end   mb:setmapping(0, 0, 3300, 0)   updateregisters()   prev = os.time() end mb:handleslave() curr = os.time() if curr ~= prev then   updateregisters()   prev = curr end

[SigmaTec]

If you only need read-only values on the modbus side you can simply periodically update the register values in the resident script itself. Modify register mapping in updateregisters() as needed. Each script must use a unique port (502, 503 etc).

Code:

1234567891011121314151617181920212223242526272829303132333435

if not mb then   require('luamodbus')   mb = luamodbus.tcp()   mb:open('0.0.0.0', 502)   mb:setslave(2)   function setint64(grpaddr, regaddr)     local value = grp.getvalue(grpaddr) or 0     local raw = knxdatatype.encode(value, dt.int64).dataraw     local r1 = raw:byte(2) * 0x100 + raw:byte(1)     local r2 = raw:byte(4) * 0x100 + raw:byte(3)     local r3 = raw:byte(6) * 0x100 + raw:byte(5)     local r4 = raw:byte(8) * 0x100 + raw:byte(7)     mb:setregisters(regaddr, r4, r3, r2, r1)   end   function updateregisters()     setint64('32/1/1', 3204)   end   mb:setmapping(0, 0, 3300, 0)   updateregisters()   prev = os.time() end mb:handleslave() curr = os.time() if curr ~= prev then   updateregisters()   prev = curr end

Sorry Admin, using your script (without modify it, except '32/1/1') no communication with Modbus master is possible.
Note : I suspended cyclic "Modbus handler" script and preserved "mbproxy" script in common functions.

Perhaps interesting for you 'timeout ?' > founded by monotoring in log : 

res, err = mb:handleslave()
log("mb:handleslave :",res,err)
* arg: 1
  * string: mb:handleslave :
* arg: 2
  * nil
* arg: 3
  * string: timeout

[admin]

First, check the Error log for any errors.
You can use this test script for reading from local TCP slave with port 502 (check Logs tab for the read result or any errors):

Code:

12345678910111213141516171819

require('luamodbus') mb = luamodbus.tcp() mb:open('127.0.0.1', 502) res, err = mb:connect() if res then   val, err = mb:readregistervalue(3204, 'int64', 'b')      if val then     log('read ok', val)   else     log('read failed', err)   end else   log('connect failed', err) end mb:close()

[SigmaTec]

First, check the Error log for any errors.
You can use this test script for reading from local TCP slave with port 502 (check Logs tab for the read result or any errors):

Code:

12345678910111213141516171819

require('luamodbus') mb = luamodbus.tcp() mb:open('127.0.0.1', 502) res, err = mb:connect() if res then   val, err = mb:readregistervalue(3204, 'int64', 'b')     if val then     log('read ok', val)   else     log('read failed', err)   end else   log('connect failed', err) end mb:close()

Result : 
* arg: 1
  * string: connect failed
* arg: 2
  * string: Connection refused

But to be sure we are on the same subject :
my deal is to use LM as Modbus slave in order to write in it some registers which will be read by a master.

[admin]

"Connection refused" means that either your slave script is not running or it's not using port 502. The test script is a master that checks if a local slave is running or not.

[SigmaTec]

"Connection refused" means that either your slave script is not running or it's not using port 502. The test script is a master that checks if a local slave is running or not.

I use your previous script, that I run. Then I run your test script.
Did I make a mistake ?

Code:

123456789101112131415161718192021

if not mb then   require('luamodbus')   mb = luamodbus.tcp()   mb:open('0.0.0.0', 502)   mb:setslave(2)   function setint64(grpaddr, regaddr)     local value = grp.getvalue(grpaddr) or 0     local raw = knxdatatype.encode(value, dt.int64).dataraw     local r1 = raw:byte(2) * 0x100 + raw:byte(1)     local r2 = raw:byte(4) * 0x100 + raw:byte(3)     local r3 = raw:byte(6) * 0x100 + raw:byte(5)     local r4 = raw:byte(8) * 0x100 + raw:byte(7)     mb:setregisters(regaddr, r4, r3, r2, r1)   end   function updateregisters()     setint64('32/1/1', 3204)   end

[admin]

You're missing a part of the slave script. It should have at least 35 lines, you've posted only 21.

[SigmaTec]

You're missing a part of the slave script. It should have at least 35 lines, you've posted only 21.

yes, it was only the first lines...
Here is full !

Code:

12345678910111213141516171819202122232425262728293031323334353637383940

if not mb then   require('luamodbus')   mb = luamodbus.tcp()     res, err = mb:open('0.0.0.0', 502)   log("mb:open :",res,err)   res, err = mb:setslave(2)   log("mb:setslave :",res,err)     log("Modbus slave IN : 502/2")   function setint64(grpaddr, regaddr)     local value = grp.getvalue(grpaddr) or 0     local raw = knxdatatype.encode(value, dt.int64).dataraw     local r1 = raw:byte(2) * 0x100 + raw:byte(1)     local r2 = raw:byte(4) * 0x100 + raw:byte(3)     local r3 = raw:byte(6) * 0x100 + raw:byte(5)     local r4 = raw:byte(8) * 0x100 + raw:byte(7)     res, err = mb:setregisters(regaddr, r4, r3, r2, r1)     log("mb:setregisters :",res,err)   end   function updateregisters()     setint64('0/0/14', 3204)   end   mb:setmapping(0, 0, 3300, 0)   updateregisters()     prev = os.time() end mb:handleslave() curr = os.time() if curr ~= prev then   updateregisters()   prev = curr end log("Modbus slave OUT : 502/2")

[admin]

Works for me.

[SigmaTec]

Works for me.

Congratulations ;-)

For me :
mbproxy = {} is in common fuctions
no cyclic script as in my first post
your script for one objet/port

And nothing...

[admin]

What do you mean by "no cyclic script"? Each TCP slave runs in a separate resident script with 0 sleep time.

[SigmaTec]

What do you mean by "no cyclic script"? Each TCP slave runs in a separate resident script with 0 sleep time.

Yes, Admin, that's what I understood from analyzing the scripts. 

I discover your answer after everything works in resident script ! 
Sorry for the response time: I had no Internet for 3 hours (fiber optic installation in progress)


Question : Is there a limit in number of resident scripts with sleep time 0?


Thank you once again for your precious help and your patience...

[admin]

There's no exact limit on how many scripts can run no matter what the sleep time is set to. All scripts consume some amount of RAM but it's quite hard to hit this limit in normal use cases.

Modbus script is in sleeping state most of time anyway when it's waiting for a client to connect. 0 sleep time should avoided for busy loops which consume all CPU resources.

[SigmaTec]

Perfect Admin !
Have a good evening.