Modbus RS485 Master w/multiple slave

[gongue2005]

Hello everyone, I need an example of how to make a master and several (8) slaves by modbus RTU RS485, please, my knowledge does not work. the slaves are soft starters and only allow 100 milliseg fixed  I'm using visilogic v9 ... and probe with unilogic but the minimum time is 100 millisec, with one slave works perfectly but with two no longer.
I am desperate, anticipated thanks for your help.to not create error in the network.  I am desperate, anticipated thanks for your help.

[Ausman]

8 minutes ago, gongue2005 said:

visilogic v9 ... and probe with unilogic

Visilogic and Unilogic are 2 different systems.  What PLC are you using?

cheers,

Aus

[Joe Tauser]

You've presented us with a dark cave and very little to go on to help.  We need illumination.

1.  Post a link of the modbus documentation for the soft starters, and anything else that may be helpful.

2. Post your code that you've done so far.

Joe T.

[gongue2005]

On 6/15/2018 at 8:33 PM, Ausman said:

Visilogic and Unilogic are 2 different systems.  What PLC are you using?

cheers,

Aus

I look for the solution in either of them. with unistream it is not capable of polearning less than 100 millisec. the slaves are PSTX ABB sofstarters with native modbus rtu port that only has two values 9600 and 19200 bps. You have to force 5 coils by FC # 1 and read 11 FC # 15 coils. I have put the time out in 7 (70miliseg) and D # 1 retries in the modbus configurator and with a sequential drum active 3 MB every 70 millisec and works well with up to 3 slaves. no more. I must repeat that the slaves have a communication failure function that is activated if there is no data exchange in 100 millisec (fixed value) already this week it occurred to me to use a V1040 with 3 RS485 ports and each one connects to 3 slaves achieving a possible solution. Unistream was limited to not being able to have more serial ports (I do not know) and not being able to power less than 100 millisec. (or if?)
something interesting that you do, is that if you look within 70 millisec of the requests, the same MB makes two requests, one on the positive pulse and one on the negative (trying to optimize the time of intervention with the slave).
I annex my development for your criticism.
I am not happy with this solution, I am very lacking in knowledge.

3 Vision 6 esclavos pstx.vlp

[Isakovic]

I don't know why you use Drums, maybe just try rapid fire like in the picture, it could go faster. It works for me, but I always use 9600 though.

Good luck.

[Ausman]

I  haven't looked at your vlp, just your description, and as I said in your Unistream Hardware post I think you have a timing overlap problem occurring.  If you take out one slave, all the others are then activating their communication failure function.  You need to do maths on total times involved in the serial transmission process to work out optimum times, and then for good measure allow at least a few scans after the system says it has finished to fully clear buffers.  ie Your timeout is 70ms and you have 1 retry so if you take out 1 slave you have effectively reached 140ms which is over your slaves' failure trigger.   Perhaps don't do any retries at all, and monitor each connection in other ways to alert you to a failure.

Aus