How to split Eastron Modbus signal for two Deyes?

[shadowmaker]

It seems I can connect either one of my Deye 12Ks to my Eastron SDM630 Modbus V2 meter to get CT readings, but not at the same time for some reason. I couldn't use original CTs that came with the inverters because i have 75m to my main panel. Here's my setup (left one connected in the picture):




The blue CAT5e carries Eastron signal (RS485B-,RS485A+,GND) to an active RS232/RS485 splitter. Actually the splitter has only RS485B- and RS485A+ terminals on it, so I'm using RS232 port's ground terminal to carry ground signal over the splitter also. It seems ground signal isn't all that important for RS485 to work, but I have read that it is better to include it than not (it doesn't seem to change anything, connected or not). I can change CT signal on the fly (RJ45 connectors to Deye's meter port) and it works just fine as long as I don't connect both at the same time. If I do that the CT readings start to fluctuate wildly. I think I have proper connections/configurations as I can get both working separately, but there still is something fishy going on here.

 

[zanydroid]

I don’t think you can have multiple masters with ModBus RS485. If you look at the protocol level details it assumes there is only one. Does the active splitter speak ModBus? Or does it just blindly forward the serial bits around.

Following this thread to see what people like to use for active proxying to isolate this. I wouldn’t be surprised if there is a RasPi implementation of one.

 

[shadowmaker]

I don’t think you can have multiple masters with ModBus RS485. If you look at the protocol level details it assumes there is only one. Does the active splitter speak ModBus? Or does it just blindly forward the serial bits around.

Following this thread to see what people like to use for active proxying to isolate this. I wouldn’t be surprised if there is a RasPi implementation of one.

I thought the Eastron meter is the master and these two Deyes are the slave ones. I think the active splitter is just forwarding, nothing more.

 

[zanydroid]

I don’t know the master/slave terminology of ModBus very crisply, but I believe the inverters typically poll the meter, and having two sources of polling messages would likely confuse or cause interference.

If the meter actually periodically broadcasts the reading on RS485 without any polling then a basic forwarding splitter would likely work.

 

[houseofancients]

I don’t think you can have multiple masters with ModBus RS485. If you look at the protocol level details it assumes there is only one. Does the active splitter speak ModBus? Or does it just blindly forward the serial bits around.

Following this thread to see what people like to use for active proxying to isolate this. I wouldn’t be surprised if there is a RasPi implementation of one.

i thought so the same, as i told shadowmaker per pm..

did test it today after getting shadowmakers mesg yesterday..
as soon as i use the splitter, thing become unstable and measurements are all over the place

@shadowmaker i think you really need to come up with a way to do proper paralleling , otherwise one issue after another will keep comming up

 

[houseofancients]

I don’t know the master/slave terminology of ModBus very crisply, but I believe the inverters typically poll the meter, and having two sources of polling messages would likely confuse or cause interference.

If the meter actually periodically broadcasts the reading on RS485 without any polling then a basic forwarding splitter would likely work.

no the meter is the one broadcasting, without waiting confirmation.
what end up happening is one packet is "caught" by one inverter, the next 2 by the other..
makes thing unstable

 

[zanydroid]

Is it possible to shift everything to ModBus TCP with appropriate hardware proxies so at least stuff works slightly more like regular networking?

EDIT: and ModBus TCP naturally supports multi-master unlike ModBus RTU

 

[zanydroid]

This thread may be helpful

https://control.com/forums/threads/modbus-rtu-to-2-master-devices.46291/

And as I said a few messages above, make sure your understanding of the protocol and physical layer is super crisp. The descriptions at least in this thread are a little too handwavy and vague. Could just be in translating what you understand correctly -> English.

 

[houseofancients]

Is it possible to shift everything to ModBus TCP with appropriate hardware proxies so at least stuff works slightly more like regular networking?

when you parallel these, one can set the modbus id.
however , when in a parallel setup, the master passes the readings on to the slave even without a direct connection to the eastron
i think both would solve OP'S issue, however he stated that paralleling no option is for his situation.

 

[zanydroid]

Yes, I would expect with paralleling there would be a single master interrogating the meter, and then just sharing within the inverter stack via the internal synchronization bus.

Depending on how the meter reading is used in OP's situation, having the inverters independently take action on the feedback may or may not work. Hard to say without more detail on the end goal.

 

[shadowmaker]

With FLA and LFP battery chemistries paralleling is not an option, at least not right now. I thought Deyes were "read only" devices (=slave), but Eastron manual says that the meter can be remotely controlled via Modbus, which means it has to be a slave device.

I think easiest and cheapest way to solve my CT problem is to get another Eastron meter and have two independent sets. Kind of stupid setup, but this way I should get my system working automatically without constant babysitting.

Thank you zanydroid and houseofancients.

 

[houseofancients]

With FLA and LFP battery chemistries paralleling is not an option, at least not right now. I thought Deyes were "read only" devices (=slave), but Eastron manual says that the meter can be remotely controlled via Modbus, which means it has to be a slave device.

I think easiest and cheapest way to solve my CT problem is to get another Eastron meter and have two independent sets. Kind of stupid setup, but this way I should get my system working automatically without constant babysitting.

Thank you zanydroid and houseofancients.

a second meter shoud indeed solve your problem too..

i saw the same in my eastron manual, and have been using a packet inspector, yet havent seen any packets going towards the meter.
doesnt mean it cant, but i havent seen them, therefor my assumption that the meter always pushes it's measurements to modbus id 0 ( which is the id a master inverter has)

 

[zanydroid]

Can you force both inverters onto the same ID, and program the meter to send to that ID? As long as the inverters don’t reply (ie those are write registers) it might be fine. I’m not sure if ModBus has an acknowledgement for writes, if there is that would cause confusion.

If you have enough space to put two CT clamps on, it’s probably fine to do both meters. If you don’t then one of those multi master or proxy solutions could help.

 

[houseofancients]

Can you force both inverters onto the same ID, and program the meter to send to that ID? As long as the inverters don’t reply (ie those are write registers) it might be fine. I’m not sure if ModBus has an acknowledgement for writes, if there is that would cause confusion.

If you have enough space to put two CT clamps on, it’s probably fine to do both meters. If you don’t then one of those multi master or proxy solutions could help.

as of right now both inverters have the same id.
and then pickup the broadcast up at random, which causes OP'S issue

i would have to look in the packets, but there must be a sequence number, or lack there of ..

in any case, OP seems to have an idea to a solution

 

[zanydroid]

as of right now both inverters have the same id.
and then pickup the broadcast up at random, which causes OP'S issue

i would have to look in the packets, but there must be a sequence number, or lack there of ..

in any case, OP seems to have an idea to a solution

Thanks. For future reference what are you using to sniff the bus?

 

[houseofancients]

FREE Serial Protocol Analyzer: Windows Com Port Monitor & Packet Data Sniffer

Free Serial Port Monitor for Windows: Freeware COM Port Packet Sniffer and Software RS232 Protocol Analyzer

freeserialanalyzer.com

and wireshark should it be tcp

 

[zanydroid]

Ah ok you just use a USB/COM to RS485 adapter and put it on the bus?

 

[houseofancients]

Ah ok you just use a USB/COM to RS485 adapter and put it on the bus?

yes, for lack of a rs485 switch/hub

 

[zanydroid]

Ok. I took a quick look at a ModBus spec again, and if you look at page 5 and one of the Write commands like Force Coil, you can see that the protocol expects acknowledgments. If there are two slaves on the same address they’ll probably corrupt each other’s acknowledgments a lot of the time.

https://modbus.org/docs/PI_MBUS_300.pdf