This site is now in read-only archive mode. Please move all discussion, and create a new account at the new Victron Community site.
Hi,
I have a question about my methodology for running a long RS485 cable, beyond 30 meters, especially in regards to the USB-RS485 adapter and the wiring for the 120 Ohm termination resistors either end.
I have read through the available documentation from Victron and FTDI and I have arrived at what I *believe* is the correct method I should be using.
Can someone please confirm that my understanding of this is correct based on my attached picture.
After reading a lot into this I have arrived at the following wiring method being the correct one.
It is important to avoid accidentally creating a ground loop through the shield and black GND wire.
The only way you can reliably earth both ends (if required by local regulations) is to break the ground loop by means of USB or RS485 galvanic isolator.
USB is notoriously difficult to provide galvanic isolation for because of strict capacitance requirements imposed by the signalling standards of USB, therefore a market for USB isolators is limited and expensive.
Don't be tempted by cheap Amazon USB isolators they barely work let alone meet required safety standards. A good industrial USB isolator is going to be £100+
It is much easier to find better and cheaper RS485 galvanic isolators. RS485 is less strict
in regards to extra capacitance introduced by galvanic isolation methods.
I am no expert and I'm just an electronics hobbyist at best, but I believe this is the correct method, I encourage anyone who is a subject matter expert to correct me here.
I have just installed my ET112 and have terminated an RJ45 on the end of my cat6 cable so it plugs in nicely to the ET112. I have just used the standard 568B Wiring.
ON the other end i have used an rj45 coupler and terminated the USB cable:
Black to Pin4
Yellow to Pin 5
Orange to Pin 6
I haven't used any resistors, I also haven't cut the USB cable down yet as I haven't finalised the location but will shorten this to around 20cm.
Run is over 20m and no issues.
A quick update:
I received my Victron USB-RS485 adapter cable and I have tested the brown and green wires with a multimeter, and I can confirm that the brown and green wires are the two ends of a 120 Ohm resistor.
+------- (Terminator 120 R, Pin 1) Wire 4 (Brown) | +-+ | | | | 120 Ohm termination resistor | | +-+ | +------- (Terminator 120 R, Pin 2) Wire 6 (Green)
i need to bring this up again.
So... how to correctly terminate the USB/RS485 Adapter?
A. just connect brown with green cable? (120Ohm resistor is already build internal)
B. put a 120Ohm resistor inbetween brown and green cable?
OK, i can answer it myself now. If i measure between brown and green cable, i see 120Ohm.
i think thats the same JC said one post above.
So simply connect those wires and the end-termination on GX side is done.
Other and last question:
Black cable GND:
This one should go to the GND Pin on the other RS485 device (gridmeter), thats clear.
But:
do you connect it to earth also?
- if so, only at one end of the RS485 bus?
- or at both ends?
or is it better not to earth the black cable, but rather, only connect the shield of the Bus-cable to earth.
Again same question here:
only at one end. Or at both?
(my RS485 bus cable will be ~ 120m long, so i want to make it as correct as possible)
do make my question super clear:
What is correct?
Version A:
not ground black cable:
Version B:
do ground black cable:
if, so: only on one end? or on both ends?
Version C:
not ground black cable, but ground cable shield:
if, so: only on one end? or on both ends?
thank you!
thank you!!!
After some research in the meanwhile, i totally agree with you in all points.
Also thanks for correcting the 120 Ohm "activation" method at the GX / USB side.
i originally thought that works like a jumper and is already internally connected to the data wires. But its not -> your version makes more sense.
Maybe another feasible solution if you have wifi/Internet at cerbos. I use rs485 to TCP server/client adapters from waveshare. (from Waveshare)
Converted rs845 to tcp/ethernet using one adapter then transfered over lan to another adapter back from TCP to rs485. I imagine latency is increased but doesn't seem too much.
why would you do that?
There are Ethernet Versions of the grid meters available.
But thats not what i wanted. If your local network fails, there is no communication anymore. But the RS485 version would still work. So, for me this seams to be the more robust version. There is nothing inbetween -> direct connection.
Also i have read about several issues with the Ethernet version
Although I know rs485 can use cat6 I then can only use that wire for rs485.
Much less versatile than a lan connection (ie I have poe camera next to inverter also), however you are right a running an additional rs485 cable would result less points of failure. (but for me that means a lot of digging to the shed)I also had a 95m run between my meter and inverter which wasn't a concern with tcp/Lan but would've needed more planning for rs485.
@jc2010 makes a good point. I hadn't considered as in my area and for max output size it wouldn't be an issue. My connections are all wire and if the Lan failed and all power was temporarily exported it would be below my DNO export limits However that may not be the case for everyone.
@Pedaaa The RS485 TCP solution proposed by @matt1309 is another good method to get around this distance issue, it is just as reliable as the RS485 method, perhaps not as simple and with a slightly larger failure domain but it's still a good alternative.
One thing to note about using wifi or wireless bridging via zigbee is that this is not an 'advised' method of communication for EESS as per the requirements of g100 in regards to export control. Everything ideally must be hardwired. So you can use the TCP server for RS485 but it has to connect via an ethernet cable ideally.
But that is a moot point in regards to Victron's systems because it shuts down the EESS in under 5 seconds upon losing contact with the control unit. Therefore using a wireless method with a victron device is permissible here.
"The ELS may be formed of discrete units, as shown in Appendix B, or integrated into a single packaged scheme. Where discrete units are used they should preferably be interconnected using metallic or fibre optic cables.
Alternatively the units may be interconnected using secure radio links but where this is the case these links shall be licensed (by OFCOM) and have a planned availability of 99.9% or higher.
Irrespective of the media used for interconnecting between the discrete units, if the communication path fails the generation output shall be reduced to a nominal value stipulated by the DNO within a set response time
(see section 5.5) to prevent the Agreed Export Capacity from being exceeded."
From:
ENA_EREC_G100_Issue_1_Amendment_2_(2018)
I would actually say that thinking on this, the RS485 TCP server is perhaps the superior solution here, because with the encapsulation of RS485 inside TCP and transport over native Ethernet you can make use of a cheap 10-100 Mbps Multimode/Singlemode fiber optic transceivers/media converters and cables, providing you with the best possible galvanic isolation available, along with the best signal integrity.
