Power Meter Destroys Modbus RTU Bus Communication?

C

Thread Starter

connected

Hi everyone

This is my first message at control.com, but I've been reading here for a while. Seems to be a great site.

I'm an electrician in my daily work, and have stumbled upon an interesting problem which I'm supposed to solve.

We have a SCADA system which includes supervising of electrical power flow for a residential building. The subsystem consists of:

1 x Moxa Mgate MB3180 ethernet/serial interface
1 x Schneider power meter PM750
27 x Carlo Gavazzi EM24 power measurements.

They are interconnected through daisy-chain 2-wire RS485 which are running Modbus RTU. The problem seem to be that the Schneider PM750 destroys the communication on the line.

Through a web interface communicating with the Moxa Mgate, I got the message "Illegal Package (CRC)" with the PM750 is connected. This is true for every component on the bus. When the PM750 not connected (identical wiring), there are no error message, and the communication works as intended.

I've checked the baud rate, parity and adresses of components, but haven't found the solution to the problem yet.

I've also experimented a bit with different physical configurations regarding terminating resistors and shield termination. Since the wiring is identical when the communication works and when it don't work, I'm suspecting the Schneider PM750 to be erronous. I've also monitored the bus activity on oscilloscope and the physical signals seem to be fine.

The PM750 is located as the last component on the wiring, and this setup have been working for some time. We have almost identical setups in different buildings which are working, but there the PM750 is located at the beginning of the bus wiring.

I haven't experimented with line biasing or isolating the PM750 through RS485 repeater.

I would appreciate any thoughts to the problem!
 
L

Lynn August Linse

It is most likely a grounding issue - that the 2 vendors approached the 'reference signal to ground' differently. Mixing vendors on a single RS-485 is always tricky.

How is the third signal ground-wire handled?
 
D

David Wilson

Some notes:
> They are interconnected through daisy-chain 2-wire RS485

The shield and grounds are part of the circuit. Check for consistency and common mode problems.

> I've also experimented a bit with different physical configurations regarding terminating resistors and
> shield termination. Since the wiring is identical when the communication works and when it don't work

Where are the terminating resistors and what value?

> The PM750 is located as the last component on the wiring

Do you move the terminating resistor when connecting and disconnecting the PM750 for testing?
 
> It is most likely a grounding issue - that the 2 vendors approached the 'reference signal to ground'
> differently. Mixing vendors on a single RS-485 is always tricky.

> How is the third signal ground-wire handled?

The cable shield is continuous in its entire length. I have tried 2 different connections of the ground/shield:

- only connected at the start of the bus (the Moxa Mgate end)
- connected at all components
 
> Where are the terminating resistors and what value?

Resistors are placed at the cable ends, 120 Ohm value.

>> The PM750 is located as the last component on the wiring

> Do you move the terminating resistor when connecting and disconnecting the PM750 for testing?

No, the resistor placement are the same. I connect/disconnect the PM750 by the termination plug on the component.
 
D

David Wilson

This link shows the shield connection to the PM750:

http://download.schneider-electric.com/files?p_Reference=63230-507-200&p_EnDocType=Instruction sheet&p_File_Id=27450377&p_File_Name=63230-507-200.pdf

I have seen devices with +/- marked backwards of other devices (sometimes from the same vendor). I just replaced a third party device with a Schneider and had to reverse the labeled polarity.

<b>Moderator's Note:</b> When copying and pasting long URLs, make sure you delete any spaces placed in the URL by the forum software.
 
I've been researching this problem today and tried both biasing of the bus, different common/shield connections and changing the topology of the rs485 components.

Neither of the experiments resulted in a successful establishment of communication to the company Pm750.

The most positive result gained today was the tolerance of Pm750 on the network, i.e. it generated no error messages. But i got no response whatsoever on master requests to the pm750. Therefore i suspect a defect communication module in pm750.

I guess I will switch the Pm750 with a new one.

By the way,, does it exist any utility to check the serial interface of the Pm75? I'm thinking of a pc interface and software where you could pass a arbitrary request to the pm750 and inspect the response (if any).
 
L

Lynn August Linse

If you expect lightning/grounding damage, what I have done a few times is power everything off & node by node, measure the resistances between A/B, A/Gnd, B/Gnd with each node connected & disconnected. A unit with hardware damage will tend to have too low of a resistance (or at least too low compared to the others). If that doesn't point out a bad-apple, then (again with power off) try to hold the A/B 3vdc apart via some 1k resisters& repeat measuring the effect on the differential voltage of removing each node. You might find a node which allows the A/B to be 2.6vdc apart when NOT connected, but drops it to less than 1v when connected - this is NOT good behavior for an idle RS-485 driver chip! Such chips should have at most a few mV impact when added/removed (when powered off)

Sadly, I've even had cases where a bad node talks FINE all by itself, but when added to a multi-drop, things add up & nothing works.
 
> By the way,, does it exist any utility to check the serial interface of the Pm75? I'm thinking of a pc interface and
> software where you could pass a arbitrary request to the pm750 and inspect the response (if any).

There is a free open source Modbus example here that should work on a Windows or Linux PC:
http://ctrlterm.com/custom.htm
 
I've seen similar problems a few times before -- where one device brings down the RS485 bus. One issue that may or may not be the problem that you are experiencing is with the common mode differential voltage between devices on the bus. Here's how I test for it...

With the PM750 disconnected from the RS-485 bus, and using a DC voltmeter -- measure the DC voltage from its 'A' Terminal to the 'A' wire of the bus. Do the same for the 'B' Terminal to the 'B' wire of the bus.

This will give you an idea of the common mode differential voltage between the PM750 and the rest of the devices on the bus.

If the PM750 has a common-mode offset of greater than 16VDC, then it will likely cause a bus failure. Most RS-485 transceiver chips are limited to a 16V common-mode differential voltage from the bus, and they will fail if exceeded.

If this is the problem, then you will need to place the problem device behind an RS-485 isolator, such as a B&B Electronics 485OPDRI.

(Another solution for devices in general is to make sure to power the problem device with an isolated DC power supply. However, in your case, since it is a power meter, it probably generates its own DC power from the connected AC power, right?)

Hope this helps!

-Jim Mimlitz, SCADAmetrics
 
Thank you for your kind and relevant responses.

I've been approaching the problem with biasing of communication lines and common mode differences in mind.

Here are som result from my last research:

- It seems to be a potential of about 3V between PM750 ground and Moxa Mgate ground levels, with PM750 potential as the lower one.

- When removing cable shield/ground wire connection on Moxa Mgate, and attaching on the PM750 end, the PM750 stopped destroying the communication (Illegal packet message gone). Even so, the PM750 will not talk with the Moxa Mgate, also with the PM750 as the only component on the bus.

- The communication between the Moxa Mgate and Carlo Gavazzi EM24 seems to be reliable, even without shield/ground attached in any place.

I will try to measure DC potential between A and B lines on the PM750 when detached, relative to the bus.

My boss tells me to fix this, so I've ordered a repeater to isolate the PM750. If that don't do it, I'll have to order a new PM750 I guess.

I do not understand why it wouldn't talk at all now, when it have done so earlier. I've double-checked communication parameters.
(It could of course be broken as I've mentioned earlier)
 
I finally found a solution to this problem.

Based on the response here, I've measured the voltage differential on A and B lines between the PM 750 and the network. DC levels seemed to be approx. 7V, and AC levels close to 60V (!)( RMS reading) After connecting RS485 repeater between PM750 and the rest of the network, the communication recovered and is now stable.

I still wonder why the communication have been working for approx a year, and then fails without changes in network hardware. It's hard to tell I guess, but my thoughts are changes to the electronics inside PM750 or in the electromagnetic surroundings.

I haven't checked, but the power source to Moxa Mgate and PM750 could be fed from different phases (400V TN network), and that might be a problem even if I don’t think it should be.

Anyway I want to say thank you again for your response, most appreciated :)
 
C
Differences are common, if there is nothing to tie their references together. If one end or both is well isolated it only takes a tiny leakage current to create large potentials. In fact, there is no reason to expect that the systems would be at the same potential. A change in capacitance, moisture level, inductive coupling, phase of the moon :^). Consider that you can raise a foam coffee cup to thousands of volts dragging if off the desk. If the isolation is good enough it could assume almost any voltage until something conducts, usually a protection diode. These are called phantom voltages and can be misleading with high impedance measuring equipment. If you check it with an old Simpson multimeter it would probably go away due to the 20k ohms/volt resistance. But it is really there as far as the common move voltage range is concerned.

Regards
cww
 
Top