Sie sind auf Seite 1von 8

Trouble shooting Profibus - A practical example

Troubleshooting Profibus PA
A practical example

James Powell, P.Eng.

On a recent field assignment to troubleshoot a network, we were reminded


of how robust Profibus is and how, even though troubleshooting is a science,
it is not black and white. One of the reasons that we found this site visit so
interesting was that things were not as they first appeared. In this article we
will go over the process we followed and the things we learnt.

The Site and were capable of transmitting


long distances. However, the envi-
The site was a specialty chemical ronment that they were being used
company producing chemicals used in was not very friendly to wireless
in the petro-chemical industries. technology. There were a lot of ob-
All the instruments were located in stacles and electromagnetic noise to
hazardous environments and the in- overcome.
trinsically safe aspect of Profibus PA
was used. The site consisted of a Sie- Each link module is configured for
mens APACS DCS with two Profibus 20 to 25 instruments with about six
DP networks that uses Satelline-3Asd to eight instruments per Profibus
wireless modems to connect to a PA segment using Profibus DP/PA Ex
total of 15 Profibus DP/PA Link mod- couplers.
ules, located in cabinets all around
the plant. The two Profibus DP networks are
both set up to run at 9600 baud,
The wireless modems used a li- and with special Profibus DP bus pa-
censed frequency for transmission, rameters due to the radio modems.

Industrial Communication

Answers for industry.

1
Trends in Instrumentation

The retry limit was set to 5, a value the


Server room
customer had received from the wireless (2nd floor)
modem manufacturer. This is high, espe-
cially for 9600 baud. Simatic PDM

The layout of the two networks is shown


in the Figure 1. Control room
(1st floor)

Reported problems APACS + DCS

Our customer reported a number of is- Profibus DP Profibus DP


network 2 network 1
sues:
- communicating to some nodes gave
problems
- Profibus errors on the DCS
- PDM sometimes looses connections

However, the site was functional and PA PA PA PA


segments segments segments segments
was getting all the data in within a
couple of seconds, which met the site
requirements.

Troubleshooting Tool Figure 1: The control system includes two Profibus DP networks with a Siemens APACS DCS

Before you can troubleshoot a network,


you need the correct tools (Figure2). refers to how you transmit the ones networks multiple times and took traces
The two key tools for Profibus are a bus and zeros. Profibus DP uses a modi- and monitored performance.
analyzer and an oscilloscope. We used fied RS-485 physical layer which uses
a Profibus bus monitor called ProfiTrace a differential signal to transmit the in- The visual inspection proved to provide
which has a special PA probe so that formation. Profibus PA uses a far more the most information. Technically, at
you can view the messages right on complex method called Manchester En- installation several Profibus DP wiring
the Profibus PA side of the network. For coded bus powered which is defined in guidelines were violated by mixing cable
troubleshooting, we used a low cost IEC 61158-2 standard. This is a method types, and not using powered termina-
USB oscilloscope which worked well for where power and communications can tors (see Figure 3). However, the net-
this purpose. The oscilloscope is useful be done on the same two wires. work was running at 9600 baud and the
because you can get a lot of information copper wires were going less than 100
from looking at the Profibus PA wave- The fist step in troubleshooting Profibus meters, which meant that these viola-
form. DP was to do a visual inspection of both tions would have had no impact on the
Profibus DP networks. Then we connect- error rate. Also during the visual inspec-
Since there was great number of field ed the Profibus bus monitor to both DP tion, we noticed that a couple of the
devices, Simatic PDM was a key trouble-
shooting tool. As it turned out the key
problems at this site were not instru-
ment based, so we ended up just using Figure 2: Trouble shooting tools
Simatic PDM to confirm that all the in-
struments were configured properly and
had no errors. This is important because
a badly set-up instrument will generate
diagnostic requests on the bus which DP cable
might look like a bus problem at first Oscilloscope
sight.

Troubleshooting Profibus DP ProfiCore

When troubleshooting a large Profibus


PA system, you should always start with CP 5512
looking at the Profibus DP side. One of
the advantages of Profibus is that in
general you use the same troubleshoot- PA Probe
Tap cable
ing equipment for both Profibus DP and Oscilloscope
Profibus PA – the only real difference connection kit

is the physical layer. The physical layer

2
Trouble shooting Profibus - A practical example

Figure 3: One of the panels shows violation of Profibus DP wiring guidelines Figure 7: “T” connector junction box with M12 connectors

antennas were not mounted correctly. It Below is a summary of the four major In the case of segment 2, the waveform
looked like they had not been tightened problems that we found, and how they peak to peak is over 800mV. The distor-
enough and had slipped. Once we looked were determined. tion at the top and bottom is similar to a
at the Profibus bus monitor and looked at segment that is missing a terminator. Just
the retry rates to the different stations, Problem 1 looking at the waveform, we thought
we saw that the ones with the largest that we were examining a long run with
retry rates were also the ones with the Figure 4 shows the waveforms from seg- a missing terminator. However, the run
mounting issues. ment 2, which with its angled peaks and was under 500 meters, which is not that
valleys shows there is a problem. Figure long; so what else could be causing that
Fixing the antennas improved things to 5 shows an ideal wave form in Profibus level of distortion?
a point where the DCS was no longer PA. Figure 6 shows the waveform of a
getting Profibus error messages. We were short run with no terminator. After verifying that there were indeed
still getting retries, but given the high two terminators on this network, we
electrical noise and the number of ob- We call this a problem in the sense that started looking at how it was wired in de-
stacles in the Chemical Plant, there were the waveform is not consistent with the tail. This required reading some specifica-
limits to how much we could improve the ideal waveform; the peak to peak volt- tions and doing a few lab bench tests.
performance of the network. age is too high and the peaks and valleys
are distorted. When we attached the bus We found the answer at the T-connectors
Troubleshooting Profibus PA monitor to the segment, we saw zero and cable types that had been mixed
retries and zero errors on this segment. (Figure 7). The site was using Siemens
The Profibus PA segments proved to be All the instruments were properly func- T-connectors and Belden Profibus PA ca-
far more interesting and far more chal- tioning, but we were not content with ble. The Siemens T-connectors are vam-
lenging. The first phase of the investiga- the overall result. We were also getting a pire type connectors and are designed
tion was to attach the Profibus bus moni- similar waveform from most of the other to match the impedance of the Siemens
tor and the oscilloscope to each Profibus segments, many of which were having Profibus PA cable. The IEC 61158-2 stan-
PA segment to get an overview of what communication problems. Something dard does not state an exact specifica-
was going on. was not right. tion for class A cable. It states a range.
Therefore one manufacturer will pick one
Since these segments were located in An ideal waveform has a peak to peak set of values and another vendor will pick
intrinsically safe areas, we needed a hot voltage of around 800mV (Figure 5). If another. When you mix cable types, you
work order before we opened each cabi- you are missing a terminator, then you get reflections similar to what you get
net. Each area was checked for chemical get a peek to peak voltage of around when you are missing a terminator; only
vapor and all necessary people were noti- 1.5 V and distortions at the top and bot- to a lesser degree. In this case, the site
fied, which took about an hour for each tom (Figure 6). Also, as you increase the was unknowingly mixing cable types.
cabinet. Checking out a large network in size of the network, the peak to peak We concluded that given the size of their
an intrinsically safe area requires a con- voltage is lowered. wire runs, this was not a problem for
siderable amount of time. communications; the bus monitor sup-

3
Trends in Instrumentation

Figure 4: Segment 2
waveform; the angled
peaks and valleys indi-
cate problems.

Figure 5: Ideal wave


form; a symmetrical
form shows perfect
communication of
signals.

4
Trouble shooting Profibus - A practical example

Figure 6: Waveform
when missing one
terminator.

Fig. 8: Segment 1.

5
Trends in Instrumentation

Fig. 9: Trace from Segment 1: showing framing errors and illegal messages Fig. 10: Temperature transmitter incased in water

ported this. Had this been a non-IS envi- it was natural to assume that it was the transmitters, water started to shot out
ronment with long wire runs, then this electrical noise that was causing this from around the seal and we saw what
wire practice would have been a prob- problem. However, once we solved that the problem was (Figure 11). The unit
lem. The ultimate solution would have problem by drying the connectors and was installed correctly except that where
been to switch out the T-connectors for tightening them, the communication the conduit ended it was not sealed (Fig-
Profibus PA junction boxes. problem with the Sitrans LR250 did not ure 12). Water collected via the conduit
correct it self. From prior experience, we until the temperature transmitter was
Problem 2 know that another cause of the framing completely under water.
errors and illegal messages could be two
Figure 8 shows the waveform from seg- devices with the same address. So we To our amazement we could still com-
ment 1. went up to each field device and veri- municate to all the radar devices on this
fied the address. We found two devices segment and it was in full data exchange.
When we looked at this, we figured that with the same address. The solution was When we tried to do a download with Si-
there were interferences from power simple to change the address in one of matic PDM, we received communications
cables or Variable Frequency Drives the devices to the address that it was errors on the long messages, but given
(VFD). However, when we traced the wire supposed to have. Once we did this, the that several of the instruments were un-
runs, we could not find any sources for bus errors disappeared. derwater, we would not have expected
electrical noise. During our wire tracing any communications at all.
procedure, we opened the cabinets that Problem 4
held the T-connectors with the M12 con- These were also the instruments that the
nectors. These cabinets were not sealed On segment 10, we had the power light customer was reporting losing connec-
correctly and had water dripping from on the Profibus DP/PA Coupler blinking. tion with – not a big surprise. Once the
the M12 connectors (Figure 7). Also, the Since we had only ever seen this light as instruments that were underwater were
M12 connectors were connected with a solid green, we checked the manual. replaced and the conduits were properly
only a half turn. A tight connection re- The manual said that this indicated that sealed, this problem disappeared.
quires several complete turns, and these the segment was drawing too much
connectors were just barely touching. power and the current was being limited. Conclusions
After we dried tightened the connectors On this segment we had a number of
the waveform cleaned up. Sitrans LR250 level and Sitrans TH400 This site visit definitely showed how
temperature transmitters. We were com- robust Profibus is. With the site installa-
Problem 3 municating to the level devices but had tion issues we found, the network should
lost communications to some of the not have even worked. Yet the operators
On this same segment, we were also hav- temperature transmitters. Therefore, we were getting data which met their update
ing trouble communicating to a Siemens thought we were probably dealing with requirements.
radar level transmitter (Sitrans LR250). an overloaded segment – too many de-
We connected the Profibus bus monitor vices and/or too long. The systematic troubleshooting method
program and saw Figure 9. that we followed worked very well. How-
Given the waveform shown in figure 8, When we opened one of the temperature ever, interpreting the data required some

6
Trouble shooting Profibus - A practical example

Fig. 11: Conduit open at the top for rain

art. A test result can have more than one


cause. Noise on the waveform that can
be caused by electrical noise, can also be
caused by a combination of poor connec-
tions and water. A waveform that looks
like there is a missing terminator can be
caused by mixing cable types. Bus errors
that can be caused by a noisy line, can
also be caused by having two devices
with the same address. A segment that
is in overload condition can be caused by
having too many devices, and can also
be caused by having a device with water
problems.

Things are not always as they first ap-


pear. However, once you fully investigate
a problem methodically, the real cause is
not hard to find.

Reference:
‘A Guide to Troubleshooting Profibus PA
Networks’, James Powell, P.Eng. James Powell, P.Eng. is Product Manager
Communications Systems at
Siemens Milltronics Process Instruments Inc.

© 2008 Siemens Milltronics Process Instruments Inc.

www.siemens.com/processautomation

7
Trends in Instrumentation

Siemens AG Printed in Canada


Industry Sector © Siemens Milltronics Process
Sensors and Communication Instruments Inc. 2008