DanW

My original reply should have read: ------------------------- If device C (not D) is on an ethernet network, it has an IP address. The IP address (not the slave node ID number) is used for identifying the slave. ------------------------- Your device A can not specify more than one transaction, where each transaction uses a different IP address?

Device A is the master. Device C is the slave. If device D is on an ethernet network, it has an IP address. The IP address (not the slave node ID number) is used for identifying the slave. For reading registers 0-9 from any device, Device A master needs to query each device by its IP address, for instance, first Device D at 192.168.xxx.xxx and the second Device D at 192.168.xxx.yyy

I'd be upset if it were a new installation with racks installed out-of-numerical sequence. But for an update that needs an additional rack, the new one goes where there's space the existing ones don't get touched. The potential for problems involved with moving and probably re-wiring the existing racks isn't worth the satisfaction that sequential numbers give to those of us with OCD (Obsessive Compulsive Disorder). Now if it involved covering cracks in the wall, that's a whole different requirement . . .

I googled SCADA topologies (spelled correctly) and got 1) Guide to Supervisory Control and Data Acquistion (SCADA) a US government by the US government's NIST (the governmental standards organization) http://www.cyber.st....ty%20(2007).pdf 2) Rockwell Automation's SCADA guide http://literature.ro...sg001_-en-p.pdf 3) The US Army's SUPERVISORY CONTROL AND DATA ACQUISITION (SCADA) SYSTEMS FOR COMMAND, CONTROL, COMMUNICATIONS, COMPUTER, INTELLIGENCE, SURVEILLANCE, AND RECONNAISSANCE (C4ISR) FACILITIES APPROVED http://armypubs.army...pdf/tm5_601.pdf You can't get much more North American than NIST, Rockwell Automation and the US Army

Google can help with your homework.

Yes. A master can be anywhere on a multidrop network. Note that the biasing resistors and the termination resistor are at one physical end, with the other termination resistor at the other end, without regard for where the master is.

In general, ultrasonic is not steam friendly. Ultrasonic signal strength fades badly when the transducer gets condensation on it. The suggestion for Rainex is a common suggestion and it does have a good effect, but it's high maintenance. Radar is somewhat better in steam. If you have access to the bottom, have you considered an impulse line to a head pressure transmitter? A damping value will smooth out some of the turbulence that pressure 'sees'.

Not all devices comply with NAMUR, but most have some form of fault detection and alarming outside of the "normal under/over operating" range. which as NAMUR considers 3.8mA to 20.5mA. But you have to check each field device type to see what the fault range is. Your meter might use 3.85mA and 20.8mA According to the Publication 1756-6.5.9, ControlLogix Analog I/O Modules (Cat. No. 1756-IF16, IF6I, IF8, IR6I, IT6I, OF4, F6CI, OF6VI, OF8) http://www.pacpro.co...ix_If16_of8.pdf your AI card has a range of 0-21mA, so you scale 4-20 mA and recognize that negative ranges less than (the eng unit equivalent of ) 3.80mA or positive ranges greater than (the eng unit equivalent of ) 20.5mA (using NAMUR values) are alarm states and alarm accordingly.

I can't answer for specifics of A-B analog inputs, but there is a general specification that might address what you're tackling - NAMUR. NAMUR is a German association that publishes recommended practices for instrumentation. Many field instruments (optionally) use NAMUR settings for fail-safe conditions, I say optionally, because the factory settings are typically not NAMUR, but NAMUR fail-safe levels can be selected if the option is available. Here's a graph of the NAMUR NE-43 range for 4-20mA. I suspect that some systems (PLC/DCS) allow for an input selection for NAMUR which allows detection of these fail safe levels, but doesn' t affect the scaling of the PV signal in the 4-20mA range.

Two technologies: 1) Head pressure with a submersible pressure transmitter. Sensor head comes with a long coil of cable and a vent tube. Sensor is dropped into the pond and it rests on or near the bottom. The head pressure of the water above it produces a 4-20mA output signal. The measurement is the water level over the transmitter. It is powered with a 24Vdc power supply (industry standard). The 4-20mA can go long distances over copper wire. Somewhat vandal resistant because the cable on the shore can be run in conduit to a panel on pole, with the sensor in/under the water. 2) Ultrasonic level. An ultrasonic level is mounted on a stationary fixture a couple feet (600mm) above highest level conceiveable. The sensor generates an ultrasonic piug that bounces off the water/floating debris/ice and is converted to level (assuming some arbitrary zero at some depth) or distance, the distance from the stationary transducer) and provided as a 4-20mA signal. The instrument is usually AC powered, but can be had with DC power that can run from battery/solar, particularly if the update is in frequent, like hourly or twice a day. This technique is also used extensively over a flow weir where the level is used to calculate flow rate. The PLC would need an analog input compatible with the 4-20mA output of the sensor.

The lack of hard biasing, pull -up or pull down resistors that drive the bus lines to a known state during the idle condition when no driver is active can cause a false data bit, which the UART can interpret as a start bit. I suspect that the absence of data following a start bit could show up as FF because I think a logic zero shows up as a one in serial voltage logic. It might be worth looking into biasing resistors. Pages 7 and 8, Failsafe Biasing, in the Analog Devices RS-485 Circuit Implementation manual gives a sample calculation. http://www.analog.com/static/imported-files/application_notes/AN-960.pdf

Pennsylvania was the hub of process instrumentation from WWII through the 1990's. Moore Products (now Siemens) was/is in Springhouse PA. Honeywell (what's left of them) is in Fort Washington, PA. Red Lion Controls (HMI's, comm boxes, panel meters, a modular control system) is in York, PA. Phoenix Contact is in Middletown, PA. Brooks Instruments (mass flow meters) is in Hatfield, PA Kobold is in the Pittsburgh area, but I think they import/private label most of their stuff. The analytical instrument field, IR/UV spectrophotomers, mass spectrometers, chromatographs, etc. all use embedded processing. They all exhibit at the Pittsburgh Conference, which is never in Pittsburgh.

Good luck with IEEE. I might have missed the connection, but I've found that IEEE is not a big in the systems integration world. PLC manufacturers employ engineers who might belong to IEEE, but I'm not sure IEEE offers much to automation guys. Linkedin just hid its "groups" under "interests" on the navigation bar. Some of the automation groups are Automation & Control Engineering Industrial Automation and Controls Network Automation Automation Engineers MCAA Measurement Control and Automation Association PLC / SCADA / MES Worldwide Community PLC SCADA Engineer Process Control technology-specific groups Industrial Wireless Level Measurement Vendor-specific groups like DeltaV (Emerson's DCS)

Maybe try using even or odd parity, in hopes that the bogus leading byte will be rejected.

Your response is missing the one of the two CRC bytes; it should be 4E 03 02 00 4E 6D BE but I don't think that's your problem when the low order half the CRC is correct. It appears that the master is looking for a response too soon, earlier than the response is actually arriving and has time to receive what it thinks is a start bit and the following line logic state that it interprets as all ones. I'll bet that you're probably running 8-N-1, with no parity so there's not even a parity check on the bogus data word. There's probably a setting (sometimes called turn around) that tells the master to wait x milliseconds before looking for a response. Try increasing that value to see what happens.

Check out the answers to the same question on control.com: http://www.control.com/thread/1367509710

I don't know if electrical noise issues, which would force continual re-sending of packets until the data gets through, can cause a swithch to go from full duplex to half duplex or not. But if so, when I sat through a Profinet seminar, the issue of electrical noise from drives was brought up. The consensus was that shielded cable for ethernet communications and shielded power cabling were essential for efficient communciations.

George, Good for you. Hopefully you can bring some insight and answers about DCS systems to the forum. Because of the proprietary nature of DCS and the limited distribution of documentation, many inquiries regarding a DCS go unanswered. Dan

I don't mind, but it isn't mine. I found the graphic in a NAMUR slide or document some time ago, so they're the owners, and I can't give permission to use something that I don't have title to. But if you right click and select "save picture as" . . . .

Bernie, Isnt' it amazing how many inquiries stop at the first step? Do you suppose they can't find their way back to where they posted the question? Did someone look over their shoulder and point out the hidden right click option (or whatever?) Do they just live with the problem ? It's a mystery to me . . . Dan

You could start with Prosoft's training video http://www.youtube.com/watch?v=nt_dSat7Q20

No. The purpose of 4mA as a minimum is to allow the transmitter to draw its power from the loop itself to avoid the requirement for power wiring. This is known as 'two wire loop powered'. The demand for two wired loop powered instrumentation came from the process industries that dominated the Instrument Society of America (ISA) at the time: refineries, steel mills, power plants, and paper mills. Those industries were central control room panel board oriented and home run wiring to the control room (or marshalling panels) was the norm. Running a 3rd or 4th wire for power would have been a deal breaker for anyone contemplating 3 or 4 wire field instruments. In 2 wire loop powered applications, the transmitter consumes about 3.6mA for its sensing and operation. All field instruments offer a choice between fail-low or fail-high indication, so the region around 3.6 to 3.8mA is used for fail-low indication, as the graphic from NAMUR shows. The 20mA span followed the the ratio of the pneumatic predecessors, 3-15psi, where the span to elevated zero ratio is 5:1. The indication of open circuit at 0mA is a great feature of 4-20mA but it wasn't the driving factor, 2 wire loop power was.

A nice reference set of wiring diagrams for open collector, specifically for Signet, but really universal. http://www.icenta.co...ctor_output.pdf National Instruments has a 2 or 3 page document on frequency/quadrature input I/O, which is what your board is. http://www.ni.com/white-paper/7109/en But I suspect you already know that from what you've explained, since you're wired to A/Areturn and source 24Vdc from the card. This is E&H's minimalist documentation: If you put A and A return across the pull down resistor, then you should have the voltage drop across the input. I know 10K ohms is common for pull-up/down resistors, limiting current to 2.4mA. I'm guessing that this statement says 25mA is the maximum current rating (page 25) "For continuous currents up to 25 mA (Imax = 250 mA / 20 ms)" but it's not crystal clear to me. A 1200 ohm resistor in a 24V circuit would limit the current to 20mA. If it were me, I'd start high and work down. Better to limit too much current than too little. (R = E/I; 1200 = 24V/0.02A)

I so seldom run into Modbus ASCII that I have to stop and think. Do you REALLY mean Modbus ASCII, not ASCII characters as data over Modbus RTU? I assume you do because you have the colon delimiter, but it's important because - Modbus ASCII uses 7 bit words, RTU uses 8 bit words - Modbus RTU uses CRC check and Modbus ASCII does NOT use CRC, it uses a one byte LRC value for error checking (does not include delimiters, colon or LRCF). - An ASCII message must start with a 'colon' ( : ) character (ASCII 3A hex), and end with a 'carriage return – line feed' (CRLF) pair (ASCII 0D and 0A hex). - Only allowable data characters are hexadecimal 0–9, A–F (ASCII coded). - timing differences that I can't recall. I think the RTU/ASCII message formats are common, but at this point, it's foggy. The graphic below is an example from the Modbus spec. The slave node ID is not included in the example. If the formats are the same, then you're missing the byte count register and the data bytes all of which contribute to the LRC calculation. The Modbus spec (MODBUS over serial line specification and implementation guide V1.02) has info on LRC calculation Dan

What's a 'containment locker'? What is 'gas level' and its associated setpoint? Liquid seeks its own level but I'm not sure what 'gas level' is.