Radar level gauges-Intermittant comm failures in loop

[Perrinoly]

All,
We recently installed a number of radar level gauges and we have the problem of ~3 units per loop (7-8 devices per loop) intermittantly dropping communication for short periods of time. They are placed in a sort of 'star' setup, and they come through a DIGI card at our input. Feel there is probably noise problems and I would welcome any ideas (hopefully from someone who has seen a similar problem) on methods of improving the integrity of this system. Thanks in advance.

 

[jsummerfield]

Is this a proprietary communications scheme like Enraf or an open scheme like Foundation Fieldbus?

John

 

[buzzp]

Network protocol needs to be known before anyone can help (RS485, RS232, ?)

 

[HEC]

What are the cable lengths, Protocol, Industry? Is it only one unit that drops off or is it equally shared? There is not enough information here to give sensible advice. Have you tried basic trouble shooting like shield earth loops. As a default installation standard I generally only earth shield at the board and have it cut back at the field device.

Mark Hutton
hutton4eng@picknowl.com.au

 

[Perrinoly]

Thanks for the replies. More info that might help is that the protocol is RS485 and that as stated before there are only 3 devices in each loop that are being affected. (our installation consists of 2 separate loops, each with a total of 8 devices). They both come in to a multi-port DIGI card that I am not too familiar with. The physical setup is a star topology. Cable length is a possibility but highly doubtful as it was spec'd out well and each loop is apparently capable of handling up to 16 devices. Thanks again for any input.

 

[buzzp]

Do you have any terminating resistors on the network? What about shields - where are these grounded? Which 3 devices of each loop are affected - the first ones on the bus? The last ones?
I realize everything was spec'd well but what about the cable length? Are we talking a few hundred feet or a couple thousand?

 

[CMfgE1]

P,
What kind of distributed control system are you using? (Siemens, ABB (taylor), Delta V, etc.)
Is it block programming or ladder logic?

And what protocol for sending the signal are you using?
(Foundation Fieldbus, Profibus, Analog (4-20ma), Asibus)

These are the kind of questions needed to determine a helpful solution. Also what kind of radar unit are you using? Manufacturer and cone setup.

Good Luck

Quote: "Its not what you know, its who you know" - anybody trying to find a decent job

 

[djs]

Hello;
RS-485 devices should be connected in a daisy chain fashion, not star. However this is really dependent on the transmission speed and cable lengths.
The problem with star connected networks is that there is no way to properly terminate the network. An unterminated network will have reflections which show up as noise. As the transmission rate increases, these reflections contribuite more noise.

 

[blackwed]

I totally agree with djs. And don't let anybody (boss or client) BS you otherwise. You may get a 485 system to operate in a star configuration for a while. The question though is not if but when it will fail. Sooner or later it will. And to do the daisy chain right, you should run your wiring from device to device not from box to box with jumpers from the boxes to the devices.