DF1 Polling Master with Radio Modems

Anyone have any experience, observations, insight etc. when configuring radio modems to function with an Allen-Bradley DF1 Polling Master with half-duplex slaves? I'm okay with Data-Linc, Esteem, Prosoft etc. However, a client of mine needs to stay on their licensed UHF (450Mhz) band due to path profiles, propagation, unable to install repeaters etc. Because of this, we are playing with 3 Kantronics TALON UDC models. They're set up in LINESUB mode (basically transparent) but aren't responding to the DF1 transmissions (no keying or TX) unless I set the PACTIME to 300ms or less. Even with that, recieved data from the slaves is not good enough quality for any SCADA application.
Anyone have any input? Rockwell and Kantronics have no insight, surprisingly! I've played with so many combinations of DF1 driver settings for a cause and effect excercise over the last three days I doubt that there is anything there. I've also been using/observing an RS232 LED (mark & space) sniffer/monitor with not much insight there either.

I've done DF1 half duplex over radio before. I've used a variety of radios in most vhf and uhf bands that are available for data, but think I've always used MDS when I've had AB hardware (And BTW, MDS does make a 450MHz model, and MDS is pretty much the caddilac of radio modems ... something to think about). I have never heard of kantronics, so i wont be able to offer much insight there. If you post a link to a manual I'll be happy to look and see if anyhting stands out to me (Feeling too lazy to look one up myself).

Off the top of my head, did you say the radio is not transmitting? Are your cables correct? the standard AB cable is a null modem, in this case you'll need straight through cables (most likely, depending on how wierd those kantronics are), and make sure the RTS/CTS lines are going through too and not jumpered back on to each other. Most devices dont need those anymore, but radios still tend to like em. You will also want to set the PLC ports to use handshaking.

Re the sniffer, have you tried it without the sniffer? Some of those things can draw enough voltage trying to power the leds that the recieving device doesn't see the activity. I always try to use the externally powered ones.


-jeff

I've never heard of these Kantronics radios either, but all serial analysis in my book takes more than "test and fiddle" approaches.

I like to set up at least one master and slave with a hardwired RS232 connection between them to absolutely verify that the controllers and ports and serial configuration are correct (for hardwire). The simplest configuration is to have the Master set for Standard Polling and just plug in the slave, and be sure its slave status bit goes high in the Master's status table.


Next, introduce the modems and begin examining modem settings, port handshaking settings, and wiring.

Post in detail some expansion on your comment:

They're set up in LINESUB mode (basically transparent) but aren't responding to the DF1 transmissions (no keying or TX) unless I set the PACTIME to 300ms or less. Even with that, recieved data from the slaves is not good enough quality for any SCADA application.

Click to expand...

What do you mean when you say that the modem is not responding to DF1 transmissions ? The slave controller should be the one doing the responding. Or are you referring to a serial or radio activity LED on the modem ?

What is this "PACTIME" setting ? [EDIT: I downloaded and read the user manual. PACTIME is a radiomodem setting for the delay between the end of data received by the modem from the controller and the modem keying up the radio to transmit]

How do you assess that the data from the slaves is "not good enough quality " ?

I found no references to Kantronics in any RA tech support cases going back seven years.

As always, post more information. Which controller models you're using and what port configuration settings you've made would be good baseline data.

The deal with the pactime setting could be that there is just too much delay and the AB is timing out. I used to know the DF1 protocol pretty well, but I can't remember how long it would wait for a response. Anyway, with a pactime of, say 500ms on both radios, that would mean a minimum of 1 sec before any response got back to the master. In reality, with 300 ms, when you factor in the other delays, you may be close to 1 sec anyway. And how fast is the PLC retrying? If its faster than the pactime, I wouldn't expect the radio to ever transmit.

-jeff

The "ACK Timeout" value in an SLC-5/0x controller set as DF1 Half Duplex Master is in units of 20 milliseconds, and the default is 50, so the Master needs to get a poll acknowledgement within 1 second.

If you're polling Node 1, for example, you'll see:

"10 05 01 FF" from the Master controller, with a reply of
"10 04" from the slave controller at Node 1.

Because the modem is not using handshaking, it needs some way to tell if the connected controller is through with its transmission. That sounds like what the PACTIME value does.

You should be able to turn the PACTIME value down pretty far, since the controller does not pause between bytes of a message. I see radiomodems run with data detection delays of 20 to 100 milliseconds regularly.