Pitor my comments embedded ----- Original Message ----- From: "Piotr" <[email protected]> To: "EMC-PSTC" <[email protected]>; "Chris Wells" <[email protected]> Sent: Monday, February 09, 2004 7:06 AM Subject: Re: RS485 and CM choke
> > ----- Original Message ----- > From: "Chris Wells" <[email protected]> > To: "'emc-pstc'" <[email protected]> > Cc: "Piotr" <[email protected]> > Sent: Saturday, February 07, 2004 3:06 PM > Subject: Re: RS485 and CM choke > > > > Are you avoiding Optoisolation (and associated DC:DC) due to cost, > space > > both? > > Due to cost. OK - remember the cost of various cable options too. Customer cares if both your product and installation are good. Course he will see your cost first :-) > > > There are few cables available that have the 3rd wire integrated > without > > having to use a 3rd wire off a second data pair for this purpose. > > What most industrial RS485 users do is use the shield as the return. > > And so have a driven shield. > > The shield would be AC referenced to ground at each node > (.0047 -.0022 uF Y2 > > to ground and 1M bleed off resistor) > > I expect this will raise the hair on the backs of the shield purists > here > > but I challange you all to come up with a solution that would allow > the > > customer to use a single twisted pair cable with shield that would > support > > the needs of RS485 and various code application issues. > I'm not sure what the charachteristic impedance is. It could be really off Note that if you used the solid twisted (instead of stranded) you could drive some current on it. I did this on one display design. RS485 on one pair and then power on two others. I kept the power lines twisted. I took it about 500M but only 19.2K baud > I supposed the most popular is the Ethernet cable 4 pairs or 2 pairs > no shield. One can get shielded Ethernet cable..... But the balancing of the lines and the number of twists to reject Mag field is almost as important. In fact I have seen UTP beat STP in comparisons for Ethernet. I thought of both wires from second pair used as 3rd wire. > As no shield I was trying to find solution with my CM chokes. You can do a lot with just a Ferrite bead around the cable Now I am > thinking rather (not sure yet) of shielded cable, and second pair as > third wire, This is where the installation cost comes in. The extra pair will add about 1/4 to 1/3 on cable cost. That is why shield is often used. a shield may be AC referenced to ground, blocked at some > level by transils. All happy ? (I suppose not) I am using shield as are many industrial companies. > > > Distance > > At 115Kbaud - 1KM is a long distance > > I will consider reducing my needs to 500m. > > > You will have to consider the RC loading of the cable (~35pf/M?), > loading of > > the transient protectors and the loading of the transcievers > themselves. > > Check out the 1/8 loading transcievers like this one from Maxim > > http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1522/ln/en > > > > You will want to specify a low capacitance cable for this distance > and limit > > the number of nodes on the link. > > If you use 1/8 loading transcievers and limit system maximum number > to 32 > > you may be able to go the distance at this baud rate. > > However you have to consider some transient protection. > > If the cable is well terminated (R=Zo) than its pF/m seems be > unimportant. NO - not true RS485 needs a crisp rising edge on square wave for serial port to catch proper edge of asynch byte. If the rising edge is rounded more than 10% of bit width then many bit and framing errors will ruin your link!. Use datalines coated with polyetheylene (spelling correct?) they are much better than PVC. The transient protectors loading I am afraid but in the > sense of Zo discontinues and reflections, but not in the sense that it > adds to transceiver loading Not using proper cable impedance missmatches Tx to cable. It will also cut down on output signal level and so distance.... But 4 ohms each data leg helps you use smaller TVS. which I think is the DC problem and not > AC. Before end decisions I will check selected transient protector > loading in practice Look at the capacitance of TVS TVS 600 joule is about 700 pF while special ones are only 30pf. The are really TVS in series with regular diode that has low junction capacitance and can take the hit. (I have here 1200m of 2 wire cable). I don't know > how to calculate its effect on transmission line. I hope that using > transceivers with it data rate reduced helps a lot. I can estimate the > time needed to load theirs capacitance and compare with my bit time. > > > About 4 ohms on each data line and 100 ohm on return to terminal > helps a > > lot. > > I supposed the 600W TVS is OK (if 18V) for transmission lines (1kV > Surge with 42 Ohm source resistance (=25A) ) and need no current > limiting resistors. > If I use shielded cable my needs goes down. Yes shield helps! > > > It will effect your impedance matching but the protection is worth > the > > compromise. > > I don't believe in impedance matching of RS485 transceiver. If it is > receiver it don't loads line - OK. standard RS485s are about 12K input reciever impedance The 1/8 loading are 8 times higher input impedance. If the power is off the impedance drops significantly 12K becomes 5K The reciever loading does mater The minimum Rx detection level is ~ 200mV and it can be reached easily due to Rx loads. Note one trick that helps is to use RC termination. Not sure of the exact value - something like 100 ohms in series with 100 pf. This helps keep the signal level up. If it is transmitter I think it has > lo impedance. The signal travels towards both ends and hopefully don't > reflects. As only one transmitter can be at any time it is OK. If > there were two each of them would make reflections of signal send by > the other one. > > > You must use a low capacitance TVS or your baud rate will not work. > > Check out biderctional 500/600 joule TVS with built in series diode > that > > pulls capacitance way down > > http://www.microsemi.com/datasheets/MSC1701.PDF > > http://www.microsemi.com/datasheets/srlc05.pdf > > http://www.microsemi.com/datasheets/sd26a.pdf > > Thanks, I was waiting somebody points it. I couldn't find in my notes > where to find lo capacitance TVS. I think series diode gives not a lot > until the main TVS capacitance is loaded. So beginnings of > transmissions can be corrupted. But I have never checked this my > opinion. > > > You said you need to deal with CM noise up to 100MHz > > You could simply add two ferrite beads on the data lines along with > the > > series resistance. > > I know it. I see the main problem at low frequency band end. > > > Check out http://www.steward.com/ and look for beads to give you ~ > 50-90 > > ohms at 100MHz. > > Opt for current rating that won't saturate quickly (1 Amp). > > Why 1A. I was thinking about 1kOhm/100MHz and about 120mA DC current > rating. > Note that CM currents at high RF can get to be much larger than 120mA I would push for ferrite with 1AM rating. > Piotr > > This message is from the IEEE EMC Society Product Safety Technical Committee emc-pstc discussion list. Visit our web site at: http://www.ieee-pses.org/ To cancel your subscription, send mail to: [email protected] with the single line: unsubscribe emc-pstc For help, send mail to the list administrators: Ron Pickard: [email protected] Dave Heald: [email protected] For policy questions, send mail to: Richard Nute: [email protected] Jim Bacher: [email protected] All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc
