Post by thread:[EE] possible ways of dealing with noise on a seri

This question is completely theoretical. I'm just curious as it pertains to something I may be building in the near future. Lets say I have 25 devices spread throughout a building connected through a low speed (50Khz) synchronous serial bus at 5 volts. Lets say we have 50 feet (or so) of cable sloppily daisy chained in between each of the devices (around high amounts of florescent lighting). Would noise be a likely issue with the bus ? If so what would be an effective way of dealing with it. (differential signaling, perhaps ?) Thanks Jason Whit

Hi ... I couldn't give you specific electrical calculations for what the noise might be, but I'd never propose building a setup like this. My first suggestion is do it with rs485 (differential) which is designed for exactly this purpose, run it at 57600 baud and you're all set, you'd get near the same bandwidth as the 50Khz synchronous. Or use 115200 baud and have near 2x the bandwidth you'd need. Wiring-wise definitely definitely use a twisted pair for the rs485, and for extra certainty shielded cable. You'll need a ground wire between all the units as well, unless you use isolated rs485 (a lot more expensive, like $14 a transceiver vs. $1). Shielded cat6 cable makes great rs485 cabling and it is pretty cheap. But for these low speeds and relatively short lengths just having the rs485 be a twisted pair and you're all set. Is there a reason to use sync serial vs. async? Note that using a data format that can detect errors may avoid the need for any consideration at all of signal integrity. If you're sending packets of data out to or between these devices, add a CRC and have acknowledgement replies from the receiver back to the sender, and if errors happen the sender then does retries etc. and then you can have a poor link and no problems. If 9 out of 10 packets are garbled, but you just retry until success, well, hey that may be just fine. (or not, depending on your application...) Anyway my 2c. J Jason White wrote: {Quote hidden}> This question is completely theoretical. I'm just curious as it > pertains to something > I may be building in the near future. > > Lets say I have 25 devices spread throughout a building connected through a > low speed (50Khz) synchronous serial bus at 5 volts. Lets say we have 50 feet > (or so) of cable sloppily daisy chained in between each of the devices > (around high > amounts of florescent lighting). Would noise be a likely issue with > the bus ? If so > what would be an effective way of dealing with it. (differential > signaling, perhaps ?) > > Thanks > Jason Whit

> Is there a reason to use sync serial vs. async? Cost savings, sync. serial is easy to implement in software. I plan on using a very low end PIC without a UART module. > If you're sending packets of data out to or between these devices, add > a CRC and have acknowledgement replies from the receiver back to > the sender, and if errors happen the sender then does retries etc. and > then you can have a poor link and no problems ... I think that would be ideal for this application. Though bus bandwidth will suffer, it suppose its well worth it. -- Jason White

At 01:05 PM 5/25/2012, Jason White wrote: >Lets say I have 25 devices spread throughout a building connected >through a low speed (50Khz) synchronous serial bus at 5 volts. Lets >say we have 50 feet (or so) of cable sloppily daisy chained in >between each of the devices (around high amounts of florescent >lighting). Would noise be a likely issue with the bus ? If so what >would be an effective way of dealing with it. (differential >signaling, perhaps ?) First question: are all nodes electrically isolated from Earth Ground? In other words, are they all floating with respect to each other except for the serial cable? If so, I'd be tempted to start off with current-mode signalling (similar to the old Teletype signalling method). I've had GREAT results with this over multi-thousand foot runs between multiple nodes. The advantage is that you use a single twisted-pair cable for complete half-duplex communications. I recall using about 10mA current with 9600 baud data rate (24Vdc supply). The current detector was a simple PNP transistor (E-B voltage drop) with another transistor for data transmission. Obviously, idle state is with current flowing. Several variants of this used 4-conductor station "Z" wire (22 AWG) where I sent both power supply and data down the 4 conductors. One system was two nodes only - transmitting from master to display. That variant used: +DC, Gnd, Clk, Data, where the receiver supplied the current and the master sunk the current to Gnd. Current sensors were PNP transistors from the display incoming +DC rail. Other variants used opto-isolators for the receive section. The main thing is to ensure that you have enough current to ensure that the cable capacitance is completely discharged when the current drops to zero. Its easy and it works. I used a fairly low data rate (9600 baud) for my stuff but there is no reason that it won't work at higher rates. dwayne -- Dwayne Reid <spam_OUTdwaynerTakeThisOuTplanet.eon.net> Trinity Electronics Systems Ltd Edmonton, AB, CANADA (780) 489-3199 voice (780) 487-6397 fax http://www.trinity-electronics.com Custom Electronics Design and Manufacturing