[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Serial Connection

Stan Hoeppner wrote:
Moczik Gabor put forth on 3/28/2011 12:01 AM:
Stan Hoeppner wrote:
We bought DB25 plugs in bags of 100, and used spooled CAT5 as the noise
rejection is many times that of CAT3, allowing greater distances across
sprawling warehouses.
RS-232 uses single-ended signaling and requires shielded cable, twisted
pair doesn't help either.

Interesting to see you state this, considering the experience of many IT
folk stuck supporting legacy serial terminals and printers proves the
opposite to be true.  Back in 1996/7 I worked for such a legacy
equipment using company, a chemical producer in St. Louis.  We were
routinely (weekly) making 200+ ft. CAT5 UTP (24 AWG solid plenum cable)
runs across warehouses and production buildings.  The connections were
between RS232 Wyse terminals, thermal label printers, and ethernet
terminal servers at 9600 bps, using 3 conductors, one twisted pair
(orange/white-orange) and one single conductor of another pair (green)
for ground.  We could reliably run 115,200 bps on runs up to 50 ft.
using CAT5 UTP.  The cables distances we routinely achieved were
obviously many times the RS232 spec maximums.

Actually, I regularly use 5..15m (~16..50ft) RS232 cables at 115kbps without problems. Shielded non-TP cables, of course. :-)

Twisting RX and TX signals together is the worst that can be done. If it runs long enough, the transmit signal of one end will induced as noise to the transmit signal of other end.

Twisting RX and GND (TX and GND) together is maybe acceptable. I'm sure that it is worse than one-by-one shielded wires, but maybe better than two unshieleded.

For example, if you replace the RS-232 transceivers with RS-485, you can transmit the same serial port data on the same (TP) cable on the same bitrate, but magnitudes longer distance. 1200m (4000 ft) or maybe more achiveable, even at 115200bps!

The common-mode noise rejection of twisted
pair only works if the signals connected to a balanced input receiver.

Twisted pair cabled achieves its rejection performance against EMI/RFI
simply due to its design, not whether the signal is balanced or not.

Simply no.
Please read about balanced and unbalanced transmission.

An unshielded TP cable with unbalanced transmission is nothing better than a single wire.

The cable pair wires must be driven AND received with equal impedances to have great common-mode rejection. It doesn't matter that the signal levels according to GND are perfectly symmetrical or not, but the impedances must be equal.

Here are two (electronic) articles about balanced transmission (of audio signals, but the principles are the same):

any non-negligible length of the cable is in proximity to a noise
source, having twisted conductors minimizes the potential pick up area
of any one conductor.

No. The principle is that BOTH conductors of a pair receives the SAME noise, but the differential receiver substracts the two signals, cancelling noise.

If you measure (with oscilloscope) both signals according to GND, you will see that both wires pick up the same polarity and amplitude noise.

UTP have been used for telephones for more than 6 decades, and I'm
pretty sure everyone knows telephones don't use balanced signals.

Actually telephone is balanced. Telephone does not use the local ground at your house, it's inherently balanced.

The term "balanced" is often confused or mystified by electronic engineers too, but if you drive and receive two wires with equal impedance, and compare the signals to each other and not to (local) GND, that is a balanced transmission.

Reply to: