"Balanced signalling" vs "Differential signalling"?

[cupoftea]

Hi
As you know, “Balanced” comms systems have three wires, Ground, Signal1 and Signal2.
Signal1 being high means a logic high
Signal1 being low means a logic low
Signal2 is always equal and opposite in sine to signal1
Ie if signal1 is +10V, then signal2 is -10V

With “balanced” comms, the common mode voltage is zero?

Do you agree that a “Balanced” comms signal is a “differential” comms signal by definition?

...though with a “balanced” comms system, the two signals have equal impedances to ground, whereas with differential signalling, the two signals may not necessarily have equal resistance to ground?

Do you agree that balanced differential signalling, where the two signals are a twisted pair, and the ground is run along the cable shield (which surrounds the twisted pair along the cable length), is the most noise immune form of digital communications?

 

[D.A.(Tony)Stewart]

balanced refers to impedance source, to cable to load, like coax single ended or differential single channel with 2 wires not 3 and shield is RF ground on STP, using CML or VML with cable that supports the BW x length required. It is not balanced if each wire has a different impedance.

Differential means complementary entary signals and can be done in many configurations depending on interface with UTP short haul or STP .

CM chokes are std for ethernet and many other comm channels to improve CMRR where balanced mismatches are inadequate. For ethernet this is the most common mode source of CM immunity, not shielding. That is a supplement for better performance at high GHz-m products.

If UHF BW, RF ground shld. blocks grid & some SMPS noise using 75R + 1nF to centre tap to gnd at each end. Of the isolation Tx hybrid that also may have the CM choke depending on the speed and PHY connector or discreet parts.

 

[betwixt]

Caution - be sure you are understanding the use of the words in the correct context. Sometimes the term balanced is used to mean there is no overall DC in the signal stream, for example when a '0' is zero volts but a '1' alternates between +V and -V. This implies there is a reference zero connection but balanced data can be carried on two wires.

Brian.

 

[FvM]

In literature, the terms "differential" and "balanced" are treated as synonyms in certain context, see e.g. this quote from an USB 2.0 test spec:

The transformer converts the unbalanced (also known as single-ended) signal
from the signal generator which is typically a 50 Ω output to the balanced (also known as differential) and likely
different impedance loaded presented by the cable.

In another context, the meaning can be completely different, as shown by betwixt.

With “balanced” comms, the common mode voltage is zero?

Not generally.

Do you agree that a “Balanced” comms signal is a “differential” comms signal by definition?

Not neccessarily. The USB interface uses a balanced cable (mechanical & electrical), carries mostly differential but also single ended signals, e.g. a "single ended zero".

Nevertheless the intended differentiation between "balanced" and "differential" seems artificial to me.

 

[D.A.(Tony)Stewart]

For single supply differential the Vcm is always be positive DC, yet it will be balanced differential impedance to reduce reflection distortion. The Zcm can vary greatly with the design and frequency yet the CMRR is related to Zcm/Zdiff * Icm/Idiff = Vcm/Vdiff

 

[cupoftea]

Thanks, i must admit i have been googling for a chart showing each comms method and showing the voltages on the signalling wires, and how many wires there are, what data rate, and whether asynchronous or not. In particular, how many wires carry the "1"'s and "0"'s? Eg with differential signalling you sometimes have the complimentary signals alone being sent on the cable (2 wire signalling).....but sometimes you send the ground with them aswell (3 wire signalling).

But this is very difficult to find such a chart. Do you know of one?

I am trying to contrast CAN, Ethernet, USB, RS232, RS485, RS422, UART, SPI, I2C to see which is the most immune to noise. In particular with reference to signals that leave a PCB and go onto a cable.

Its very hard for example, to find the exact differences between CAN and USB.

It also appears that sending eg a 0-5V single ended signal over a cable is asking for trouble.....but is this still the case if you send it via a coax cable?

 

[KlausST]

Hi,

how many wires carry the "1"'s and "0"'s?

You must not see the signals as two individual signals, like the one is "1" and the other is "0".

example: 3.0V as well as 3.3V are considered as a valid TTL HIGH levels
* if an RS485 pair carries P = 3.0V and N = 3.3V then
P = valid TTL HIGH
N = valid TTL HIGH
but RS485 = valid LOW (Vdiff = -0.3V)
***********
* if an RS485 pair carries P = 3.3V and N = 3.0V then
P = valid TTL HIGH
N = valid TTL HIGH
but RS485 = valid HIGH (Vdiff = +0.3V)
***********
* if an RS485 pair carries P = 0.2V and N = 0.5V then
P = valid TTL LOW
N = valid TTL LOW
but RS485 = valid LOW (Vdiff = -0.3V)
***********
* if an RS485 pair carries P = 0.5V and N = 0.2V then
P = valid TTL LOW
N = valid TTL LOW
but RS485 = valid HIGH (Vdiff = +0.3V)

--> the absolute voltage value does not count for a differential pair
--> it´s the differnce voltage that counts
RS485 difference input voltage threshold usually is +/-200mV (may vary on type. Especially fail safe types use different thresholds)

Klaus