RS485 Pull Up Resistors

[fifty60]

Why do some RS485 circuits require pullup/pulldown resistors be installed in the circuit, while some RS485 circuits do not require the pullup/pulldown resistors? Is it because some devices provide the internal pullup/pulldown resistors already?

Or, is this also factor of how many devices are connected to the RS485 network? For example, if there is only a single master and a single slave is there a need for the pullup/pulldown resistors?

 

[LMAO]

Why do some RS485 circuits require pullup/pulldown resistors be installed in the circuit, while some RS485 circuits do not require the pullup/pulldown resistors? Is it because some devices provide the internal pullup/pulldown resistors already?

Or, is this also factor of how many devices are connected to the RS485 network? For example, if there is only a single master and a single slave is there a need for the pullup/pulldown resistors?

Never heard of RS485 "pull up/down" resistors; I believe you meant "termination" resistors. Termination resistors are used at each end of the bus to avoid signal reflections.

 

[rlundsrud]

To add to what LMAO said, the resistor is used to effectively dissipate the signal so there is no reflection back down the com line. This resistor must be matched to the impedance of the wire itself, this is typically 120 ohm for low-capacitance EIA485 wiring, but you should verify this before you install one. The way to imagine the signal cancelation without the resistor is think of emitted sound waves being canceled by reflected sound waves (like from hitting a wall). They won't cancel all of the sound but it will affect the shape of the resultant wave. In the case of the signal on your Com circuit, it is a square wave with the information across the top. The reflected signal removes the edges of the square wave giving it a rounded or a wedding cake appearance. This is enough to clip the data on top of the wave rendering it useless.

I hope that wasn't more confusing than helpful.

Bob

 

[templdl]

To add to what LMAO said, the resistor is used to effectively dissipate the signal so there is no reflection back down the com line. This resistor must be matched to the impedance of the wire itself, this is typically 120 ohm for low-capacitance EIA485 wiring, but you should verify this before you install one. The way to imagine the signal cancelation without the resistor is think of emitted sound waves being canceled by reflected sound waves (like from hitting a wall). They won't cancel all of the sound but it will affect the shape of the resultant wave. In the case of the signal on your Com circuit, it is a square wave with the information across the top. The reflected signal removes the edges of the square wave giving it a rounded or a wedding cake appearance. This is enough to clip the data on top of the wave rendering it useless.

I hope that wasn't more confusing than helpful.

Bob

Ditto. A good description. This was common with the Westinghouse system. This would be similar to the need to terminate an unused 75ohm TV cable line using a 75ohm resistor to absorb the signal instead of it being reflected causing signal interference to the rest of the system.

 

[GoldDigger]

The relationship between pull up/pull down and termination resistors comes about when you have a single ended rather than a balanced differential signal.

The termination resistance is needed to control reflections caused by AC components of the signal, particularly square edges (high frequency components.)

With a balanced signal, you can run a termination resistor from one lead of the pair to the other and as long as the driver can handle that load you have done the job of avoiding reflections.

But with a single ended signal you have the problem of where to connect the other end of the terminating resistor.
One example is a TTL logic level where + is ~5V DC and - is ~0V DC, but the switching threshold is between 1 and 2 volts.
If you connect the terminating resistor to either +5 or to ground, you run the risk of creating a DC load which can be too much for the driver to handle properly. One common solution is to use a combination of a pull up resistor to +5 and a smaller value pull down resistor to ground which together produce ~1.5V DC via the voltage divider effect and whose parallel combination resistance is the needed termination resistance.

For a large number of terminations, you might use a single resistor for each signal tied to a common well regulated 1.5V source instead of individual resistor networks.

 

[Bugman1400]

I've always thought that in serial-type comm (rs-232, rs-485, etc.), many of the devices that do not have RTS or CTS test the wire before they start transmitting on TX. So, to be sure that another device is not already transmitting on the wire, they test to see if the wire is a source, a sink, or Hi-Z. I thought the pull-up or pull-down resistors were to help with this.

 

[zbang]

I've always thought that in serial-type comm (rs-232, rs-485, etc.), many of the devices that do not have RTS or CTS test the wire before they start transmitting on TX. So, to be sure that another device is not already transmitting on the wire, they test to see if the wire is a source, a sink, or Hi-Z. I thought the pull-up or pull-down resistors were to help with this.

Real RS-232c (12v) and RS-423 (6v) are bipolar-unbalanced signals and only allow a single transmitter but multiple receivers. Some systems only send a unipolar signal, which might work, but isn't to-spec. RS-232c.
RS-422 is a bipolar-balanced/differential pair, also single transmitter.
RS-485 is also a bipolar-balanced/differential pair, but you can have multiple transmitters on the line (they're supposed to go to a high-impedance state when not sending).

Of those, only 485 can sense that another node is sending, and AFAIK only if it specifically implements it.

The only serial bus architectures that I can think of which implement collusion-detection as part of the spec are things like Ethernet, Arcnet, and some of the coax/twinax-based PLC and data networks.

======================================

I think we're being unclear about the purposes here- pull-up/down and termination serve different functions and go in different places- terminators go at each end of the line/bus, and the pull-up/down can go anywhere, but usually only once. The pull-up/down network may incorporate termination as the middle resistor, but doesn't have to.

The relationship between pull up/pull down and termination resistors comes about when you have a single ended rather than a balanced differential signal.

With a balanced signal, you can run a termination resistor from one lead of the pair to the other and as long as the driver can handle that load you have done the job of avoiding reflections.

But with a single ended signal you have the problem of where to connect the other end of the terminating resistor.

I'm with you on the balanced line termination, except that for almost any type of signaling, the terminating power dissipation is usually rather small since the drivers limit the current and you shouldn't use more terminators than called for. For an unbalanced line, there is no problem of where you connect the other end of the terminator- you connect to the signal reference.