18
\$\begingroup\$

I have a serial comms project coming up. A number of boards with RS-485 interfaces need to communicate. Date rate is 57600bps and cable length will be less than a few hundred feet.

Will shielded CAT5 cable work well enough in this situation or should I pay the extra for "true" RS-485 cable such as this one?

enter image description here

I cannot easily visit the install site ahead of time as it's interstate. I do have the same boards here in the lab I can test with however.

\$\endgroup\$
  • 4
    \$\begingroup\$ What's "true RS-485" about the Belden cable? \$\endgroup\$ – stevenvh Jun 8 '12 at 6:30
12
\$\begingroup\$

While often shielded, CAT5 can be of the UTP type, meaning unshielded twisted pair. The "true" RS-485 cable you link to has two twisted pairs and a shield. If I remember right, CAT5e (and above) has a shield, at least most cables I have seen have - the exact standard can vary. I guess those would do the job just nice. Just make sure you use the twisted pair like in the original RS-485 cable (your data sheet looks like [1 White/Orange Stripe] and [2 Orange/White Stripe] are a twisted pair, and [3 Blue/White Stripe] is a third, untwisted wire.)

There seems to be a small impedance mismatch (100 Ohms for CAT5, 120 Ohms for RS-485). This will cause reflections at the driver and at the receiver, but I am pretty sure your application will still work. While you may read that 120 Ohms is typical for RS-485, the termination network uses 120 Ohms between the differential pair and 2 * 680 Ohms to VCC and GND.

RS-485 termination, source: wikipediaSource

Thus, the value of the termination that the cable "looks into" is smaller than 120 Ohms anyway: (120||(680+680)) Ohms = 110 Ohms.

If you have the chance to test, you could go for it. I would probably not even hesitate and use STP CAT5(e). If your contract says you pay money for every minute your installation fails, you probably want to use cables with the proper specification. (The latter will still not make sure nothing ever fails, but you're probably in a better position if you can blame it on the "true" cable instead of having someone else blaming your "wrong" cable. But you see how this last paragraph has nothing to do with physics...)

\$\endgroup\$
  • \$\begingroup\$ Cat5 cable has 100 ohms impedance, correct? How will this affect things if the RS-485 drivers are designed for 120 ohms? \$\endgroup\$ – SeanLabs Jun 7 '12 at 21:29
  • \$\begingroup\$ It will probably still work, my answer is updated now. \$\endgroup\$ – zebonaut Jun 8 '12 at 5:58
  • \$\begingroup\$ +1 for the last paragraph. Unless you're saving a ton of cash with no bad juju if your stuff fails, use the real stuff. If you're contracted to fix it / warranty it you don't want to give anyone a stick to beat you with. You also don't want to give anyone an opportunity to try and plug your RS485 thingy into an ethernet port and damage one or both - they WILL manage it one day. \$\endgroup\$ – John U Jun 8 '15 at 12:46
6
\$\begingroup\$

Generally speaking CAT5 s fine for RS485. IME the first limit you hit is the series resistance driving a termination over a long cable. I've run 250kbaud over 100m reliably. Things started getting shaky at around 200-300m.

\$\endgroup\$
4
\$\begingroup\$

Per reference to Maxim APPLICATION NOTE 3884 How Far and How Fast Can You Go with RS-485? from Jul 25, 2006 (cited 2104-05-28):

mentions rs485 and cat5 with measurements.

The performance of a Maxim driver (the MAX3469 in this case) and an equivalent >driver from another manufacturer are presented

What Factors Limit the RS-485 Data Rate?

The following factors affect how far one can reliably transmit at a given data rate:

  • Cable length: At a given frequency, the signal is attenuated by the cable as a function of length.
  • Cable construction: Cat5 24AWG twisted pair is a very common cable type used for RS-485 systems. Adding shielding to the cable enhances noise immunity, and thereby increases the data rate for a given distance.
  • Cable characteristic impedance: Distributed capacitance and inductance slows edges, reducing noise margin and compromising the 'eye pattern'. Distributed resistance attenuates the signal level directly.
  • Driver output impedance: If too high, this limits drive capability. Receiver input impedance: If too low, this limits the number of receivers that the driver can handle.
  • Termination: A long cable can act like a transmission line. Terminating the cable with its characteristic impedance reduces reflections and increases the achievable data rate.
  • Noise margin: Bigger is better. Slew rate of driver: Slower edges (lower slew rates) enable transmission over longer cable lengths.

System designers often choose a driver and receiver from two competing manufacturers, but most designers are primarily interested in how far and how fast the RS-485 driver can drive a signal. The performance of a Maxim driver (the MAX3469 in this case) and an equivalent driver from another manufacturer are presented

Signal integrity is tested by observing the driver's differential output. Set the oscilloscope to look for trigger points between the 80mV and -400mV thresholds. (These thresholds are chosen because receivers have an input range of 20mV to -200mV, plus a noise margin.) Then, when pulses (bits) begin to 'run together', use eye patterns to determine the overall contributions of distortion, noise, and attenuation to the parameter called intersymbol interference (ISI).

ISI forces you to reduce the bit rate to a level that allows an adequate distinction between pulses. Tests of the Figure 1 circuit show a consistent and clear correlation between trigger points and eye patterns. The eye patterns exhibit 50% jitter, measured using methods documented in National Semiconductor's application note 977[4]. Measuring jitter at 0V differential and ±100mV differential yields the data shown in Figures 4 and 5.

...
Depending on drivers you get different result

At 39Mbps and 340 feet of Cat5 cable, the driver output of Figure 2 exhibits an eye pattern in which signals cross in the middle of the eye—a condition indicating possible bit errors. The Maxim device at the same data rate, however, (Figure 3) shows no such condition. The Maxim transceiver offers better performance due to symmetrical output edges and lower input capacitance.

\$\endgroup\$
2
\$\begingroup\$

If you're using chip-level drivers on boards of your own design, tailor the termination networks for the CAT5 cable; essentially make it 100-ohm rather than 120-ohm. I used to do that all the time. My problem was often getting long enough put-ups of the esoteric 485 cable. I could get 500 ft. or even 1000 ft. put-ups of CAT-whatever cable. It wasn't meant to be used in runs that long, but it worked and it was a fraction of the cost.

\$\endgroup\$
2
\$\begingroup\$

We use CAT5 in the Security industry ( Card access controllers and cameras) for RS485 without any problem... 9600-19200 Baud.. with some cable length over 2000 feet.. works like a charm

Mario

\$\endgroup\$
  • \$\begingroup\$ That's good. Do you ever have to use isolators at all? \$\endgroup\$ – SeanLabs Oct 8 '15 at 0:16
1
\$\begingroup\$

The standard calls for a cable with an impedance of 100 Ohms at frequencies above 100kHz, DCR < 240 Ohms, and mutual pair capactiance < 20pF, with an AWG of 24 or larger. That's about it. Most CAT5 cable has an impedance of 100 Ohms (give or take), definately less than 20pF, and I don't think I've ever seen a cable with a DCR of 240 Ohms. So, in theory you should be fine. In practice I've used CAT5/5e/6 both UTP/STP with no issues on various devices from small serial to Prominas.

I tried to find the specification but I found a TI data shete that goes over the balanced/unbalanced test procedure, etc. http://www.ti.com/lit/an/snla137a/snla137a.pdf

The only exception that I can think of, is MIL-STD. It will call for a particular cable and the cable jacket should reflect that.

\$\endgroup\$
  • \$\begingroup\$ It'd take a fairly long cable to get the DC resistance to 240 ohms. (kilometers long) \$\endgroup\$ – Jasen Apr 22 at 21:37
1
\$\begingroup\$

twisted pair is twisted pair. so cat5 is 4 channels of rs485. It seems if there is a lot of RFI in the installation, shielded twisted cable can be justified. Rs485 is low impedance balanced , so twisting the pair helps with noise immunity as well as reducing the attenuation effect of higher baud rates due to the parallel capacitance of the wires in the cable jacket over the long lengths. These distances I'm talking about are greater than 100 meters.

Now there was certain instances I used STP wire for rs485 because I had a lot of devices connected in a long run chain (DMX512 lighting fixtures). And the results before was a degraded signal that the devices near the end would interpret the signal incorrectly.

unshielded cat 5 should work in your situation because of the voltage you will be running (5-10V) and your signal is not high frequency (above 500Mhz) in the first place.

But You stated you are going to install this at a commercial setting, so it always better to use the worst case environment situations, and use shielded cable.

I would insist also to shop around. beldon 3106A is what I typically use to wire balanced audio. Yes, it would work for that too, but 24 awg cable with plenum grade jacket is the standard.

\$\endgroup\$
0
\$\begingroup\$

From my experience with various HVAC controllers (Bacnet MSTP and Modbus), most of the time UTP cable works fine for short runs in low noise environments but can have intermittent and incredibly difficult to diagnose issues on long multidrop runs with differing equipment and incorrect termination. The labour involved in troubleshooting RS485 issues can quickly exceed the cost of using the correct cable. My advice is to be careful to use the correct cable and termination from the beginning or be prepared to replace the cable if it proves to be a problem.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.