Published July 1999

THE ART AND SCIENCE OF RS-485

by Bob Perrin

Start • Arm Yourself • RS-485 101 • Getting Grounded • Shielding • Topology • Termination • Idle-state Biasing • Transients • Review Time • Sources

TRANSIENTS

ESD and capacitively or inductively coupled transients are a fact of life often overlooked when designing communication networks. Recently, I was part of an investigative team of engineers sent to a customer’s site to assist in determining why 200–400 of their 4000 RS-485 nodes were going down daily. The problem turned out to be transient voltages on the data lines.

The network had a mix of RS-485–based equipment on it. Several different manufacturers supplied the various pieces of equipment. The failures were mostly isolated to RS-485 receiver chips but were not isolated to just our equipment.

The failures had existed at a nuisance level for several years. Then late last year, the customer experienced a drastic increase in failure rates. By the time we were called, 10% of their nodes were going down each day.

Over the last few years, several network consultants had been brought in to address the network failures. None of them met with much success. By the time we arrived, the failure rate was at a catastrophic level.

The customer had done almost everything by the book. The network cabling was commercial CAT-5. The network topology was straightforward. The lines were adequately terminated. Each node had a power supply isolated from earth ground. The network cable had a wire dedicated to connecting signal grounds between nodes.

Each individual network consisted of 50–150 nodes and each node used a 1-UL receiver. Although this violated TIA/EIA-485-A, an oscilloscope verified that the transmission lines were carrying nice clean square waves of reasonable magnitude and offsets. And besides, the receiver chips were blowing, not the transmitters.

Most of the receiver chips were dual or quad devices. Autopsies performed on the damaged chips revealed that often only one receiver on the chip was blown; the others were usually functional.

After a while, it was clear that transient voltages were finding their way onto the data lines. We were not able to identify any single source or to nail down any single coupling mechanism. Even if we were, the facility was fixed and we probably couldn’t have altered the system to mitigate the source(s) or coupling mechanisms. We had to devise a method of eliminating the problem at the board level.

First, we had to find a method of mimicking the symptoms in the lab. To accomplish this, we used a Shaffner NSG-435 ESD gun to simulate transient events on the transmission lines. After building a small network in the lab and discharging energy into the data lines directly, we found that the most common receiver in the customer’s system, a TI 75175 quad receiver, was always destroyed with a single 2-kV air-gap discharge into either or both data lines. We saw one part fail as low as 1 kV. The most common threshold seemed to be 1.4–1.7 kV.

It’s interesting to note that a 1-kV air gap discharge is right on the edge of human perception. This means the receiver chips could be destroyed by ESD that may not even be noticeable to a human technician.

We tried two TVS schemes with the existing receivers. Both increased the ability of the receivers to tolerated transient events.

Figure 7a shows the simplest and most effective method. The circuit in Figure 7a seemed to protect the 75175s to about 8 kV. The tradeoff for good transient voltage protection is a fairly high capacitive loading. The TranZorbs used had an open-circuit capacitance of 500 pF.

Figure 7b shows our second experiment, which only protected the 75175s to about 4 kV. The circuit uses a bridge with a low capacitance (about 13 pF) in series with the TranZorbs. This is a fairly common circuit used to protect high-speed data lines.

a) (Click here for figure 7)	b)
Figure 7a—TVSs directly on the data line provide the highest level of protection and the highest capacitive loading of the transmission line. b—This is common circuit for protecting high-speed data lines.

Our experiments were done in haste, and although we maintained as much laboratory discipline as we could muster, further experiments should be run before the above thresholds of 4 and 8 kV are accepted as gospel. However, the results are certainly valid in a qualitative sense. Both TVS schemes provided significant improvement in the ability of the TI 75175 to withstand transient voltage events.

Our last experiment involved a Maxim part, the MAX3095. The datasheet for this part claims a ±15-kV protection using IEC1000-4-2 air-gap discharge, ±8 kV using IEC1000-4-2 contact discharge, and ±15 kV using the Human Body Model. Even though the Maxim part has only been out about a year, availability is good.

Using our ESD gun, we methodically zapped the Maxim part but were unable to destroy or even notably degrade the performance of any of the MAX3095 parts we tested. In a last ditch 4:30 A.M. attempt to get a failure point for the Maxim data set, we hammered one of the parts with 50 shots of 16.5-kV air-gap discharges. The NiCad battery pack on our ESD gun ran down, but the MAX3095 didn’t even blink.

We only had a small group of five sacrificial Maxim chips. So, once again, the limited sample set puts the quantitative value of our data in the dubious column at best. However, it is clear qualitatively that the MAX3095 is a rugged little part.

Maxim is infamous for long lead times, super-high prices, and lackluster customer support, but I’ve never heard of Maxim lying on a datasheet. I’m not a fan of Maxim’s aloof manner of doing business, but I do believe their datasheets and I’m totally sold on this little receiver.

Maxim has parts with high ESD ratings that are pin compatible with the widely used MC1488 and MC1489 parts for RS-232 applications, as well as other ESD-hardened interface parts.

In the end, we recommended trading out the TI 75175 for the MAX3095. These two parts are not pin-for-pin compatible in all applications, but for our customer’s equipment, the MAX3095 dropped right into the existing 75175 sockets and fired up.

The MAX3095 is a ¹/₄ UL part, which meant that we were also reducing the load on the network by 4×. The longest runs of 150 nodes were still slightly above the TIA/EIA-485-A allowable limit of 32 ULs (150/4- ~38). After installing the Maxim parts, the signal levels on all the transmission lines improved significantly.

At the time of this writing, our customer has over eight million machine hours on the MAX3095s and not a single failure of the Maxim parts. This was as close to a silver bullet as I’ve ever seen. Only time will tell if the MAX3095s will weaken with age and have to be placed on a preventative maintenance schedule, but it doesn’t look like that will be the case.

I learned one other interesting lesson from this trip. Beware the local customs. The customer’s maintenance crew was fairly sharp. Years ago, the technicians learned that the most delicate part was the receiver chip, so they adopted the custom of carrying tubes of these parts around and replacing the parts in situ.

This facility was one of the worst imaginable environments for ESD. Humidity was 10–17%. The crews were required to wear polyester uniforms and most of the facility was carpeted.

The maintenance personnel were not trained in basic ESD precautions. In the process of replacing damaged ICs, they were damaging the new ICs they were installing. Furthermore, the technicians would handle bare network cable during the repair, which meant they would discharge static electricity into the transmission, damaging other nodes on the network. Remember our lab tests where the TI 75175 failed at level of ESD that was barely perceptible to humans?

Also, cable contractors were often employed by the facility. These contractors would install or modify network cable to suit the needs of the facility’s ever-changing geometry. The contractors were handling bare network cable, with hundreds of nodes connected, and using no ESD protocol.

Our customer has since trained their maintenance personnel in proper ESD protocol. As a matter of contract, outside cable consultants are required to undergo the same ESD training and exercises as the in-house staff. These procedures have significantly contributed to the reduction of failures.

PREVIOUS • NEXT

Circuit Cellar provides up-to-date information for engineers. Visit www.circuitcellar.com for more information and additional articles.
For subscription information, call (860) 875-2199, subscribe@circuitcellar.com or subscribe online. ©Circuit Cellar, the Magazine for Computer Applications. Posted with permission.