Design Article
Comment
TM123
GREAT-Terry
How about aiding the whole system with active clamp circuit like LTC4366? By ...
Surge protection—Stop fried electronics
Mike Fahrion, B&B Electronics
8/8/2012 1:57 PM EDT
Surges and spikes on data lines can fry your communications boards and garble data. This article describes the operation, installation and selection of what is probably the most common method of data line protection. Surge suppressors divert excess energy away from the port being protected into a ground connection. The operation of these devices relies on a high quality ground connection in order to safely shunt away unwanted energy.
Surge Suppressor Operation
Shunting harmful currents to ground before they reach the data port is the job of components such as Transient Voltage Suppressors (TVS, often referred to by the trade name Tranzorbs),
Metal Oxide Varistors (MOV) or gas discharge tubes. These devices all work by turning on at a set voltage. Once the clamp voltage has been exceeded, the devices provide a low impedance connection between terminals. These shunting devices are most often installed from each data line to the local earth ground, and should be selected to begin conducting current at a voltage as close as possible above the system's normal communications level. For RS-422 and RS-485 systems, the voltage rating selected is typically 5 – 7 volts, in RS-232 systems 12 - 15 volt devices are appropriate. These devices typically add some capacitive load to the data lines, similar to adding additional cable to the system. This should be considered when designing a system by reducing the total allowable line length. Several hundred feet is usually an adequate figure in RS-422 and RS-485 systems.
What does a surge look like?
While transients may not always conform to industry specifications, both the Institute of Electrical and Electronics Engineers (IEEE) and the International Electrotechnical Commission (IEC) have developed transient models for use in evaluating electrical and electronic equipment for immunity to surges. These models can offer some insight into the types of energy that must be controlled to prevent system damage.
Both IEC 1000-4-5: 1995 “Surge Immunity Test” and IEEE C62.41-1991 “IEEE Recommended Practice on Surge Voltages in Low-Voltage AC Power Circuits” define a “1.2/50µs - 8/20µs combination wave” surge which has a 1.2 µs voltage rise time with a 50 µs decay across an open circuit. The specified current waveform has an 8 µs rise time with 20 µs decay into a short circuit. Open circuit voltage levels from 1 to 6 kV are commonly used in both the positive and negative polarities, although under some circumstances voltages as high as 20 kV may be applied. Figures 1 and 2 illustrate the combination wave characteristics. In addition, IEEE C62.41 also specifies a 100 kHz “ring wave” test. The ring wave has a 0.5 µs rise time and a decaying oscillation at 100 kHz with source impedance of 12W as shown in Figure 3. Typical amplitudes for the 100 kHz ring wave also range from 1 – 6 kV.
Installation
Surge suppressors must be installed as close to the port to be protected as possible, and must have an extremely low impedance connection to the local earth ground of the unit being protected.
|
|
|
|
|
jonnydoin
8/9/2012 5:23 AM EDT
The article mentions 8/20 surge pulses as 8ms/20ms. That's wrong. The time scale is in microseconds. A pulse of 8kV with 20ms decay is a *very* large energy. TVSes are capable of absorbing power up to 400~5000W, depending on the model, but a 20ms pulse would require a much larger device.
- Jonny
Sign in to Reply
Mike Fahrion
8/9/2012 12:05 PM EDT
Great catch Jonny - the mu symbol in the text got lost in the web translation. We'll get that corrected.
Sign in to Reply
cmathas
8/9/2012 12:48 PM EDT
Sorry, I didn't catch the change in the translation. It is fixed now. Thanks for letting us know.
Sign in to Reply
David Ashton
8/12/2012 8:10 PM EDT
While you're fixing stuff Carolyn, on page 2 2nd para is: "6 feet of 18AWG wire has approximately .039 W resistance." W should of course be the ohms symbol - which is Capital W in Symbol Font.
Sign in to Reply
TonyDave
8/10/2012 4:24 AM EDT
Hi,
Surge? "An uninterrupted voltage increase that lasts more than a few seconds is usually called a "voltage surge" rather than a spike."
Nice article about Transient protection, what about the surges. Try 150% for 1/2S (with a profile). Now thats a surge.
Stokersson
Sign in to Reply
EREBUS
8/11/2012 4:59 PM EDT
Personnally, I like my circuits fried with a touch of tobasco sauce.
Seriously, you need to put protection on all of your inputs or pay the consequences. Regardless of surge or spike, over voltage destroys semiconductors in femtoseconds.
Anyone remember the old UV erasable proms? I could destroy one from five feet away just from the static charge I would build up walking over the tile floor.
Zaap! It's done.
Sign in to Reply
I_B_GREEN
8/12/2012 5:45 PM EDT
Only thing missing is an inductor in place of or in series with the resistor. Allows for much more energy withstand. fulcrums energy from (referenced in voltage)shunt protection downstream to GDT via inductor voltage rise due to di/dt. Care must be taken to not saturate core, air gap recommended.
Sign in to Reply
TM123
8/12/2012 6:18 PM EDT
I have found that allowing the entire voltage supply (including 0 Volts) to float enables the circuit to float around any surge/spike.
In this case, there are a number of capacitors connecting various voltages, inputs and outputs to the real ground.
A device I use lots of is the good old Pi Filter (or capacitor input filter) on every power supply rail, input and output, along with extra series inductors - everywhere.
To date, I have protected uProcessors, and low voltage electronics (5 Volt, 12 Volt) successfully against voltage discharges around the 8 MV mark (we have electronics inside an Ion Accelerator).
Warning - do not use MOVs as when they fail, they fail short which will require replacement. Use protective devices that are self recovering so the protective components do not become part of the problem.
Sign in to Reply
David Ashton
8/12/2012 8:06 PM EDT
Point taken about MOVs, but they are capable of absorbing a LOT of energy, even if they fail shorted. I used to live in Zimbabwe, which has bad lightning during the summer. A company I dealt with there used big MOVs in their mains surge arrestors and they worked a treat. Better to replace a $ 35 mains surge protector than a TV/Video installation.
Same company did some tests and found that a knot in the 3-core mains supply cable provided a lot of protection as well - added inductance. I used to knot all my cables after that and never had a problem with my mains inputs.
Sign in to Reply
GREAT-Terry
8/14/2012 4:54 AM EDT
Good. Is there any difference between putting 2 unidirectional TVS and a single bidirectional TVS? I usually use 2 uniderectional but frankly have no idea if a single bidirectional has poorer. Agree also to add air core inductor in series with the resistor,
Sign in to Reply
TM123
8/15/2012 6:15 PM EDT
GREAT-Terry, A Bidirectional TVS is matched to provide performance in both directions - equally. A pair of Unidirectional TVS will provide protection but will not necessarily be matched to balance the intrusion levels either side of your supply or common rails.
David Ashton, fair comment on the MOV front - our particular application is inside a pressure vessel and it takes a bit over a day to pump out the gas (SF6), purge and put in walkways to access the electronics. MOVs do work but are not appropriate in our situation.
Sign in to Reply
GREAT-Terry
8/15/2012 11:15 PM EDT
How about aiding the whole system with active clamp circuit like LTC4366? By using this kind of clamping device, the TVS can be smaller I think.
Sign in to Reply
TM123
8/21/2012 7:59 PM EDT
There is a bit more to it than just shunting off surges/spikes.
The article has addressed the Ohmic requirements, but what about capacitive and inductive effects.
Depending on cable lengths, there will be some considerable capacitive coupling between conductors for the duration of the surge/spike.
Also multiple conductors (presumably in the same cable/conduit/duct) will suffer from inductive coupling similar to a transformer - after all we are dealing with a changing Voltage/Current waveform i.e. ac - but only for restricted cycles (hopefully only one).
Some care also needs to be taken to prevent these effect through the use of shielded cables and earthed at both ends, to transform a simple shielded cable into a faraday cage in an effort to reduce these other effects.
This raises a can of worms because we now have the probability of earth loops. To protect we must make some compromises.
I know I would rather deal with noise reduction rather than replace everything because it has been fried!
Sign in to Reply