Many military aircraft provide on board power via a +28 VDC bus whose actual voltage is permitted to deviate from its nominal value according to certain specified limits, including voltage transients. Those transients can be quite destructive of electronic equipment.
Protection of electronic equipment can often be provided by transient absorbing diodes, but such diodes cannot simply be connected across the +28 VDC bus. They require some kind of current-limiting series impedance between themselves and that line, usually a resistance of some value, R. Without current limiting, the diode itself will almost certainly be destroyed when the transients occur. See Figure 1.
Some of the most energetic and potentially damaging of the bus voltage transients are those of MIL-STD-704A, Figure 9, Curve 1 in which voltage pulses of as high as eighty volts are permitted. There exists a descriptive equation, see Figure 2, which closely approximates the Curve 1 values of voltage transient magnitude versus the time durations of those transients.
Let us call this the "first" descriptive equation.
The circles along the equations curve are data points read directly from Curve 1 and are placed in this sketch to illustrate the first equation's descriptive validity.
In Figure 3, we select the 1N6284A from a family of transient absorbing diodes.
The ability of our chosen diode to provide protection is related to its ability to dissipate pulses of power for particular durations of time. The vendor's published Peak Pulse Power Rating Curve shown in Figure 4 can be used to derive a descriptive equation for this peak pulse power capability. Let us call this the "second" descriptive equation.
The two coefficients K1 and K2 in Figure 3 turn out to be very nearly K1 = 0.45036 and K2 = 70.3823.
The second descriptive equation for the chosen diode is P = 70.3823 / T 0.45036 where P is the peak pulse power in watts and T is the duration of that pulse power in seconds, the same T as in the first descriptive equation. Hold this thought; we'll be back to it shortly.
When we substitute back as a check, we show for 0.1 µSec duration, a power capability of 99996.2 Watts and for 10 mSec duration, a power capability of 559.994 Watts.
Next, we examine the values of R which bring the peak pulse power dissipation of the transient absorbing diode to its maximum permissible value as a function of time, T.
The maximum value of R that emerges from the iterative calculation process of Figure 5 is the minimum value of R which will protect the chosen transient absorbing diode from all possible conditions of the defined input transients.
Going back to the numerical coefficients K1 and K2 for the 1N6284A, we do a numerical example and find that R >= 11.70 ohms to protect against the worst case pulse power stress. Surprisingly, that worst case stress arises for a 65.16 volt transient which lasts for 0.41 second.
Note that it is not the eighty volt transient which represents the greatest threat as we might have thought. The real diode assassin is roughly 65 volts for just over 0.4 second.
With an elevated ambient temperature of course, we must apply power derating to the diode as follows:
At an ambient temperature of 100 deg C, the derating coefficient, ß, becomes 0.5 for which the required value of R increases as follows.
This post reminded me of my old aerospace days, at Allied Signal (now Honeywell). Useful information for protecting against similar transients in other types of equipment, as well. Thanks for the reminder!
Posted by: John Jovalusky | July 15, 2011 at 01:04 PM
Hi John, I believe you did not give your Vz, the ideal clamp voltage in your description.
Note that the cheap but rather simple solution of a series resistor will only work on the smallest load draw circuits.
As soon as any significant current in drawn, the power can't be sourced through such a high series resistance. The most logical way to protect a circuit and allow it to draw significant power is to give it a wide-input power supply, if necessary a DC/DC converter that can handle at least 28-80V input without changing the 28V output. You can still add protection, but the protection should clamp at voltages higher than 80V so the time duration makes the energy miniscule and therefor the dimensioning much easier.
Note that instead of only a dropping resistor, you could also employ an impedance (inductor+resistor) to avoid dissipating all energy in a resistor and relax the dissipation.
Posted by: Cor van de Water | July 15, 2011 at 03:03 PM
Hi John,
I like your write-up. Could you offer this in .PDF format?
Thanks,
Chris
Posted by: Chris Hudgins | July 15, 2011 at 09:43 PM
Hi, Cor.
The applications to which this resistor approach had been put were indeed relatively low power. One fellow I worked with did have an application of somewhat higher power though, so having read this in an intermnal memo I'd written, he found a clamping TVS diode of somewhat greater heft and asked me to do a resistor calculation for that device. It came out to be only five ohms and he walked away happy.
When I was checking the literature about this topic, I did find a single mention of using an inductance to achieve TVS protection, but there was no theoretical underpinning provided and I didn't have time to get into that analysis. Regrettably, I never did get into that. Maybe one day soon.
Posted by: John Dunn | July 15, 2011 at 10:46 PM