Keith Armstrong of Cherry Clough Consultants, and an editorial board member of Interference Technology, spoke with Clay McCreary about his paper, “Design of Multiple Stage Avionics Lightning Protection for DC Power Input Lines Using a GUI” at this year’s IEEE International Symposium on EMC.
McCreary’s paper provides a major advance in understanding how to use Gas Discharge Tubes (GDTs) to more effectively protect electronics from lightning surges on DC power supplies.
He found that GDTs with a specified DC holdover voltage of 52V, used on 28Vdc aircraft power supplies, could easily continue to conduct after the surge had gone, shorting the power supply out and setting the PCB on fire (or at least badly charring it).
According to Armstrong, everyone assumes that if a GDT’s spec says the DC holdover voltage is 52V, then it would not continue to conduct after the surge, providing the DC power voltage was less than, say, 48V (to be on the safe side)— but this turns out not to be true, and the resulting overheating can cause serious fire hazards.
McCreary discovered that the test spec used to measure the DC holdover voltages of GDTs uses a 200 Ohm resistive source impedance, and that lower source resistances (such as an aircraft’s 28Vdc power bus) produce higher surge currents that reduce the DC holdover voltage!
The lower the power supply’s source resistance, the higher the current in the GDT’s plasma, and the lower the DC holdover voltage—which explained how his “52V” specified GDTs could burn up PCBs when suppressing a 28Vdc supply that had a very low source resistance. This is a very important issue when using GDTs – one that many engineers know little about.
McCreary discussed the GDT ‘switch-on’ voltage—the voltage that has to be achieved to start the plasma and the surge clamping action. He found that a given GDT had a very variable switch-on voltage, varying by as much as 200V depending on the application, and his colleagues just told him that that was what GDTs were like. But it is difficult to protect electronics from surge damage if the transient voltage that is ‘let-through’ by the GDT varies by that much, Armstrong said.
Investigating further, once again McCreary found that the switch-on voltage was specified by the manufacturer after testing with a specified source resistance—in this case 1000 Ohms— and that different source resistances give different switch-on voltages.
After discovering the test condition used to rate switch-on voltage, it was determined that the GDTs being used were not designed for high current applications resulting in the large variation of switch-on voltage. Thus, a GDT designed for high current applications was chosen and the variation in switch-on voltage was reduced to less than 20V — a ten-fold improvement.
As for the issue of switch-on voltage variations, those that knew didn’t tell anyone else so that they could design more effective lightning protection.
Clay’s poster paper is co-authored with Brian Lail. He has an article on these GDT issues coming out in a future edition of the IEEE EMC Society Magazine.
