This is what we do at VK4BA, A 12V battery, charged by solar, then a simple 5V supply for the transmitter. The antenna is a vertical, grounded right there to a ground rod and radial field. The battery negative is grounded as well (If I remember correctly). The GDT is mounted as described above. We have been trough a few good (lightning) storms this season without fault. However, nothing is going to save you from a direct hit most likely, but I think the GDT helps with static charge dissipation and induced currents from nearby strikes. But of course, maybe we've been lucky :)

I wonder if we are now seeing problems with the power supplies being reset/current limit due to some coupled spikes (common mode spikes maybe). I would try to isolate the power supply from the Transmitter for more GDT testing. Testing the transmitter using batteries and no connections to a PC/Pi etc. Would 3 * AA cells @ 4.5v work to power it or would you need a simple linear regulator (7805) and a 6v battery? I am talking about a simple power supply, not a USB power bank as that may be affected as well. Worth trying both I guess. Having the USB power "safety tester" that will current limit and reset if the unit locks up is a good fall back. A note on most USB ports - they are supposed to have hardware over current protection in the chips that will shut off the power if the device draws too much current. I think the Pi is detecting this happening and reporting it. This is a good indicator that the WSPR Transmitter is latching up somewhere.

That is interesting. I would not parallel GDTs - just seems like it could cause trouble when you consider there are several interacting modes of operation and wide tolerances.. But I have no real knowledge behind that. How are you zapping the board to test? could there be a power supply spike induced on the power lead as well? Can you see the GDT spark over? Normally you can see a flash thorough the ceramic shell. Are you testing the unit inside the metal case or bare boards? I wonder if there are other coupling mechanisms that the case helps to prevent?

@Harry Zachrisson Like I said, its more complex but I think about them that way for a basic mental model. If you need to actually design proper protection systems it is a lot more complex and a GDT is just one stage. The 90V rating is a nominal DC arc over voltage (it is normally a wide tolerance +/- 20%). GDT should NOT be used at DC where the voltage will get anywhere near that level. If it fires it will stay on and basically self-destruct. You are right about the higher trigger voltage for transients. For instance the datasheet I'm looking at for a TDK (Epcos) unit lists spark over voltage at about 500V for a 100V/us pulse. However once transitioned into conduction, the voltage across the unit is held quite low. They specify 15V for an arc current of 1A, higher currents would have a higher voltage. This low arc voltage is why care has to be taken in the presence of DC. Transient pulses are a lot harder to reason about and simulation is probably the best option without access to specialised testing equipment. The low pass filters on the output of the transmitter will help suppress any pulse getting into the actual transmitter until the GDT kicks in. But simulating it with some of the standard pulse models (8/20us for instance), would be an interesting exercise. Parasitic inductance and capacitance is important to model for this, and is also why testing with high power pulse generators is important for verification. For use as an antenna transient suppressor, basically just ensure the RF power output can't trigger the GDT by a wide margin. High power RF becomes a lot more tricky, but for lower power things this is fine. I've seen these used a lot in cellular phone systems when I worked for a company that built tower mounted amplifiers and filters. We did significant testing with lab ESD pulse generators to verify our protection circuits. However in these applications, we were building bias-Ts into the device, the GDT was placed after the RF choke so would operate at DC only. It was there to protect against induced voltage between the inner and outer conductor of the coax. We also had specially designed power supplies powering the devices which would detect current spikes like an arc and totally disconnect the power for a period before re-powering to clear any arc faults - a bit like re-closers on the electrical grid. They would retry a small number of times before locking out that power feed up the tower and raising an alarm. 73s, Ash. edited: typos

The voltage rating is their nominal "strike" voltage. Below that voltage they are essentially open circuit, once the voltage exceeds the rating (more complex, but close enough) they flash over and become practically a short - a bit like a neon bulb. You would want to use the 90V unit in this case, it will fire at 90V not 230V. If this were a high power transmitter, you may need the higher voltage one to account for the voltage going into the coax from the RF.. I would guess you could also use an actual neon bulb, but I expect it wouldn't be as robust.

Actually sorry.. don't put the TVS diode over the power supply - that would mean any fault currents have to go via IC pin protection diodes and that would be bad.. Looking at the schematic, perhaps the right place to put it would be after the PA and before the filters.. Maybe TP5? I used TP6 for the GDT. That way the GDT will kick in and limit to spikes, then the filters will provide further attenuation of spikes, and then the TVS diode would limit any excursions before they hit the PA. 73's, Ash.

Harry, it was a pretty simple thing to add on our end (I had some of the GDT's on hand) and there was a nicely placed test point on the antenna input so soldering it in was quick and easy. Used the SMA ground pad on once side for the ground, the TP as the antenna input. Thought it would be good cheap insurance given the electrical storms we have here. If you wanted to add a second protection stage like a TVI diode, I think it would need some isolation from the GDT so it doesn't prevent the GDT firing. They are designed for significant current handling, but have a higher strike voltage.. Perhaps the TVI could be place across the power rails, but not the antenna input? The other thing that is important is grounding the antenna coax shield at the connector so high current paths would bypass the transmitter and not go through the 5v power supply. Then the GDT can take care of any induced shield to centre pin voltages. We have also thought about adding a high value resistor to bleed static charge from the antenna to ground, or use an isolating transformer on the antenna to ensure there is a DC path to ground. Just some thoughts.. hope it helps! Ash.