The article says:

"The device he came up with consisted of an antenna and a cylinder with a thin membrane that acted as a microphone. Soviet agents stationed across the street from Spaso House would turn the device “on” by focusing a radio signal on it, which then bounced back to their radio receiver. When the ambassador or anyone else in the study spoke, the sound waves caused the membrane to resonate and alter the signal that returned to the Soviets, allowing them to hear the conversation."

The wikipedia link in a sibling comment has even more information.

There's a lot of info out there if you search at all. It acted a lot like an RFID chip it was passively powered by the receiver transmitting on given frequency. The mic was capacitive and when powered changed the signal that was being returned by the bug.

https://en.wikipedia.org/wiki/The_Thing_(listening_device)

> Soviet agents stationed across the street from Spaso House would turn the device “on” by focusing a radio signal on it, which then bounced back to their radio receiver. When the ambassador or anyone else in the study spoke, the sound waves caused the membrane to resonate and alter the signal that returned to the Soviets, allowing them to hear the conversation.

It could contain more detail but sounds like it was altering a radio signal that was completely external to the device and therefore had no battery at all.

FTA:

“It had no power pack of its own, no wires that could be discovered, no batteries to wear out...”

just to clarify the wording of your essentially correct supposition, the radio signal emanated from across the street "completely external to the device", but when this signal encountered the device, the device resonated (like an opera singer breaking a glass across a room, or even like your eardrum vibrating while listening, except EMF waves not sound) and the listeners across the street could detect the signal from this modulated resonance... so ultimately, not "completely external" to the device.

this type or degree of "vibrating resonance" is passive "signal electronics" rather than the active power electronics inside RFID, which uses a stronger signal resonance to generate power to power up an active broadcast and computing circuit.

it's not entirely unlike bouncing a laser off the outside of a window to detect the conversation inside, though that higher tech and frequencies means the system does not need to be tuned to a resonant frequency.

I think this works on the basic wave mixing principle. You mix two frequencies together and the products include the sum and the difference. My theory is this gadget was able to reflect the RF wave sent to it mixed with the sound waves from the room.

Reading the article would provide some of that information…

Michael Ossmann gave a great talk at DEFCON on RF Retro-reflectors. Provides a good technical overview of the approach they used here.

Starts at 3:32 -- https://youtu.be/mAai6dRAtFo?t=212