unexpected delay in a TWT

The point of a TWT over regular tubes is changing the paradigm on how electrons do work, so that should be due to propagation of the electron beam itself.

As I recall, TWTs work by shooting an electron beam (= constant velocity) down a helix, where, due to spooky action and black magic, the beam turns into bunches and somehow does work on the electromagnetic field. The result, gain. The delay should be essentially the propagation of the beam. Maybe with a velocity factor for the helix, or somewhat shorter for the feedpoints being somewhere along it, etc.

Whatever the case, I'm guessing your anode voltage is close to 160V.

Tim

Our TWT has an anode voltage of -14kV and a current in the range of 60mA. And yes, the signal propagation of the electron beam matches the signal velocity on the helix. I was puzzled that there are 1400 wavelengths stored in the helix, while I expected much less.

Rene

Tim Williams wrote:

Wow. The beam velocity is about c*sqrt(2*16keV/511keV) or 7.5e7 m/s, so for a 30 cm tube, the transit time should be just about exactly 4 ns. Are you sure you didn't drop a decimal place somewhere?

Cheers

Phil Hobbs

Seems a lot. How did you measure the delay?

John

beam

ity)

ns

beam.

charge

Does the device to which you refer as a 'TWT' actually contain a TWT and some filters, at least to constrain the amplitudes of harmonics when the device is operated anywhere near to saturation? If so, the filters will contribute to the delay.

Chris

Thanks John and Phil,

I have a CW generator and a fast switch that switches within nanoseconds that is connected to the TWT unit, not the tube. The unit contains whatever plus the tube. I have no idea, whether a filter is included or not. Even if there was a filter, it cannot contribute to the delay as it has as bandpass a bandwidth of perhaps 2GHz.

After the unit, there is an isolator plus a directional coupler and another directional coupler. Then a detection diode. I measure the delay between the control pulse of the switch and the detection diode with a sufficiently fast scope doing a few GSample. The pulse shapes there are sufficiently rectangular.

Rene

Hmm. A sharp enough filter with a 2 GHz bandwidth could contribute a couple of nanoseconds, but 40 is a stretch. Is it possible to tune the source just a bit and measure d(phi)/d(f)?

Cheers

Phil Hobbs

On a sunny day (Sun, 18 Apr 2010 17:13:48 +0200) it happened Rene Tschaggelar wrote in :

Did you take into account that the speed of the electrons in that tube is

Check my post from the 13th. If the total delay were due to the electron beam, it would be powered by about 160V.

Tim

On a sunny day (Sun, 18 Apr 2010 19:19:14 -0500) it happened "Tim Williams" wrote in :

Circular motion? Been many years since I had to do with a TWT amplifier. That was for a satellite uplink.

On a sunny day (Sun, 18 Apr 2010 19:19:14 -0500) it happened "Tim Williams" wrote in :

Circular motion? Been many years since I had to do with a TWT amplifier. That was for a satellite uplink. PS because when you uplink to a sat at 40,000 km height, then 40 ns delay is nothing compared to what you get. Big echo when receiving back your own signal :-) And with digital processing the latency is even bigger.