Droop, pulses

A dc restoration circuit is needed following the output coupling capacitor to make the drive voltage referenced to the power switch s common. The supply voltage of the driver should be well bypassed so that its voltage does not droop during the drive pulse. 

Hviid T, Nielsen SO. (1972) 35 volt, 180 ampere pulse generator with droop control for pulsing xenon arcs. Rev Sci Instrum 43 11928-1199. 

Dependence of Residual Linewidth on Errors of Individual Pulses, Power Droop, and Offset... 

A power droop of the transmitter leading to a slow variation (decrease) of the flip angles during the pulse train... 

We now turn to the consequences of a power droop. When a m.p. sequence is started, the transmitter must suddenly switch from the off state, where the grid, screen, and plate currents are virtually zero, to the on state, where it must fire thousands of closely spaced pulses for some tens of milliseconds. Usually the transmitter is a class C tube amplifier, which means that especially the plate current cannot be drawn for this length of time from buffer capacitors placed close to the power tubes. In response to the sudden change of the plate current, the plate voltage will sag to some extent and this causes a droop of the rf power and hence of the flip angle /3. This droop affects the m.p. spectrum in two ways ... 

The scaling factor of a m.p. sequence depends on p (Haeberlen, 1976). A variation of p during the sequence therefore causes a chirp of each resonance. Our simulation program allows us to quantify this effect also. In Fig. 10 we show a simulated BR-24 spectrum of our model system that assumes an exponential power droop that amounts to no more than a 1% decrease of p after 100 BR-24 cycles, that is, after 2400 pulses. Note the asymmetry of the lines and the wiggles at their feet that are indicative of the chirp. In Section IV we present experimental m.p. spectra that display exactly these features. 

Fig. 10. Effect of a droop of the transmitter power on m.p. spectra. An exponential decrease of all flip angles, which amounts to no more than 1% after 100 BR-24 cycles, that is, 2400 pulses, is stipulated. Note the wiggles at the feet of the lines.

A close inspection of the best resolved MREV spectra (b) and (c) in Fig. 22 reveals small wiggles at the feet of the lines. A comparison with Fig. 10 suggests that these wiggles result from a (small) droop of the rf power along the pulse train. Such wiggles are less pronounced in the BR-24 spectra, which comes as no surprise because the duty cycle was only 1 /6 in the BR-24 but 1 /3 in the MREV experiments. 

In addition to the effects from the probe there is the electronic deadtime, including pulse ringdown (100 ns), preamplifier recovery (800 ns), filter overdrive recovery (1 p.s) and ADC conversion droop (200 ns) (Hoult 1979). Magnetoacoustic ringing (up to 200 p.s) can be very significant if careful probe design (e.g. coil wire) is not considered. Samples that exhibit peizoelectric behaviour can lead to very long response times of up to 10 ms. 

A constant-current or drooping-characteristic power supply is required for GTA KJ either DC or AC and with or without pulsing capabilities. For water-cooled torches, a water cooler circulator is preferred over the use of tap... 

---