Here, the sample is subjected to continuous irradiation about the x axis. While being irradiated, the magnetization vector rotates in the z-y plane at a nutation frequency proportional to the pulse power. The magnetization on the -y axis is defined by a sine function. Fourier transformation of this magnetization gives an antiphase doublet centered at zero whose splitting Δν is twice the nutation frequency. The reciprocal of the nutation frequency is the time it takes the magnetization vector to rotate one complete cycle in the z-y plane and therefore the time it takes to rotate by one quarter of a cycle (i.e. the 90° pulse duration) is defined as 1/(2 Δν). The problem with continuous irradiation is that the sample must be irradiated at the same time magnetization is being detected. To eliminate this problem, a scheme similar to homonuclear decoupling is used where the radiation is turned off long enough to sample a data point. This is depicted in the figure below.
Here each dwell period is divided up into a period for irradiation and a period for detection. The duty cycle for the irradiation is the fraction of time for which the sample is being irradiated. The magnetization is sampled when the power is off. As in the case for continuous irradiation, the magnetization vector still rotates in the z-y plane however, the rotation is slower as it is scaled according to the duty cycle. The duration of the 90° pulse is d/(2 Δν), where d is the duty cycle for irradiation. An example of this is shown in the figure below. 
The nutation spectrum was measured for HDO using a duty cycle, d = 0.10 and a power level of 12 dB (Bruker). Since the response of the amplifiers is linear, the 90° pulses at higher power levels can be calculated. Each decrease by 6 dB cuts the duration of the 90° pulse in half. In this case the 90° pulse at 0 dB was calculated to be 10.93 µsec at 0 dB based on the measured 90° pulse of 43.71 µsec at 12 dB. This pulse agrees to within a couple of percent of that measured by the more traditional method however, the measurement took only a few seconds. You can use a program called "pulsecal" on newer Bruker spectrometers to do this in complete automation.--
* Peter S.C. Wu and Gottfried Otting J. Mag. Res. 176, 115 (2005).
This sequence uses a simple two step phase cycle to subtract out the effect of the direct 13C Boltzmann magnetization. The first part of the sequence uses a (90°-y) pulse to return the CP enhanced magnetization to the z axis. The decay of the enhanced magnetization down to its Boltzmann value is followed using a (90°x) pulse with detection of signals on the -y axis. The second part of the sequence uses a (90°y) pulse to put the CP enhanced magnetization on the -z axis. The recovery of the enhanced inverted magnetization back to its equilibrium Boltzmann value is followed using a (90°x) pulse with detection of signals on the y axis. The addition of the first and second parts of the experiment by way of the phase cycle allows for a simple calculation of the 13C T1 with both the advantages of CP enhancement and the ability to collect more scans per unit time. An illustration of this method is shown in the figure below where the 13C T1's of glycine were measured. (The small peak in the spectrum is a spinning sideband of the carbonyl carbon)
where νsample is the absolute frequency of the sample resonance and νreference is the absolute frequency of an agreed upon reference compound. For 1H, 13C and 29Si NMR, tetramethylsilane (TMS) is the agreed upon reference standard. When the scale is plotted in this manner, the peak positions are relative to that of the standard compound. This scale is particularly useful, as it independent of a single absolute frequency and therefore does not depend on the magnetic field strength, which varies from laboratory to laboratory. Spectra recorded using magnets of unequal field strength can be compared more easily.
The shielding scale increases to the right.
The second figure demonstrates this experimentally. It shows the 500 MHz 1H NMR spectrum of HDO using the pulse sequence shown in the figure as a function of the % gradient strength (100% ~ 0.5 T/m). The gradient pulses were 1 msec in duration and rectangular in shape. One can see that the intensity profile matches closely to that predicted in the bottom of the first figure. The sample is almost entirely dephased using only 2% of the maximum gradient strength.
I regret to say that 
The transmitter frequency is set on the water resonance. A non-selective hard 90° pulse is applied followed by a 1 -2 msec gradient pulse. The gradient pulse dephases all of the resonances. A composite pulse (consisting of 6 hard pulses seperated by a delay, τ) is then applied which acts as a 180° pulse for everything except peaks on resonance (i.e. water) and any peaks at frequencies n/τ away from the transmitter, where n is an integer. τ is chosen such that 1/τ lies outside of the spectral width (typically several hundred µsec). The second gradient pulse (equal in magnitude, duration and sign, to the first) further dephases the water resonance at the center of the spectrum which was unaffected by the composite pulse but rephases everything else which was inverted by the composite pulse. The gradients and composite pulse act as a gradient spin echo for all but the water. The FID is then collected with the water resonance suppressed by the two dephasing gradients. An example of the application of WATERGATE is shown in the figure below. 
The top trace shows a standard 500 MHz 1H NMR spectrum of phenylalanine in H2O / D2O scaled to the water peak. The resonances of the phenylalanine are not visible on this scale. The middle trace is the same spectrum as the top trace with the phenylalanine resonances on scale. The huge water resonance is truncated. The bottom trace shows the WATERGATE spectrum. The water signal is greatly suppressed.
During the (D2 - π -D2)n period the intensity of lines with short T2 (broad lines) diminishes much more quickly than that for lines with long T2 (sharp lines). The CPMG sequence is therefore useful for enhancing the sharp features in a spectrum by suppressing the broad features. This is demonstrated in the figure below. The top panel of the figure shows a portion of a conventional 500 MHz 1H NMR spectrum of a polymer sample contaminated with small amounts of smaller molecules. The broad lines (truncated in the figure) are due to the polymer whereas the much smaller sharp lines are due to the impurities. The bottom panel of the figure shows the CPMG spectrum of the same sample with D2 = 4 msec and n = 32. One can see that the broad polymer lines are greatly suppressed and the smaller sharp lines are much more obvious. 
A sample of CHCl3 in acetone-d6 was placed in a 500 MHz magnet equipped with a probe using air as the VT gas. The magnet was shimmed and the spectrum acquired is shown in the top trace. The air source was then replaced by a source of nitrogen gas at the same flow rate. The spectrum was measured again without re-shimming the magnet and is displayed in the lower trace. The difference in line shape and width is due to the difference in magnetic susceptibilities between the two gases. It should be noted that a spectrum of similar quality to the one obtained using air can be obtained after re-shimming the magnet to correct for the susceptibility difference.
The "T" connector and 50 Ω terminator are not required if the oscilloscope has in input impedance setting of 50 Ω. In this case, the connections can be made according to the following figure.
The output power (in Watts) is determined by the peak to peak voltage, Vpp , of the pulse as follows.
The top trace shows the data for a rectangular gradient at 100 % strength. The middle trace shows the results for a rectangular gradient of 50% of full strength and the bottom trace shows the data using a sine bell shaped gradient pulse of 100 % strength. One can see that for the rectangular gradients the full intensity of the line is recovered in as little as 10 µsec. When the sine bell shaped gradient pulse is used, the full intensity of the line is recovered in less than 1 microsecond. This faster recovery is the result of the gradual rise and fall of the gradient strength in the sine bell shaped pulse.
The top trace shows the data for a rectangular gradient at 100 % strength. The middle trace shows the results for a rectangular gradient of 50% of full strength and the bottom trace shows the data using a sine shaped gradient pulse of 100 % strength. One can see that in all cases a reasonable line shape is recovered in ~ 400 msec. The shape of the gradient pulse does not seem to influence the time required to recover a good line shape.
The data were obtained by measuring a 300 MHz 1H NMR spectrum of HDO as a function of transmitter frequency. The 
For this measurement, the temperature of the probe was set to 50°C with an air flow of 800 L/hour. Once the thermocouple read 50°C, a sample of D2O was placed in the probe and 30 minutes was allowed to pass, after which the sample was presumed to be at thermal equilibrium. The lock was established and the magnet was then shimmed. The sample was removed and allowed to sit at room temperature for 30 minutes. It was then reintroduced to the probe at 50°C. 1H NMR spectra of the residual HDO were then collected at 30 second intervals for a period of 10 minutes. As soon as the room temperature sample is reintroduced to the warm probe, it begins to warm up. During the this time, the thermal gradients and convection currents are large and the line width is adversely affected. As the sample temperature approaches 50°C the thermal gradients are smaller and the line becomes narrower. After approximately 6 minutes the width of the line changes very little. The sample appears to be at thermal equilibrium after 10 minutes.
Imagine a spectrum consisting of two singlets. If the transmitter is set to the frequency of one of the singlets and a 90°x pulse is applied, both magnetization vectors are rotated to the -y axis. During a delay equal to one quarter of the reciprocal frequency difference between the singlets, the "on resonance" singlet will remain stationary while the "off resonance" singlet will rotate by 90° onto the x axis. If a long high power pulse is now applied along the y axis, it will behave as a spin locking pulse for the "on resonance" singlet and a purge pulse for the "off resonance" singlet. An example of this is shown in the figure below for a sample of methylene chloride and chloroform where the transmitter was set on the methylene chloride resonance.
The top trace represents a simple one pulse measurement. The spectrum in the bottom trace was collected by applying a 90°x pulse followed by a delay equal to one quarter of the reciprocal frequency difference between the methylene chloride and chloroform. A 1 msec y pulse was then applied at the same 
The second figure is the 19F MAS spectrum of the perfluorinated polymer, Nafion at 11.7 T (top trace) and 21.1 T (bottom trace). The MAS rates for each spectrum were chosen such that the dipolar coupling between the fluorines was effectively averaged by the MAS and that the spinning sidebands would be coincident (in ppm) in the spectra. Again, one can see that there is little if any improvement in the chemical shift resolution at higher field.
The answer is the very complicated spectrum B. The spectra were calculated with the following parameters:
At the same time the miscalibrated subsequent pulses lead to significantly distorted spikelet patterns (second figure). 
The 180o pulse misset by as little as 20o-30o, could produce considerable oscillations in the spikelet intensity across the envelope. This illustrates that 