Pulse calibration & paropt.

Background.

• When I observe what Andy does, especially when there appears to be trouble, he immediately starts running paropt instead of looking at fids.
• We've experienced several cases now where the "nominal" 90 degree pulse time was far off, causing the "360" pulse time to really be close to another null.  This causes an ugly problem where you inadvertently zero on something like a 540 degree null and then divide by 4 and get a "90 degree" time that isn't even close.  In principle, you'd notice this if you always checked the "180" and "90" after obtaining the "360", but in practice the checks tend to get left out.  So instead on finding the 360 by hunting and pecking and then checking, I propose to first paropt over 360, and then around the 360 at finer scales to zoom in on the time in a way that is already validated.  The causes of far-off nominal times may have thus far have included poor tuning, typos or obsolete values in pulse program comments, consequences of probe damage, or instrumentation problems in the amplifier or preamp unit.
• When I try to set the null time more finely than ~ 1%, I have to blow up the apparently null fid deflection and try to decide when the initial deflection goes to zero.  But there are enough other deflection including some pre oscillation that the Avance machines have and following deflections from radiation damping that I can't reproducibly pick a  zero initial deflection at this scale.  This may be an exercise in futile application of an irrelevant level of accuracy, but I'm bothered by not having an objective measure of the tolerances available for setting these parameters.  In any case, I found that by first transforming the signal away from the null to save phase, and then retransforming around the null point, there was a clear and reproducible point at which the residual dispersive-like signal was symmetrically distributed above and below the horizontal axis.  This is basically how paropt presents the data, so it seems like I might as well just use paropt.

Principle of paropt

Execution of paropt

• First acquire the calibration data a little away from an expected null, transform by ef, phase, and save phase by <return><save and return>.  For precise calibration specifically of water ¹H signals, obtain the peak for phase setting as approximately 1/4 of a 90 degree pulse.  This is to avoid phase disturbances by radiation damping.  Be sure the peak does not have disturbances in the shape around the base of the peak.
• Zoom [left click then middle click to either side of peak] on the transformed peak (or peak pair for ¹⁵N or ¹³C calibration). I find about a 1 ppm interval is convenient, but the interval is not important.  If you zoomed during the phasing operation, zoom again after the <save and return> operation.
• Click on <dp1>.  [The equivalent command line command is DPL1 in lower case].  Agree with 3 pop up screens that ask confirmation of the left and right x coordinates (also known as F1, and F2 or F1P and F2P), and to confirm automatic vertical scaling of the result.  DPL1 is described in the Bruker documentation as controling the coordinates used in subsequent plotting operations.  However, through setting F1P and F2P it controls the XwinNMR display, and supplies the paropt macro with the portion of the spectrum to repetitively scan.  Paropt does not require setup or use of a printing or plotting device.
• Then type paropt, and enter the parameter to be varied, the first value, the increment, and the number of measurements in the series.  The first scan should be to vary time over 360 degrees to confirm the estimated 360 time (or over 180 in the ¹³C and ¹⁵N calibrations).  This is true even if the intent is to alter the power to match a fixed pulse time.  One can then do finer increments around the 360 varying either pulse time or power as desired.  If you do a major power adjustment, then do a final paropt run varying pulse time at the presumptave optimal power to be sure you're on the null you think you are.  Expand the vertical scale as necessary to judge the null point as the point with symmetrically distributed residual signal.You can either use the scaling buttons on the left or the equivalent command line commands to rescale the display while paropt is running.
• You'll have to mentally count the number of increments over to the target peak, so choose start values and increments that make the arithmetic easy.
• Each time paropt runs, it leaves you in procno 999.  Each time return to the experno and procno from which you are conducting the calibration using the re command (typically re 1 1). You can review the last paropt result by re 1 999.

Common problems with paropt.

• Aborting paropt.  Give the command kill at the command line.  Click on the <paropt exec> in the list.  If you try killing it by the <display> <show and allow killing> route, the pull down menus tend to disappear before you can find the right item to kill.  Killing the paropt exec should kill its various subprocesses, but occasionally I've notices a subprocess to continue.  So I recommend typing kill a second time, if there seems to be a question about the state of the system.
• No peaks displayed. The peaks will be displayed starting at an artificial coordinate of about 15 ppm (or whatever is at the left of your sweep width).  If you inadvertantly zoom on a calibration peak while process 999 is in effect, the first part of the paropt display will be out of range, and all you will see is a pulsating horizontal line.  Click on the |<>| button over the zoom tool.
• Only vertical lines displayed.  The vertical scale is too high.  Click </8> or </2> to taste while paropt is running.  You may have forgotten to to do the <dp1> operation, or it may be that the scale was set on a much smaller peak than the ones you're displaying.
• A subsequent single acquisition is wierd.  If you leave the experiment set at 1 999 and do a single determination (zg and acquire or zg and ef), the display will be uninterpretable.  Type re 1 1.

Notes:

• When I first ran paropt, there was some SI parameter that wasn't properly set for procno 999, but was correctly set in procno 1.  It was SI, which should be 64K.  I suspect I caused this somehow, but it may be a parameter that needs set the first time you use paropt with a new calib dataset.
• The maximum intensity values reported by paropt are not true 90 degree, or 270 degree pulse times.  They are markedly shifted towards the 180 by radiation damping.
• Another issue surrounding calibration that I don't understand is that the phase correction necessary to shuffle all the signal into the imaginary channel in the fid display wasn't consistent with the phase required to phase the transformed water peak.  Where is this other offset coming from?  Look at parameters offset and tdoff.  It seems to be consistently about 25 degrees and affects H and X calibrations.
• I'm adopting a rule of thumb that finding a null to within 1% of the pulse time is close enough until some data to the contrary emerges.
• I have closely compared the results of paropt with doing one point at at time.  They come out the same.  Hence there is no artifact being introduced by, say, incompete relaxation between measurements.

• The relevant Bruker manual is au.pdf.  paropt is barely mentioned, but the syntax of the au macro language is described and could be used to interpret or alter the source.
• edau paropt to see and possibly edit the macro.  I doubt if users have permission to edit, but they may be able to save a copy of a different name and under their own ownership to edit.
• Source for macro at avance600 machine in /u/exp/stan/nmr/lists/au/src/paropt
• These two web sites give an account of it:
• http://www.chem.ucla.edu/~NMR/NMR/pulses.pdf
• http://cic.chem.wisc.edu/nmr/Guides/Bavug/bruker_avance_360--natoth_3.pdf