Office of Research

Center for NMR Spectroscopy

Overview of the Console

Next to the magnet (which is often the most expensive part of an NMR spectrometer) all NMR spectrometers have a way of generating radio frequencies at or near the Larmor frequencies of the nuclei which we want to observe. Inherent with this apparatus is the need to receive the tiny NMR signals coming after the excitation pulse and a way to record those responses in such a way as to be interpretable by the spectroscopist. All of the necessary parts (with the exception of the probe or antenna) are generically lumped into what is called the "console". A generalized schematic of the console components is shown below:

A typical arrangement of such components can be seen in the stylized console cartoon found below:

We won't comment much on the "host" computer other than to say that on most modern spectrometers the host computers use the Unix operating system. Most of the console components are below the host computer and are generalized to an "Acquisition System" which is commonly referred to as the digital cardcage on most spectrometers and an "Rf" or radio frequency cardcage. Today's spectrometers as with most electronics are becoming smaller and smaller with the Mercury 300 in the NMR Center being about the size of a filing cabinet.

The "Acquisition System" or digital cardcage can be further represented by the following block diagram:

The common features in all modern NMR Spectrometers in the digital cardcage are some form of Acquisition CPU which controls the overall execution of the NMR experiment. The acquisition cpu in the Varian spectrometers are Motorola microprocessors (PowerPC or 68040) and in the Bruker spectrometers are Intel 960 microprocessors. The acquisition cpu communicates with the "host" computer via ethernet, receiving instructions and uploading data. Timing for various events which make up an NMR experiment is either handled by the acquisition cpu or by a separate Timing control microprocessor (in the Varian the timing is controlled by the Data Acquisition Board and on a Bruker by the TCU). It can be said that all NMR experiments are nothing more than accurately timed delays in which a radio frequency transmitter is either switched on or left off. The accuracy of the timing is becoming ever more critical to today's NMR experiments and this accuracy is related to the frequency of the master clock . In most modern spectrometers a 40 MHz or 80 MHz clock is used to form the basis for all timing. Generally the various boards in the digital cardcage are connected to each other via a bus (typically a VME bus).

Timing instructions are sent to the various Rf devices (and other components such as gradient amplifiers) over some form of bus. In this diagram it happens to be called the AP (analog port) bus.

The NMR signals (analog voltages) which are coming in from the probe are converted to a digital output via the analog to digital converter (ADC) board. Modern ADC boards have at least 16 bit resolution which is often increased to an effective 20 or more bit resolution via oversampling the data followed by digital signal processing to filter and subsequently downsample the data. All modern spectrometers digitize both the real and imaginary data simultaneously allowing for true quadrature detection. If the spectrometer has oversampling capability a trick to remove quadrature artifacts via frequency shifting the receiver can be done. After digitizing the data most spectrometers have a mechanism for storing data temporarily until it can be uploaded to the acquisition CPU and finally to the host workstation.

In order to detect an NMR signal a spectrometer must first be able to excite the spins. This is the job of the Rf cardcage, which is shown schematically below:

All NMR spectrometers have a way to generate the correct radio frequency to excite the target nucleus. In al cases a reference generator (a carefully thermostated crystal oscillator) provides a basis frequency which is split and fed to a frequency synthesizer. Spectrometers may differ as to the frequency synthesis scheme (as to whether the synthesizer only supplies a coarse base frequency or whether it is used to generate the smallest resolution at the synthesizer) but in all cases the frequency generated is used to feed a transmitter device where phase information for the pulse and the gating for the pulse take place. A very important frequency which is called the L.O. (local oscillator) is also generated at this point which has the observe frequency plus the intermediate frequency added. Phase information is usually added to a separate source of the intermediate frequency (IF) which when mixed down in the transmitter with the local oscillator (LO) produces the observe frequency which now carries the phase information. Amplitude scaling may also take place on the transmitter board before being sent through a series of attenuators and into the amplifier.

The amplifier is a very critical portion of modern NMR spectrometers. Most if not all modern spectrometers use amplifiers which are "linear". These amplifiers typically have outputs of 50-100W for 1H and 300 W for low band nuclei on liquids spectrometers and have outputs of 1 KW for solids spectrometers. The important fact to note about linear amplifiers is that they amplify the input signal in a linear fashion making them ideal for low power pulses and "shaped" pulses. More on the use of linear amplifiers for pulse width / transmitter power calculations.

Once the observe radio frequency pulse has been amplified it is sent to the probe via a fast PIN diode switch which routes the high power pulse (up to hundreds of Vpp) to the probe while protecting the sensitive receive path (meant for microvolts).

The active switching mechanism for a high power pulse going into the probe is shown in the following diagram:

All NMR spectrometers have a way to generate the correct radio frequency to excite the target nucleus. In al cases a reference generator (a carefully thermostated crystal oscillator) provides a basis frequency which is split and fed to a frequency synthesizer. Spectrometers may differ as to the frequency synthesis scheme (as to whether the synthesizer only supplies a coarse base frequency or whether it is used to generate the smallest resolution at the synthesizer) but in all cases the frequency generated is used to feed a transmitter device where phase information for the pulse and the gating for the pulse take place. A very important frequency which is called the L.O. (local oscillator) is also generated at this point which has the observe frequency plus the intermediate frequency added. Phase information is usually added to a separate source of the intermediate frequency (IF) which when mixed down in the transmitter with the local oscillator (LO) produces the observe frequency which now carries the phase information. Amplitude scaling may also take place on the transmitter board before being sent through a series of attenuators and into the amplifier.

The amplifier is a very critical portion of modern NMR spectrometers. Most if not all modern spectrometers use amplifiers which are "linear". These amplifiers typically have outputs of 50-100W for 1H and 300 W for low band nuclei on liquids spectrometers and have outputs of 1 KW for solids spectrometers. The important fact to note about linear amplifiers is that they amplify the input signal in a linear fashion making them ideal for low power pulses and "shaped" pulses. More on the use of linear amplifiers for pulse width / transmitter power calculations.

Once the observe radio frequency pulse has been amplified it is sent to the probe via a fast PIN diode switch which routes the high power pulse (up to hundreds of Vpp) to the probe while protecting the sensitive receive path (meant for microvolts).

The active switching mechanism for a high power pulse going into the probe is shown in the following diagram:

The converse arrangement where the switch is open for the high power path and closed for the small NMR signals coming back from the probe is shown below:

The final stop on the console tour is the receiver section where much of the action occurs. The receiver section is shown schematically below:

A slightly more detailed look at the receiver is shown in the next schematic.

To get down to the audio frequencies which are sent to the digitizer we must examine the mechanism for being able to sample and digitize two signal paths which are at right angles to each other. This is shown in the next diagram:

Here we see the observed NMR signals being routed into the preamp where they are amplified from microvolts to millivolts (typical 54 dB of gain over two to three stages) after which the signal is routed into an attenuator which is controlled by the receiver gain parameter. This attenuation step is very important when dealing with very strong signals (ie. samples in 90% H2O). At this point the signal is still being carried on the Observe frequency (the Larmor frequency) as signals of +/- delta from the carrier. The carrier is stripped off in the next stage by mixing it down with the L.O. coming from the transmitter board. The signals are now at +/- delta from the I.F. This signal then undergoes further attenuation and amplification at which point the signal is split into two paths. The identical paths are then mixed down with I.F. which has been phase shifted by 90 degrees on one of the paths to produce two separate audio signals which differ from one another by 90 degrees of phase. These mixers (often called double balanced mixers) form the heart of the quadrature detection scheme which is used to discriminate signals which rotate faster than the carrier frequency from those which rotate slower. The signals (now possibly a few kilohertz in total frequency spread) now pass through a low pass rejection filter, are amplified up to a few Vpp and passed through an audio filter which reduces noise from out of the spectral window (more on this in the Nyquist / spectrum phasing sections). If digital filters are employed the audio filters are set to their maximum width and the signals are sent to the digitizer with the signal at "0" phase being designated the "real" channel and the signal at "90" phase being designated the imaginary channel.

Notice in the gain stage prior to the audio filters there are adjustment points for DC offset (center-glitch) and channel imbalance (quadrature artifacts). The spectrometer is adjusted to minimize these affects which lead to a spike at the center of the spectrum and or small ghost peaks equally disposed about the center of the spectrum. Unless a trick called digital-quadrature-detection is used these artifacts will usually be present at about <= 0.5% of the intensity of your NMR signals on a single acquisition. To eliminate these artifacts the receiver phase is cycled by transferring the sampled data to two separate memory locations where they are summed according to an algorithm such as the following :

If these signals were just co-added they would sum to zero. If on the other hand the signals were routed after digitization as follows:

The various signal components after cycling the receiver through 4 phases and shuffling the components come from equal parts of the two pathways and hence cancel any imbalances between them.

Office of Research, PO Box 646595, Washington State University, Pullman WA 99164-6525, 509-335-5585, Contact Us