1. INTRODUCTION
Nuclear magnetic resonance (NMR) spectroscopy is a powerful tool for elucidating the molecular chemical structure of organic compounds, with a wide application in biotechnology, medicine, the pharmaceutical medical industry, the food industry, among other branches of science and industry [1-6]. Any NMR experiment consists of placing the spin system under the action of an external static magnetic field B0, capable of aligning spins in its direction, determining the appearance of a resulting non-zero magnetization. The sense of orientation can occur in two possible states: one of low energy (parallel) and the other of high energy (anti-parallel).
The application of the RF pulse results in a net absorption, and the energy difference between the spin states parallel to B0 (which determines absorption intensity) depends on B0 intensity and the gyromagnetic ratio of the nucleus [1].
The signal-to-noise ratio (SNR), signal resolution and NMR spectra are parameters that vary proportionally with the intensity of the B0 magnetic field [7-12]. The latter facts are among the reasons why modern spectrometers have increased their magnetic field from very low fields up to very high fields, reaching even 20 T [11-20]. In order to conduct studies of the molecular chemical structure of organic compounds of interest to the Biophysics and Medical Physics Center (CBM) and to compare the results with MRI imaging studies conducted on 1.5 T machines, the need arises for developing an NMR spectrometer at this B0 intensity (64 MHz); considering that commercial equipment using this technology is extremely expensive due to the use of ultra-high magnetic fields and highly specialized components. Taking from our laboratory different NMR subsystems which worked at 4 MHz and which could be electronically adjusted to 64 MHz, our research team built a functional MR spectrometer, without major technological complications, see Fig. 1.
In Fig. 1, the Control System consists of a personal computer (PC) with a Pulse Generator Card (PGC) that generates the desired types of excitation pulses and the Acquisition Card (AC). The RF Excitation System includes the PGC, the modulator and the RF power amplifier, and the Reception System consists of a RF preamplifier, the Receiver and the AC. The Magnetic System comprises the current source, the electromagnet and the probehead.
This paper aims at presenting the steps taken to upgrade the RF modulator, which is a key element in the proper functioning of the RF system. The modulator was originally designed to operate at a working frequency of 4 MHz; therefore, it was essential to take several steps to make it function correctly at 64 MHz. The correct performance of the modulator is demonstrated by measuring its parameters in each stage, as well as through the MR signal acquired in a sample of choline dissolved in pure H2O.
The development of the 64 MHz spectrometer, as well as the upgrading of its RF modulator, changed the goal of the experimental station, from the original NRM images to NMR spectroscopy, and their applications. In addition, it became a teaching basis for learning in NMR technology, since our institution is located in the Universidad de Oriente (University of Oriente).
To present the changes made into the modulator, the document is organized as follows: Section 2, Materials and methods, describes the steps taken to upgrade the modulator and all the aspects associated with each stage; Section 3, Results and discussion, presents the significant results obtained from the measurement of the modulator’s updated parameters, as well as the discussion of results to assess the quality and good performance of the modulator. Conclusions and ideas for future work are presented in Section 4.
2.- MATERIALS AND METHODS
To achieve a field of 1.5 T in the spectrometer of Fig. 1 a high stability current source was set up with 54 A approximately, feeding a 0.58 Ω electromagnet. The spectrometer is controlled by the PC, the PGC that indicates the types of pulses desired and the AC that processes the signal coming from the receiver channel. The RF Excitation Systems consists of the PCI-766 pulse generator card, the modulator that modulates the signal coming from the synthesizers, and the RF Power Amplifier AN8063. On the other hand, the Reception channel is composed of the RF Preamplifier (RFA) AU-2A, the receiver that translates the signal from 64 MHz to baseband and then filters it, and finally the PCI-730 acquisition card. The latter card is also used to digitally control the modulator and RF power amplifier. In both channels, the synthesizer NOVATEC DDS7 (frequency output of 10 MHz) and PM5192 (frequency output of 74 MHz) are used.
2.1.- THE MODULATOR
To upgrade the modulator, the following actions were performed:
Changing the frequencies, the modulator needs to work properly at 64 MHz.
Checking that all components inside the modulator can operate well at the desired frequencies.
Checking the performance of the RF switch to attenuate the output signal when required (and design one if necessary).
Building and changing the filter at the output of the modulator.
Validating the modulator performance.
To carry out the above steps, the following software and tools were employed: the design of all schematic circuits and their simulations were implemented using the KiCad software tool, and the graphs shown in this paper were made on the OCTAVE software. The characterization of the filter was made employing the Agilent N9923A network analyser. The acquisition and measurement of the modulator data was made using the TEKTRONIX TBS1154 digital oscilloscope.
2.1.1- SETTING UP THE FREQUENCIES OF THE MODULATOR
The first task in upgrading the modulator was to change the frequencies it works with. These frequencies will be called the working frequency, the intermediate frequency and the synthesizer frequency. The upgraded modulator along with all its signals and components are shown in Fig. 2.
The working frequency (signal at RF OUTPUT in Fig. 2) is the frequency needed to perturb the net magnetization in an NMR experiment and once the main magnetic field is fixed, it can be calculated using the Larmor equation (1) [9].
where γ is the gyromagnetic constant (42,57 MHz/T for 1H), B0 is the magnetic field intensity of the electromagnet (T) and ω0 is precession frequency (Hz). For this work, B0 must be substituted by 1.5 T, and equation (1) yields a working frequency of 64 MHz; therefore, this is the frequency that must be achieved at the output of the modulator to properly excite the net magnetization at 1.5 T.
The intermediate frequency (signal at RF 1 INPUT in Fig. 2) allows working in a lower frequency and helps to obtain, through mixing, the working frequency. This frequency was set up to 10 MHz in the old modulator and is kept at this value in this upgraded version.
The synthesizer frequency (signal at RF 2 INPUT in Fig. 2) is the other stable high frequency signal that allows to obtain, through mixing, the working frequency. In order to obtain the 64 MHz working frequency mentioned above, the synthesizer frequency is fixed at 74 MHz. This 74 MHz signal mixed with 10 MHz yields, among other results, the difference frequency (74 MHz - 10 MHz = 64 MHz), which is the working frequency desired.
The baseband signal is the envelope of the RF pulse necessary to excite the probe. It is generated in the PC (Fig.1) and enters the modulator through BASEBAND SIGNAL INPUT port in Fig. 2. It may be gaussian, squared or sinc (sin(x)/x).
The general operation of the modulator (Fig. 2) is described as follows: The 10 MHz intermediate signal at RF 1 INPUT port enters to PHASE SHIFTER module and is shifted by 0, 90, 180 and 270 degrees as selected from signals entering through P0 and P1 ports. These last two signals (those that enter P0 and P1 ports) are digital and come out from the PC (Fig. 1). The 10 MHz shifted signal is then mixed with the signal at BASEBAND SIGNAL INPUT port (it may be gaussian, square, or sinc) that comes from the PCI-766 pulse generator card (PGC from Fig. 1). This resulting signal is mixed again with the 74 MHz signal at RF 2 INPUT port and finally amplified and filtered by the PASSBAND FILTER module in order to obtain the modulated 64 MHz signal. The 74 MHz signal is attenuated by the RF SWITCH when there is no baseband signal present, thus reducing noise and unwanted effects at the output of the modulator. The RF SWITCH module is controlled by the RF SWITCH CONTROL module control circuit, which is commanded by the RF ON/OFF signal coming out from the PC (Fig. 1). The modulator has 0 dBm nominal input power and 256 output levels (up to 3 dBm).
2.1.2- VERIFYING THE FREQUENCY RANGE OF THE COMPONENTS
The main components of the old modulator, their correspondent frequency range and the frequency in which they must work for the upgraded modulator are shown in Table 1:
Component | Frequency range (MHz) | Working frequency (MHz) |
---|---|---|
PHASE SHIFTER (combinations of PSJ-2-1, see [21]) | 1-200 | 10 |
MIXER 1 (GRA-3, see [21]) | DC-200 | 10 |
MIXER 2 (GRA-3, see [21]) | DC-200 | 10, 74, 64 |
RF SWITCH (PAS-3, see [21]) | DC-2000 | 74 |
PREAMPLIFIER (MAV-11, see [21]) | DC-1000 | 64 |
PASSBAND FILTER (Already designed) | DC-4 | 64 |
As can be observed, all the components from the old modulator can work well at the 64 MHz frequency, except for the passband filter, which needs to be redesigned and rebuilt for working at 64 MHz.
2.1.3- CHECKING THE PERFORMANCE OF THE RF SWITCH
The signal output from the modulator is necessary only when exciting the substance under study; therefore, it needs to be suppressed or attenuated during the reception process. This suppression was carried out using the PAS-3 part in the old modulator as shown in Table 1, but in the upgraded version, with the rising of the working frequency up to 64 MHz this component was not good enough to achieve this function and a new RF switch was used: the KSWHA-1.20 [21]. This new RF switch has a 69 dB isolation from the input to the output port when turned off and 0,75 dB insertion loss when turned on, all this for 64 MHz and at RF level of 0 dBm.
The new switch requires certain control signals and the electronics necessary to control this switch with the right signal levels was designed and built by the authors of this work. There are two pins to control this switch: to turn it on, a 0 V is required on a pin and a -8 V is needed on the other one, and to turn it off, the voltages must be interchanged. The control circuit for the RF switch is shown in Fig. 3. This control circuit is driven by a TTL signal coming from the PC and at its output there are two signals: 0 V and -8 V on each terminal and interchangeable when needed that allows to activate or not the RF switch.
2.1.4- THE OUTPUT FILTER OF THE MODULATOR
The signal from the modulator is sent to the output through a passband filter centered around 64 MHz that was designed and built by the authors of this work. This filter was designed to have cut off frequencies at 60 MHz and 70 MHz respectively and a bandwidth of 10 MHz. It was also designed to be coupled to 50 Ω.
The designed filter is of type π and its schematic is shown in Fig. 4.
The component values are calculated through the following expressions [22]
Where:
|
is the passband lower frequency limit, and is equal to 60 MHz. |
|
is the passband higher frequency limit, and is equal to 70 MHz. |
|
is the load resistance, and is equal to 50 Ω. |
|
are intermediate results. |
|
are the final values of the components. |
Once the components were calculated, the commercial values were selected as shown in Fig. 4.
3.- RESULTS AND DISCUSSION
With all these changes made to the modulator, a validation process was carried out. First, the theoretical and practical frequency response of the filter alone was determined because its frequency response is important to the system; second, the correct performance of the modulator is measured; and finally, the correct performance of the spectrometer with the upgraded modulator fitted into is tested.
The two frequency characteristics of the design and implemented filter are shown in Fig. 5, the black line being the ideal response of the designed filter on software, and the red one, the practical response. The 3 dB frequencies of the real filter, 60 MHz and 70 MHz, have an attenuation of 20 dB and 15 dB with respect to the ideal response. There are three vertical lines in Fig. 5: at 54 MHz, 64 MHz and 74 MHz respectively. The 54 MHz and 74 MHz frequencies were selected because they arise in the process of mixing and are very close to the filter bandwidth (this is not desired) as will be calculated later, and need to be attenuated. At 64 MHz there is an attenuation of the real response with respect to the ideal one of about 19 dB, and of 13 dB with respect to 0 dB. At 54 MHz there is an attenuation of the real response with respect to the ideal one of 16 dB, and of 48 dB with respect to 0 dB. At 74 MHz there is an attenuation of the real response with respect to the ideal one of 16 dB, and of 30 dB with respect to 0 dB.
To verify the correct performance of the modulator, the signals entering its ports were set up as follows: Three signals were provided to the BASEBAND SIGNAL INPUT port in Fig. 2 one at a time (gaussian, square, sinc), and each one had a 1 mV peak-to-peak amplitude and a 0.2 ms length. The signals at the RF 1 INPUT and RF 2 INPUT ports in Fig. 2 were sinusoidal signals of 10 MHz and 1 V peak to peak, and 74 MHz and 1 V peak to peak, respectively. The 64 MHz output signal obtained from the modulator was captured and digitized with a digital oscilloscope and further processed using the FFT in order to find the characteristic parameters of the modulator such as: harmonic suppression, sideband rejection and conversion loss.
Figs. 6, 7 and 8 show the measured signals and their spectra at the output of the modulator for the three pulses used: gaussian, square and sinc, respectively. In Figs. 6 a), 7 a) and 8 a), the envelope of the output signals can be observed. The important fraction of the signals in Figs. 6 a), 7 a) and 8 a) lies between 0.18 ms and 0.38 ms approximately, because it is in this interval that the baseband signal modulates the carrier, but the RF switch is turned on and allows the signal from the modulator to go out from 0.07 ms to 0.48 ms. The RF switch is turned on before the modulation process for the proper functioning of the modulator, because there is a delay to wait until the baseband signals and the other RF signals go through the electrical components of the modulator and finally get mixed and go out of the modulator. In the interval from 0 to 0.18 ms and from 0.48 ms onwards approximately, the signal is highly attenuated because the RF switch is turned off. This demonstrates the good performance of the RF switch and allows the modulator to work well.
In 6 b), 7 b) and 8 b) it can be observed that the highest power frequency component (highest level marked with a red circle) is always at 64 MHz, which corresponds to the desired frequency output and is attenuated due to the characteristic of the output filter at this frequency. The 54 MHz and 74 MHz frequencies (levels marked with a blue square) are present at the output and attenuated by the filter. For a detailed evaluation of the modulator performance, harmonic suppression, sideband rejection and conversion loss are measured and calculated from the data plotted in Fig. 6 b), 7 b) and 8 b), taking into account the main frequencies at the output of the modulator and others resulting from the mixing process.
The most important harmonics in the modulation process are the 3rd and the 5th because they are the closest to the working frequency [21]. The measured harmonic suppressions of the 3rd and 5th orders are shown in Table 2: The measured harmonics were the sum and difference between the carrier and the products of the intermediate frequency (64 MHz + or - n*10 MHz, with n being 3 and 5).
Harmonics | Orders | Levels (dB) | Suppressions (dB) | ||||
---|---|---|---|---|---|---|---|
Gaussian | Square | Sinc | Gaussian | Square | Sinc | ||
94 MHz | 3rd (64 MHz+3*10 MHz) | -59 | -57 | -59 | 44 | 43 | 43 |
114 MHz | 5th (64 MHz+5*10 MHz) | -62 | -74 | -72 | 47 | 60 | 56 |
34 MHz | 3rd (64 MHz-3*10 MHz) | -76 | -73 | -66 | 61 | 59 | 50 |
14 MHz | 5th (64 MHz-5*10 MHz) | -70 | -72 | -80 | 55 | 58 | 64 |
As it can be observed, all the suppression levels are equal to or greater than 40 dB, which ensures good harmonic attenuation and a cleaner signal, limiting the possibility of exciting the probe at other frequencies.
When measuring sideband rejection, there were two important frequencies near the 64 MHz frequency (this was the important frequency) which needed to be measured and were the result of the mixing process of the intermediate frequency (10 MHz) and 64 MHz; these frequencies were 54 MHz and 74 MHz. Rejection values are shown in Table 3.
Frecuencies | Rejection (dB) | ||
---|---|---|---|
Gaussian | Square | Sinc | |
74 MHz (64 MHz+10 MHz) | 20 | 17 | 17 |
54 MHz (64 MHz -10 MHz) | 40 | 52 | 45 |
The minimum rejection of the 74 MHz frequency is 17 dB. It is low, but it does not affect the experiment because the probehead is tuned to 64 MHz and further attenuates the unwanted frequency of 74 MHz. The minimum rejection of the 54 MHz frequency is 40 dB.
The conversion loss of the modulator was measured taking into account the two RF inputs of the modulator and the RF output. Conversion loss was approximately 20 db. It is a high conversion loss, and the cause of this loss is the filtering stage. This loss was corrected by introducing an RF power amplifier after the modulator.
Finally, to validate the correct performance of the spectrometer, the upgraded modulator was fitted into to the spectrometer and a real acquisition of an NMR signal was carried out.
To carry out this experiment, a sample of choline was dissolved in H2O. Its NMR signal and spectrum are shown in Fig. 9. The NMR signal shown in Fig. 9 a) corresponds to one acquisition without accumulations and it is obtained with 256 time points and with a sampling frequency of 1.3 kHz. From Fig. 9 b) it is observed that the choline has a characteristic spectrum with a main singlet resonance at 3.18 ppm, and water has a characteristic spectrum at 4.7 ppm, with the chemical shift (ppm) relative to DSS (2,2-dimethyl-2-silapentane-5-sulfonate), with both resonances visible in Fig. 9 b). Choline has a couple of additional resonances at 4.05 ppm and 3.5 ppm, but their amplitude relative to the main resonance are too small to be observed in the spectrum [23].
4.- CONCLUSIONS
Upgrading the modulator made it possible for the spectrometer to work properly, providing the latter with the new functionality to operate at the 64 MHz frequency; and thus, to carry out experiments at 1.5 T. This field is very common in MRI clinical machines; therefore, the spectrometer may be used to analyse substances and compare the results with those of other studies conducted on 1.5 T MRI machines.
This paper demonstrates that it is possible to upgrade the RF modulator for different frequencies, depending on the desired applications; as long as it is a modular design. This upgrading was accomplished at a low cost, with common electronic components, and within a reasonable time interval.
The upgrading of the RF modulator, which constitutes an electronically open system, allows conducting research in the fields of biotechnology, medicine and others, and could become an experimental and teaching basis for learning in NMR technology.