The search for
 new physics.

Home Built Nuclear Magnetic Resonance ( NMR ) Spectrometer

by Doug Marett ( 2010 )

Rough Notes on the Project:

   These notes are cursory due to the fact that this NMR spectrometer was built around 6 years ago. In selecting the NMR frequency, this was chosen based on the magnetic field strength of the radar magnet that was available. The radar magnet was measured and found to have a field strength of 4.33 Kilogauss in the gap. It could be varied from 4.0 to 4.6Kg using the variable regulator unit, which can swing the current +/- 3 AMPS. The device is currently set up to operate at 18 MHz since this is the approximate H proton resonant frequency in this gauss range. The H protons should ring at 4.232 Kilogauss. This requires about -0.8 -1.20 A to compensate. The F protons should ring at 4.498 Kilogauss. This requires about +1.68A to compensate.


The pulse forming circuit was derived from here.  The schematic is shown below:  

The 18 MHz oscillator is a simple can oscillator. Double balanced mixers controlled by a CMOS 4066 switch (triggered by the pulse forming circuit) switch the transmitter on and off for the 90 degree and 180 degree pulses. These are set to around 20-25us and 40-50 us in length, respectively. 

The transmitter preamplifier is a simple RF transistor amplifier to bring the RF signal up to a few watts, sufficient to drive the final amplifier stage. 

 The final amplifier is a KL 203 Linear Amplifier from RM Costuzioni Elettroniche in Italy. Maximum output power is 100 watts. It can be found at:

   The NMR power supply uses a DC supply of > 28V, which is then regulated by adjustable LM350's. This allows for the +/- 3 amps of DC current to the radar magnet coils. A switch allows the current direction to be reversed, so that the magnet variable gauss range is larger.  

Schematics:   NMR Power Supply:


The power supply LM350s are supposed to handle 4.5 amps, but tend to fail at lower currents.

NMR Receiver:


The receiver has an input protection array of multiple pairs of 1N914 diodes. A LNA-1B linear amplifier was used for stage 1 and 2, from the now defunct Erickson Engineering. The feature a broadband gain of about 25 db between 15 and 200 MHz.  Oscillation feedback between the stages was an issue, which was mitigated by the use of impedance matching band-pass filters between each stage. The final amplifier is a modification of a circuit featured once on  Norberto Raggio's website .  I included some tuned circuits in it to try to eliminate RFI, which is a considerable problem in my neighbourhood.

Experimental Results:

    Here are some examples of the FIDs and spin echo's I have generated on this device. My principle interest was in this phenomenon, and less so on the H proton spectra itself - spectra can be generated by converting the output signal to DC and then performing a scan of the DC intensity vs. magnetic field strength for a given pure compound.


An example FID signal from glycerol. Signal intensity is about 700mV, the noise is about 20mV before pulse, about 100mV after.   S/N is about 7:1 to 35: 1.

Spin-Echo: this is somewhat weaker – S/N about 3:1 (signal about 70mV) (Glycerol):


Spin Echo shown 10mV resolution, 25uS per division, glycerol sample:


Spin echo again, glycerol, 200uS per division:


Experiment with Tap Water:

   Tap water produced a FID and spin echo at about 100-200mV (S/N = 5-10:1). When a 90 degree pulse followed by a 90 degree pulse was used (each 50uS long), a spin echo followed, followed by another weaker one the same distance further on, and another one still, barely discernable. Perhaps this is like the Carr-Purcell sequence.

Experiments with Teflon:

   A piece of Teflon was rolled up and inserted into the coil, without a test tube. The peak from Teflon Fluorine protons was about 100mV in 2004, about ½ that of glycerol. It had two visible peaks in intensity with magnetic field strength. I attempted to view the spin echo, but could not find it immediately, and then the magnet power supply overheated. The Fluorine FID was found at around +1.82 amps in 2008. The Fluorine FID could be seen at about 300mV with rolled up Teflon inserted into the sample coil, but not with NaF as a solid or in solution. It was also seen with PVDF powder in a sample tube, although the signal was only at about 50mV with 25 mV noise. When the PVDF was dissolved in acetone, it could no longer be seen. No spin echo's could be found for Teflon or PVDF. 

March, 2008 - Attempted to improve the final stage receiver:

   A simpler circuit was created for the final stage receiver using LMH 6609 operational amplifiers.


The load resistor was introduced on the output of amp 1 to prevent oscillation. The Q of the series circuit was much higher in this configuration. With a second filter stage (3 amps total) The gain was 4000x, and the Q was 128:


This above amplifier was then tested alone in the NMR circuit, see picture below:


Closeup of the FID signal:


     Video's of this project can also be found on YouTube: .

About the Author:

 Doug Marett is a research scientist who has been working in industry for the last 25 years. After receiving his M.Sc. from the University of Toronto, his focus has been primarily in the design and development of new technologies and products. This has included patented inventions in the biotechnology sector such as novel medical devices, and more recently product development in the field of optics and rapid prototyping. He has been involved in a diverse range of projects including NMR spectroscopy, inertial navigation and optical interferometry. Being an experimentalist first and a theoretician second, Doug Marett adheres to the philosophy that experimental research should drive theory development and its validation, not the other way around.

Doug Marett working on a Helmholtz coil experiment, 2015.