Shuyuan Hu's background study http://hallaweb.jlab.org/equipment/targets/polhe3/lab/pnmr/pnmr_noise_HuShuyuan.pptx
from William A. Tobias @ UVa
I do have some information with me that I can pass onto you concerning a typical pulse NMR circuit we use at UVa to see 3He. I apologize for any errors ahead of time.
The attached schematic  was produced by Yuan Zheng, a graduate student that works in our lab here at UVa, and is based on a pulse NMR system developed by Scott R. (see his thesis on our website "Rohrbaugh, Scott: Virginia PhD Thesis (2008) p. 47") . Note that Scott's circuit is a bit more complex then the one we have used recently, as shown in the attached schematic. The website link is:
On the attached schematic, the gate generator (1) generates a gate (TTL pulse) that is 200 microseconds duration and triggers the RF generator (2) that puts out 55kHz RF (whatever the 3He larmor frequency for the holding field used) in a "gated burst". The cross diodes (3) rated at 1.14Vpp each and the adjustable 1kohm potentiometer (4) are used to isolate noise in case of small signals (these are not needed for large signals). If needed, we have an RF amplifier (5). Via a BNC "tee" we have an NMR probe (6) attached and coupled to the target cell either on a transfer tube or target chamber (ie. 50-turn coil). Between the BNC "tee" and the NMR probe we have a "match box" that impedance matches the probe to our NMR circuit (ie. two capacitors in parallel [one between probe and "tee" and other on "tee" end of first cap going to ground] each with value in the range of 0.47 to 0.68 microFarads that accomplishes 50-ohm impedance matching).
During the gate or TTL non-zero, the isolation switch (7) is open to allow the RF pulse to only pass to the NMR probe coil. When the TTL is zero, the isolation switch is closed to allow the picked up 3He NMR signal to pass to the receiving part of the circuit (preamps, mixer, scope).
The preamp (8) settings are typically: DC couping, channel A, 6dB high, 6dB low, 10k/100k bandpass with some appropriate gain. The reference generator (9) puts out 55kHz ±100Hz to mix down the NMR signal from 55kHz to about 100Hz (ie. we typically mix down to range of about 70Hz to 150Hz, whatever gets us 10 to 20 wiggles of FID (Faraday induction decay) on our scope). The reference generator (9) signal goes into BNC input "R" on our mixer (10). The signal from preamp (8) goes into BNC input "L" on our mixer (10). The mixer output "I" goes to preamp (11) which has typical settings: DC coupling, channel A, 6dB high, 6dB low, 100/300Hz bandpass with approximately 50 gain. The mixed down signal goes into our oscilloscope (12) which is triggered to capture the FID. The scope can be used to download the FID waveform to a PC via GPIB for further processing.
To automate the circuit without the need for signal generator to produce our TTL pulse and a scope to capture the FID, one can replace the gate generator (1) with a pulse created with a National Instruments DAQ card output and replace the scope (12) with a DAQ card ADC "scope mode" input.
In the coming days, we can provide further information on the pulse NMR circuit.
FID should be "Free Induction Decay".
For the mixer (10) I was told by Yuan Zheng that typically the mixing frequency should be sent into "L" while the signal should be sent into "R". "I" is the mixed down signal output. In our setup "L" and "R" were reversed for some reason, because in the beginning this was not fully understood. For the type of mixer we use (ZAD-8) this reversing "L" and "R" may not matter, but it may be important for other types of mixers.