i had played with the impedance of coax cable. connect the signal generator at 50Ω output and a 4meter long coax to a CRO with 50Ω input with a tee. although the web said, the coax impedance is 52, there is no observed different. the peak-to peak signal does not depend on the frequency.

then connect a 12.8MHz coax to the tee and open at the other end. in theory, it should be zero input-impedance and he CRO reading should be

V_{in}= 1 - cos( 2\pi L/\lambda)

where the CRO reading is the input voltage. when the frequency adjusted to 12.8MHz, which is quata-wavelength, the input voltage is equal to 1.

when connect the end of the coax with a 50Ω resistor, the input voltage does not depends on frequency, as expected.

i haven’t try to short the end of the coax.

after that. i going to matching the impedance of the tuner and coil. we use a short coax to connect the input to the tuner before, i replaced it with a 12.8MHz coax. the reason is, the little bit mismatching of the impedance can be saved by the length of the cable, such that:

Z_{in} = Z_0^2/Z_L \approx Z_0

and i opened the tuner to see the circuit inside:

The above is the circuit diagram. there is a fixed capacitor with 2.2pF in parallel. i cannot identify the type of the 2 variable capacitor.  the coax cable can be put in port 1 or port 2 and the coil put in the other port. different configuration has different behavior.

i found that, the input is in port 2 and the coil is in port 1, and the box doesn’t ground but just wrapped with metal sticker. i took it out and grounded to NMR system.

For the input at port 2, the total impedance of the coil and tuner is:

Z_L= - i /(\omega (C_0+C_p + C_s/(1-\omega^2 L C_s)))

The first things to notice is the impedance depends on the frequency. which mean, the impedance matching can only on particular frequency. when the Cs adjusted to matching the frequency. the impedance solely depends on Cp. when the driving voltage gone, the LC circuit will oscillate at natural frequency:

\omega_n = 1/\sqrt{ L ( C_s + C_0 C_p / (C_0+C_p) ) }

Thus, i tired to measure to inductance of the coil by a parallel resistance. but i cannot find any suitable wire to convert BNC cable to wires. after a long time finding, i gave up and wait unit work with my partner.

the input in port 2 is not a common config, so, i changed it to port 1. and the impedance is :

Z_L=i\omega (C_s +L/(1-\omega^2L(C_0+ C_p)))

The impedance also depends on frequency. and the natural frequency when the driving voltage gone is:

\omega_n = 1/\sqrt{L(C_0+C_p)}

there is one way to tune the Cp to match the input frequency and make load impedance solely depends on Cs. by using a pulse signal. and measure the natural frequency of the LC loop, such that the natural frequency is same as input frequency. However, the Low Pass Filter only let frequency less then 1MHz pass and out frequency is 12.8MHz. can i use other pulse source? may be, if i have a mixer.

so, i matching the impedance by very naive way. i use the method on testing the impedance matching. i use a continuous signal source and fixed the frequency at 12.8MHz, then connect it with a coax cable to 50Ω CRO input by a tee, then connect th tee with a 12.8MHz coax cable. the other end of the cable connect to a 50Ω resistor. this setup should be matched impedance. so, i record the input voltage on the CRO. and replace the resistor with the tuner, which port 2 connected to the coil. then adjust the capacitors (both) so that the CRO voltage is same as 50Ω resistor.

by solving the load impedance formula of port 1 configuration, there are mulitple solution for Cp and Cs to give 50Ω. and i think, as long as the impedance is 50Ω at 12.8MHz, any configuration can do the job.

later, i try to find the water NMR signal. although i cannot find any. but the noise level reduced to ±5mV. more or less equal to the background.

i played with the NMR program. the record data is counted by point, so the CRO horizontal setting should set to 5000 points over the screen.

and just before i leave, i don’t know what wrong, the program doesn’t read the CRO signal…