f, f. ;
(A) (Lo) 0 RF [Amplitude |
® | modulator
Digital
code
(B) Digital A
code y DA
Digital A
code y DA
I-channel
(C)
Q-channel
Digital A >
code y DA dissi X
Figure 2: Digital signal generator mechanizations: (A) RF phaseshifter
| (B) Direct IF generator (C) Quadrature baseband modulator.
Also the digital codes can be generated in a number of ways. One method to generate the
quadratic phase of a linear FM is for instance to implement a digital double integration of a
constant, [Eber and Soule,1975],[Postema,1987]. By changing the constant the FM rate can
be varied. Alternatively the pulsecode can be generated off-line and down loaded in a fast
buffer memory. This approach gives significantly more flexibility and the software
generating the codes can be designed to calculate non-linear FM with specified pulselength,
bandwidth and sidelobe level. The price paid for this flexibility is that a fairly large and very
fast buffer memory is required.
It is important to note that high quality signals with low sidelobes can only be obtained if the
gain of the quadrature channels are equal and if the phase difference between the channels is
90°. Also the characteristics of the DA converted signals (i.e. digital signals have a periodic
spectrum, and the "box" representation of the individual numbers in the code gives a
sinc(nf/f;) weighting of the spectrum) must be taken into account as well as other system
components such as filters.
The KRAS digital signal generator is based on a quadrature modulator with the converters
running at a 200 MHz sampling rate. The low-pass filters have a cut-off frequency of 150
MHz and a nearly linear phase characteristic. 4096 complex code words are stored in a RAM
buffer, hence supporting pulses of a length up to 20 us. The codes are generated by a C-
program that supports both linear and nonlinear FM codes. This program also hold the
capacity for incorporating predicted or measured system properties, by generating
predistorted signals, so that the signal that drives the TWT will be correct. It is also intended
to use the calibration system described below to perform on-line measurements of transfer
functions thereby enabling dynamic predistortion if required. How this can be implemented
is presently subject to a Ph.D. study at the Electromagnetics Institute.
94