Hello.

I have the AD9914 EVK operating in swept frequency mode. In this mode, you offer frequency step AND Sweep duration. I am writing to determine the chip algorithm / priority.

Given the example Fstart =400MHz Fstop=450MHz Fstep=10MHz Sawtooth mode Sweep time =24.6e-9 Sec.

This example (Used to make a point, not implimentable) sweeps 11 frequencies. Will the dwell time be 24.6e-9 / 11 or will the dwell time vary so the DDS completes a complete sine wave then moves to the next - noting 1/400MHz takes more time than 1/450MHz.

This example is detailed below. Note in order to complete one full cycle of each of the 11 steps below it takes 24.6e-9 seconds.

Thanks! Jim

Step | Freq | Duration of one cycle |

1 | 4.00E+08 | 2.50E-09 |

2 | 4.10E+08 | 2.44E-09 |

3 | 4.20E+08 | 2.38E-09 |

4 | 4.30E+08 | 2.33E-09 |

5 | 4.40E+08 | 2.27E-09 |

6 | 4.50E+08 | 2.22E-09 |

7 | 4.60E+08 | 2.17E-09 |

8 | 4.70E+08 | 2.13E-09 |

9 | 4.80E+08 | 2.08E-09 |

10 | 4.90E+08 | 2.04E-09 |

11 | 5.00E+08 | 2.00E-09 |

Per your example, the number of steps is 11 and the sweep time is 24.6ns, corresponding to 2.4ns per step. As you point out, this is not implementable because the minimum step period is ~6.86ns with a 3.5GHz system clock.

The start and stop frequencies, 400MHz and 450MHz, have periods of 2.5ns and 2.22ns, respectively.

The DDS generates whatever whole and/or fractional part of a sine wave is necessary per the step period and frequency associated with a particular step. Note a DDS is particularly suited to this type of application.

Here is how it works. A basic DDS is an accumulator followed by a trigonometric converter (phase to amplitude) and a DAC. The trigonometric converter maps the full range of the accumulator (0 to full scale) to phase angles from 0 to 2*pi and converts the accumulator output to sinusoidal amplitude.

The input to the accumulator is a frequency tuning word (FTW). The accumulator recursively sums the applied FTW at the system clock rate. Hence, the accumulator output corresponds to linearly increasing phase (which defines a particular frequency). Normally, the FTW is constant, so the DDS generates a sinusoid of constant frequency.

Now consider changing the tuning word in linearly increasing steps at a constant rate. Each new FTW increases the size of the accumulator's corresponding phase steps, which increases the frequency generated by the DDS. This is a seamless process by which the DDS cranks out phase steps according to the FTW at the input to the accumulator.

In linear sweep mode, the FTW is itself, the output of a secondary accumulator -- the frequency accumulator. Programming registers set the step size of the frequency accumulator and the rate at which the frequency accumulator increments. The only "knobs" at your disposal are the step size of the frequency accumulator and its step rate.

When you choose a start frequency, stop frequency and a sweep time, the evaluation board software must decide on an appropriate step size and step rate to meet the sweep time requirement between the start and stop frequencies. If you specify a particular step time, the software calculates the nearest suitable step size (or vice versa).

In any case, the number of sinusoidal cycles plays no part in the linear sweep operation. The DDS simply outputs the appropriate instantaneous frequency based on the prevailing FTW. The number of sinusoidal cycles (fractional or otherwise) that fit in the step period is what the DDS produces.