We analyze the normal ($N$) component of the heliospheric magnetic field data observed by the IMP and the ACE spacecraft (and in some cases as a check, measurements by the WIND spacecraft) for the period 1973 to 2020. Parameters characterizing the frequency spectrum are calculated with a novel technique and bench marked against synthetic data. This technique is based on variances calculated at incremental lags and yields the integral of a turbulence spectrum, formally from infinity to a specific frequency. While it can be used for lossy data, it can however only yield information about the energy and the inertial range for typical values of the spectral indices ($-0.75$ to $- 1.25$ for the former and $-2 $ to $-1.5$ for the latter). This technique was developed to easily and quite accurately analyze large data sets for use in ab initio modulation models. Correlation functions are calculated with a standard second-order structure function. We find that the yearly average for magnetic field magnitude for the period that includes the 2020 solar minimum is at a new low of $\sim$4.2 nT, as is the variance at $\sim$4.4 nT$^2$. Overall the magnetic variance tracks the magnitude squared of the field very well, both showing a clear solar-cycle dependence. The ratio of the magnitude of fluctuations of the $N$ component to the field magnitude has an average value of $0.52\pm 0.02$ for the whole data set, with an increase by about 10\% in solar cycles 23 and 24 compared with cycles 21 and 22. The average value of spectral index of the energy range for the whole data set is $-1.0\pm 0.1$, while that for the inertial range is $-1.69\pm0.04$ for the IMP/ACE data set. The spectral level in the energy range at a timescale of 14 hours and in the inertial range at 5 minutes both show a clear solar-cycle dependence. While the break between the energy- and the inertial range is difficult to determine accurately, an indirect indication of a solar-cycle dependence follows when the ratio of the spectra in the energy- and in the inertial range is calculated. We find a clear solar-cycle dependence for the e-folding correlation length, with a significant increase in values in solar cycles 23 and 24 compared with the previous two.