Equipment with piezoelectric transducers had insufficient signal amplitude. The Metriguard had insufficient amplification. The Fakopp worked well, but was not nearly as convenient as a hybrid system consisting of the Fakopp timing circuit and receiving transducer and a Metriguard impact hammer.

The sound velocity was also determined by impacting one endface and recording the resulting waveform at the same endface. This was analysed by Fast Fourier Transformation. This yielded the first four longitudinal resonance frequencies. The fundamental resonance frequency was used to calculate the sound wave velocity.

The logs were assessed for quality by measurement of the small and large end diameters, Pilodyn penetration, and branch index (BIX). The logs were sawn into boards whose modulus of elasticity (MOE) was determined, and the average MOE of each log was calculated.

The highest correlation between sound velocity and log MOE occurred for the hybrid system, with R2 = 0.65. The Metriguard was at the upper limit of its amplification and this was inadequate for logs that consist of at least 50% sapwood. Sapwood has high sound absorption because of its high moisture content. The coefficient of determination R2 was only 0.30.

The vibration technique with FFT analysis yielded velocity values that correlated to the log MOE with R2 = 0.62. This technique has the advantage that excitation and detection occur at the same endface, so that only one operator is required.

In one grading simulation it was calculated that if only SED and sonics were used to sort the logs, 35% of logs were misclassified. If only SED and branch index were used to sort the logs, 39% were misclassified, so that sound velocity is a better predictor than BIX for these logs. However, multivariate analysis has shown that relying on sound velocity alone for log grading is considerably less effective than including branch index, even though it is more effective than branch index on its own. Selection based on SED, sonics and BIX misclassified 30%, and increased actual recovery to 52% out of 60% of the logs. It is concluded that the best log selection system is based on SED, BIX and sound velocity, the latter using waveform analysis based on FFT.