That error is almost certainly
while auto-guiding. That's why you can see in the period field (with the check boxes), there is no component with a 638-second period (which would be the fundamental of the worm - since the HEQ5 has a 135-tooth worm wheel, one rotation is 86164 seconds / 135 teeth = 638 seconds). All of the time bases are small (<60 seconds) which means they are either seeing artifacts or gearbox noise.
To measure the mount's actual periodic error auto-guiding needs to be turned off; you need to use the PEMPro calibrate wizard as well to compensate for the star declination (or, alternatively when importing into PECPrep, you have to check the button "compensate for star declination" and input the star's declination in the little box).
The reason for this is because a star at the celestial equator (declination = 0) will show the greatest periodic error, while a star at the pole (declination = 90) will show no error. The actual error is the measured error (from the PHD log) divided by the cosine of the declination.
The only mounts that can provide <2 arc-second peak-to-peak periodic error without auto-guiding have names like Astro-Physics and Software Bisque.
