Based admittedly on a limited sample (2/2), the R&D sunshine recorders
seems to suffer from slight over-recording in unbroken sunshine, by
1-2%.

This is easily identified if you log the pulses to a datalogger, when
an hour of unroken sunshine will more often than not show up as 1.01
or 1.02 hours (i.e. 101 or 102 pulses per hour), but the overcounts
are much less easily identified if only the daily total is available.

On recent sunny days, my unit has over-recorded by a little more than
1% - for example on 2 June 15.39 h (corrected to 15.20 h by changing
all hourly values 1.00 h to 1.00). Normally this doesn't make very
much difference, but on days with unbroken sunshine it can produce a
total considerably closer to the astronomical daylength (i.e. sunrise
to sunset) than is actually the case. These units are much better near
the horizon than a Campbell-Stokes recorder, but even with clear
skies, no dew and a good horizon in my 8 years of using one it seems
they are unlikely to pick up more quickly than about 15 minutes after
sunrise, and stop about the same before sunset (corresponding to just
2-3 deg of solar elevation).

Those of you with R&D units might care to note carefully the displayed
total duration as shown on your unit over a few hours on a day with
unbroken sunshine to see whether yours does the same. Over 5 hours,
say, timed accurately (within a few seconds) I suspect you'll see
5.05-5.08 hours.

I'd be interested in whether my suspicions are correct - feedback
welcome.

--
Stephen Burt
Stratfield Mortimer, Berkshire