How credible is DPRK's nuclear stockpile?

In 2006 North Korea claimed it conducted a "nuclear test". But many analyst suspected it was a faked test mimic by traditional explosive, for 2 reasons:
  1. the energy was only bout 1/3 of the Hiroshima bomb, showing that the explosion is likely really from a nuclear bomb. Because for a chain reaction to happen, there need to be at least certain critical mass of Uranium (or Plutonium), as demonstrated first by physicists Rudolph Pierls. And the Hiroshima bomb is just over the critical mass
  2. There had not really been confirmed radioactive trace from the surrounding international observation. (The US said the signal is "consistent" with a nuclear test, but radioactively of this small amount (for underground test) is easy fake. Note also the choice of word, "consistent" is a much weaker word that "confirm", which will mean consistency in both strengh, spectrum, and also spatial distribution. "consistency" only means that there is some radiation detected and that no obvious contradiction)
The most recent claim has a magnitude of 4.7 (vs 4.2 in the previous test). The energy is about 3 times larger ( 10^(4.7-4.2) = 3.1 ), making the total energy similar to that of the Hiroshima bomb.

For a nuclear test it still seems to be very minor. Two possibilities
  1. It is a real nuclear blast, but DPRK has barely enough Pu/U, so they used all the 'available' material for this "test"
  2. It is another fake test, they figured the previous test was 2/3 short in magnitude, and got the right amount to make a more credible fake (it would to too expensive to use more)
p.s. it is hard to accurately measure the energy with seimological measure, as the rock/water of the surrounding is complex and differ from the calibration references (for both seimic calibration and also underground nuclear test energy calibration). the most reliable data is perhaps the 'relative scale' (i.e. the difference) in energy for data taken by the same probe(s).
e.g., from one source (NEIC National Earthquake Info Center, which i think is the USGS) the numbers read 4.7 (2009) and 4.3 (2006) so the difference is 10^0.4<3.>