In a report from East Asia Intel (subscription), a North Korea defector describes parts of the nuclear program, including a nuclear weapon he claims to have seen in 1992. He said that North Korean scientists discovered a method for attaining a secondary explosion from Pu-240, which he claims was used in the October 2006 nuclear test:

A North Korean nuclear technician has revealed new details of Pyongyang’s plutonium weapons program, including support from Russia and Europe for components and testing equipment.

[…]

“A nuclear bomb was completed by 1992. I saw the bomb only once in 1992. The North Korean nuclear bomb was one meter in diameter, one meter in height, cylindrically shaped and, I believe, weighs just over one ton,” the defector was quoted as saying.

“The overall weight of a nuclear bomb could not be four or five tons. It was just over one ton. As for the detonator agent, they obtained data from Russia and are producing the agent domestically. Production of the agent is not very difficult. Detonator tests were conducted after the agent was obtained.”

The defector disclosed that an accidental nuclear explosion occurred in North Korea in 1991. “The 1991 nuclear accident enabled North Korean scientists to learn how they can achieve explosions [fission] using Pu-240. While critical-point calculation is difficult, individuals such as Chancellor Pak Kwan-Ho, who is also in charge of the Supercritical Research Laboratory at Kim Il-Song University, supposedly gained access to data on the speed of chain reactions and increase in neutrons, and acquired the capability to control the chain reaction.”

The defector disclosed that the Nuclear Weapons Research Center at Yongbyon is in charge of a program to build a small nuclear bomb. He said scientists reported to Kim Jong-Il in 2000 that that they had succeeded in miniaturizing the plutonium core from eight to six kilograms.

“In 2002, I heard scientists saying, ‘If any test on miniaturization [reducing plutonium] should fail, we are thinking of a secondary system to attain an explosion [fission] via use of Pu-240.’ They definitely used that approach in the tests executed in October of last year,” the defector said.
The report estimated that North Korea’s nuclear bomb has a yield of between 4 to 15 kilotons.

“The amount of Pu-239 used in last year’s nuclear test can be estimated at six kilograms but, psychologically, I believe, scientists had no confidence that the tests would succeed and had executed a small-scale fission [premature] explosion using Pu-240,” he said. “They should exercise confidence at eight kilograms, while acquisition of valuable data in connection with the recent test should enable them to achieve success in connection with the next test,” the defector said. (emphasis added)

The Pu-240 claim is interesting, as it is not considered suitable for nuclear weapons, although that may not be the case.

Whether plutonium with higher Pu-240 content is usable in a nuclear weapon is not certain from the unclassified literature, but some sources claim that it is possible [3]. Even if technically feasible, such a weapon’s yield would be unpredictable and probably much lower than with weapons-grade plutonium given a similarly sophisticated bomb design, and the radioactivity of the Pu-240 would considerably complicate handling.

That may in part explain the low yield of the tested device. The reference above (PDF) also raises issues about the viability of light-water reactors (LWR) for proliferation prone countries, like North Korea:

The bottom line — LWRs no longer should be given to any nation that might divert the reactor’s fresh lightly enriched fuel or the plutonium-laden spent fuel to make bombs.

The report details how fresh and spent LWR fuel can be used to accelerate a nation’s illicit weapons program significantly. In the case of a state that can enrich uranium (either covertly or commercially), fresh lightly enriched reactor fuel rods could be seized and the uranium oxide pellets they contain quickly crushed and fluoridated. This lightly enriched uranium feed material, in turn, could enable a would-be bomb maker to produce a significant number of weapons with one-fifth the level of effort than what would otherwise be required to enrich the natural uranium to weapons grade.