There is some pretty sound advice on reduced loading here:
http://www.guns.connect.fi/gow/gunwriters.html, Just scroll down to the picture of the Late Peter Kekkenon with a blunderbuss.
I go of on a bit of a tangent in what follows, but it does return to guns, so stick with it. Here's my pet theory, with the supporting facts first:
As Al has said, the difference in effects between deflagration (burning) and detonation is down to the rate at which energy is released.
Put crudely, this is known as "power" - the rate of doing work.
If I can release the energy twice as fast, I have doubled the "power". The Quarries that I do work for will typically fire about 15 tons of explosive in a blast, With delays between holes to reduce the vibration levels, that is still giving about twice the power of a Saturn V rocket taking off, for about half a second!
As the previous replies have said, we don't know the exact mechanism which causes a usually predictable and slow burning powder to "detonate".
Deflagration (normal burning) progresses from the exposed surfaces of the powder grains by heat conduction.
Conduction is a slow process, think of how long it takes from the time you pick it up for the heat of a paper cup full of hot coffee to conduct trough the 1/16" or so of skin on your finger tips, so you have to put the cup down again.
Obviously heat gradient and the contained heat energy have a bearing on this too.
I'm quite willing to pick up red hot wood sparks with my bare fingers and throw them back into the fire, but I certainly wouldn't touch a glowing spatter from welding!
Never the less, the process of conduction is slow compared to detonation.
Detonation has the material turning to gas at a reaction front which travels through the material at the speed of sound in that material.
In granular solids, like smokeless powders, that is measured in several thousands of feet per second.
Sonic velocity in a material is proportional to the density (and in solids to the stiffness as well) of the material. References which describe glycerine tri nitrate as "a dense liquid", as though that were somehow un related to it being a particularly powerful explosive are missing half of the reason why it is a powerful explosive.
It's an explosive because it contains a lot of stored chemical energy, which is released when it burns, but it is a "powerful" explosive because that high density allows the reaction front to travel much faster than it would in a less dense medium.
Explosives also suffer what is termed "dead pressing".
If the density is too high, they won't detonate.
There is also an effect in weaker explosive, where a cylindrical charge has to be a certain minimum diameter before a detonation will manage to run allong it.
You see both of these effects as limitations in the ammonium nitrate - fuel oil explosives used in mining and quarrying: Dense emulsions and slurries need to have glass spheres, perlite or polystyrene granules included to reduce their density below the point of dead pressing, or they just won't go off.
You also see the problem of minimum diameter, prilled anfo explosives generally like to be in columns a few inches in diameter, otherwise you tend to get squibs.
I suspect that both of these effects are due to energy being bled away from the detonation front.
Nitroglycerine has a lot more energy per pound than most other explosives, so there is plenty of heat in that reaction front to keep the detonation going, despite the high speed that the shock wave can move through the material.
I suspect that in less energetic explosives, that the effect of dead pressing is due to the shock wave being able to outpace the chemical reaction, allowing the energy to run away from where it's needed for the reaction to continue.
Introducing voids, reduces density and so slows the sonic velocity and stops the shock from running away ahead of and outpacing the chemical reaction. (there is also the side effect of the gasses in the polystyrene beads flashing to white heat as they are compressed by the shockwave, and giving additional points of initiation).
The minimum diameter problem, is I think also due to shockwaves bleeding energy away from the reaction, through the host material. This affects prilled anfo in rock blasting. The sonic velocity of the anfo is around 3,000 feet per second, the sonic velocity of the rock is anywhere from around 6,000 for crappy broken stuff, to about 30,000 feet per second for really solid stuff like rocksalt or anhydrite, and really good granite or dolerite.
Back to guns.
Factors that seem to have favoured detonation seem to be:
- relatively large case capacity
- Relatively slow burning powder
- much reduced load
Larger case capacity gives reduced surface area relative to volume for quenching to occur in.
Relatively slow burning powders are in larger and denser and much harder to ignite grains than the porous or flaked grains of fast pistol and shotgun cartridges.
There you have two of the factors from my theories about how detonations progress or fail to.
The third, reduced loading, I don't really understand.
My pet theory is that the layer of heavy grains of slow powder lying in the bottom of the case, only gets warmed by the flash of hot flame and particles from the primer passing above it, rather than having the primer flame pass through the voids between the grains and bathing all surfaces of the grains.
My guess it that that may set the dense (so good conductor of heat away from the flash) powder grains cooking, and decomposing without really taking light - sort of like a barn full of wet hay heating up.
My guess is that over several miliseconds or more, that this fills the case with hot gas and it consumes any deterrent coatings the grains may have had. eventually, like the barn of steaming hay, the whole thing takes.
My guess at this point is that the process is maybe a bit more like pinking in a gasoline engine, that the detonation shock wave is propagated through the hot gas, rather than through the solid powder grains.
The cooking will have removed the deterrent coating though, so any burning of the powder will be much faster than if it had started burning properly, heat conduction will also be much faster due to the rapid pressure and temperature rise with the gas detonating.
That's my guess for why "detonation" is more likely with a big case and a slow powder.
By "Big" and "Slow", I've read accounts of KBs with .300 whisper(tm) and Win 296, though the question of squibs arises:
296 is a hard powder to ignite - check out all the unburned grains you get with it! I'm not sure that with a low recoiling and quiet round like the whisper, the firer would necessarily have recognized a squib and the bullet stuck a few inches up his barrel.
Anyhow, fast powders are easily lit, and seem to be the ticket to safe reduced loads.
Alternatively, get a pistol calibre carbine and remove the big capacity case from the equation
