@ peter kent
As you say, a complex subject!
Maybe the missing link is that a plane shock is not the only way to decelerate through Mach 1.0. If the nose of a body is blunt, or if the angle you are trying to turn the flow through is too big then the shock wave becomes detached from the leading edge of the body. The bit of the shock on the 'cusp' is then actually a very strong plane (normal) shock and the flow immediately behind that part is subsonic. In the case of a sharp surface with a large tuning angle this subsonic flow allows air to escape from the high pressure side of the surface to the low pressure side. This would be the case for example if the flow onto the leading edge of an intake hit it at too big an angle.
Supersonic intakes come in two basic guises - external compression and internal compression. The ramjet intakes you have been reading about are the latter type in which all the deceleration/compression takes place inside the intake. In these designs the final compression is through a normal shock situated at the minimum area 'throat' of the intake where the flow is close to Mach 1.0. This flow is delicately balanced and if some engine disturbance causes the shock to move into the converging supersonic bit of the intake the whole shock system can be expelled giving all sorts of problems (inlet unstart). Generally they are used for high Mach numbers where their higher theoretical efficiency and low external/spillage drag count for more than the additional control system complexity and performance requirements.
In external compression intakes (a simple pitot intake would be an extreme example), all the compression is done by a system of shock waves that sit outside the intake. These intakes are less efficient than internal compression intakes and they also spill a lot of air which produces external drag. Usually restricted to low supersonic Mach numbers.
Concorde's intake was a "mixed compression" design which had some features of each type. At low engine mass flow demands the flow coming on to the cowl lip could be at too great an angle to maintain attached shock waves so it behaved a bit like that described earlier. You can see this most clearly in the left hand picture where the lower efficiency and higher spillage can be seen in the graph of efficiency against intake capture (epsilon). In this state the intake behaved more like an external compression type and there was no appreciable final normal shock.
At high engine demand the angle of flow hitting the cowl was such that the shock waves remained attached and the intake functioned more like an internal compression design. Again you can see this in the right hand picture which shows most of the intake throat covered by a normal shock and in the graph where total intake flow (engine plus bleed) is constant.
On condition there was a bit of each, but since it was designed to minimise spillage you cannot see the detachment of the cowl lip shock at the scale of the diagram.
Hope this is helpful rather than additionally confusing!
PS: Looking at the centre picture again, it occurs to me that the curved shock running from the lip back and up to the reversed "D" would actually be normal to the approaching local flow which was being turned by the ramps and the isentropic compression. This would be the shock you are looking for to decelerate the flow to subsonic conditions. In other words the intake was functioning as an external compression design over this part.
Last edited by CliveL; 28th Oct 2012 at 07:28.