This has a dramatic effect on the ability of the engine to brake predictably going into a corner. In addition the Ford ECU has no feedback from the brakes so it makes it's own mind about the engine revs, emissions being the priority.
I have decided that the Idle Control Valve (ICV) is the culprit. If you pull the ICV during it's high idle the engine settles at a rate set by the butterfly end stop screw on the throttle body (usually about 500rpm). The ICV is a servo type valve controlled by the ECU, by changing the pulse width thereby modulating the current to it. By observing the waveform to the ICV on an ossiloscope it is about 150uS for a tickover of 1000rpm and increases to 300uS at 2000 rpm. By empirically adding a 33 ohm resistor you can slug the tick over response completely yet maintain a steady 1000rpm but will not surge. The down side is I suspect the emissions will be well out of order (thinking MoT in the future). I therefore decided to only add the resistor during braking. By adding a relay as follows you can force a specific idle during braking.
The down side of this modification is that when the engine is cold the idle is well below normal when braking and can result in a stall.
An initial test drive with this mod looks extremely promising, returning drivability thru engine braking. However during a day out at Knockhill I came back with an engine that would not start properly. During subsequent troubleshooting I discovered that the ICV had failed almost completely but not badly enough once the engine was started but still displaying the erratic tick-over. Replacing the ICV with a part specifically recommended for the 1997 Puma 1.7 made a dramatic cure to the throttle response negating the need for all of the above. Lesson learned: Don't use cheap ICV's from ebay. However I have learned a lot about the ECU and it's control loop. Proper ICV's are adjusted to match the ECU for that model so you must insert the correct ICV part. Also the health of the closed loop system affects the throttle response. An air leak in the inlet or exhaust can affect the signal from the O2 sensor and therefore in turn affect the ICV setting and therefore throttle response.