[Dessertrunner]

Trevor,
I know it sounds strange but in a closed loop system it is impossiable to supply postive pressure to the suction of the pump no matter how big the pipe between the two. As strange as its sound it took me a lot of head scratching to get my mind around it but it is a fact and I have confirmed it to be true.
If the pump is with out restriction then the coolant turns up at the pump inlet and any pressure the pump inlet would have been supplied with will be lost by the pump pumping faster. The pump will just pump more which will cause any pressure gains to be lost. It makes sense because at some point the coolant flow and pump revs balance.

In answer to the difference between pressure restriction and vacum restrictions. I will take a side step and give an example. One of my irrigation pump sites has 3 by 70litre per sec pumps and on the up stream side I have 4 pivot irrigators. Hang in there I am getting to the point.
If I open the valves to two irrigator which need 140l per sec and I only turn 1 pump on the cavition is so bad you near go deaf. Once I turn on the second pump the noise and problem goes away.
Problem was system up stream wants and could handle 140lper sec but as I only had 1 pump running the suction on that pump could only supply around 70lper sec. Because there was no up stream pressure the pump wanted to pump 140lpsec and tried. Now if I close the valve on the pump half way I have created a restriciton and even though I have one pump running the pressure restriction is enough to reduce flow on the pump back to 70lp Sec noise now goes away.

The answer to your question is the suction and outlet need to be in balance even or possiably more so in a closed system. The challange we have and will face is finding the balance. Tom and I have changed to the top pipes as did Subaru this then lets the coolant around faster (more flow per minute) as in my example now the cavitation happens at a lower revs.
Your suggestion to drill the impeller does fix the problem because it reduces the pumps pumping rate per min which inturn means that the suction has to do less work and is less affected by the restriction. The only problem with this is that in our engine we need evey litre of coolant we can get around the system.

I have one further example. If you sit in a bath and you want the water to mix or circulate so the temp is the same at both ends you pull you hand along and the water follows. It goes around the side of the bath and back to your hand at which time you pull it again. If you to restrict it on the down stream side of your hand a vacum pocket would form. If you were to block the up stream of your had it would not form because it can't get away from your hand (the pump)

Long answer but in short if we decrease the upstream pressure from the pump but don't improve the suction we are more likly to boil our engines because we have taken the two further out of balance