Oh that's a dip for sure Adam! First port of call is to go back to the rollers with some longer trumpets. The inlet manifold looks ok from the best I can see in the pic...short but good flow like the ones I make. On that bases I'd start with some 60mm trumpets with due note to the heat/close proximity to the radiator I mentioned before. I'd do that before any manifold change. As for the manifold, yes I'd start with the dimensions given. There is "no" hard and fast rule but I can assure you it's a fair starting point. In an ideal world you'd try a few different combos to see what your particular engine likes, but ive had good results with dimensions in that ballpark. Just like cam timing, once your in the right window there isn't huge differences. I have found that too efficient a manifold can cause issues with fueling set ups with carbs in the lower range with radical cams on 8v's, that may not be the case on a medium cam 16v. The last 16v to a similar spec to yours needed some odd size emulsion tubes on its longer trumpets and 4-1.
The dip is influenced by exhaust, cam and intake manifold dimensions on a NA engine. Almost every 16v I have tested has that phenomena, just at different rpm points. Sticking carbs on it, with their narrow window of optimum fuel air mixing, promotes more of that dip to higher engine speeds. I agree with Jason, by altering the intake tract with different velocity stacks. Make sure you have little noise the experiment, by baselining the car, in the condition it was on the day. Then make the changes. Ideally, you want to use parts that broaden the 90% + torque mark at lower and higher engine speeds. The trumpets will clawback the lower end torque by 2 to 300 rpm ( all things being equal) and the right exhaust will promote a bit of blowby at lower revs as you draw in fresh charge to promote cylinder fill at high rpm, pending your cam position. Your cams LCA and the reference position, will tilt, step up and shift the torque curve if it is not in an already optimum position also. That is the theory. Actual results depend on methodical experimentation. Looks like you have an pretty stonking engine having looked at your power curve on that type of dyno. As a reference, on the same type of Dastek rolls, I tuned a K-Jet 1.8 16v Championship MK2 car to 158 bhp @ 6700 rpm and 135 lbft @ 5800 rpm. Bare in mind this would vary on a different brand of rolls and these engines have to be pretty much stock. They was also have a dip at @ 3000 rpm and then it slowly woke up at 4000 rpm onwards.
Its very instructive to look at what other well developed engines have used, from the best tuning houses. Hugo put up a factory vw item...thats going to be on the money. Below is something developed by two of the best in the buisness, RLM Racing and Simpson exhausts. This is for a tuned Hayabusa fitted to a Radical PR6. Stepped design 4,2,1 with quite long secondaries. This engine will have some pretty hot cams and be revving to around 11K. Probably 260hp ish from a 1500. Many of the caterham and westfield type race cars also go with long secondaries, not as revvy as the busa but still probably 8500rpm type machines Below is a Westfield racer with a millington diamond engine in. But its often engine specific. Best bet is to find a car thats winning races with an engine like yours and copy what they have.
