Sorry if this appears as a hi-jack of the thread... not meant to be!
Just out of curiosity, are you running stock cam, or have you upped it? If the cam isn't stock, spec timing won't apply... you have to set initial to what the cam wants (best vacuum), then retard 1 or 2 degrees from there. That is the starting point... If mechanical 34 degrees advance puts you way over final of 34-36 all in, you need to limit it by adding a limiter plate or welding slots.
My cam likes 18 initial, and I also have 34 deg of mechanical, so I made a plate limiting it to 18 mechanical. I fashioned it off the FBO plate, but there is a minor flaw in that design.. it limits the outer throw of the weights, and I've found that it's better to shorten the slots on the inside of the slots. You still limit the total travel for the advance, but it gets you out of the 'variable timing' effect of running right on the edge of start of advance at idle speed needed for larger cams... You just adjust dizzy initial to correct the new 'start' point by turning dizzy slightly in retard direction.
My mechanical was bumping around 5 to 7 degrees at idle of 800. Couldn't get a solid idle setting to stick from day to day.
By moving the inner portion of the slot, you add a small bit of tension to the weight springs, shifting the centrifugal starting point to slightly higher RPM. You want rock solid timing at whatever idle speed you need, otherwise you will be chasing your tail trying to keep a steady idle speed... any temperature or humidity changes will affect it on a daily basis. I found 30 RPM variation made a major change in advance at idle. If you can get idle down to 600 or so, watch timing, then slowly raise the RPM... when it starts shifting off a rock solid reading, that is the RPM break point for centrifugal advance.
Mine starts at 700, so I have to raise it to start around 900 or so (just above the idle the cam likes). I have not been able to find a spring set that lets me start at 950 and get a good curve to max out at 3200... so I plan on running STOCK spring set, just changing the initial starting point of the curve. That way I get good rock solid timing at idle, and also get the exponential increase as the elongated spring kicks in at higher RPM (the elongated hole spring is stiffer, so advances curve slower per RPM than the other one). This helps hold advance back mid-throttle, and helps prevent pinging in my application.
I know I could always buy a new dizzy with variable start/end points electronically, but the mechanic in me (and my wallet) say experiment!
I'm making a plate with shorter slots at inside, and taking pics to share... will do a write up for it with results when I get everything back together.
