You made one big mistake, which is to not put diodes in the lines from zone 1 and zone 2. When zone 3 goes on, it is now back-driving zones 1 and 2. Whether that matters and what kind of damage that could cause depends on the circuit. Apparently you got lucky, since zones 1 and 2 apparently still work.
Perhaps your system is driving the solenoids with AC. In that case the diodes are only letting power get to the solenoids every other half-cycle, instead of every half-cycle when the full AC is applied. That could possibly cause the solenoids to move enough to appear to work, but also to vibrate noticably at the power line frequency (60 Hz in the US, for example).
If the issue is AC, then there are ways this can be addressed. However, it makes sense to get more information about your system before going into details that could be totally irrelevant.
Added:
Now that is seems clear the problem is that the solenoids are driven with AC, we can talk about ways to get what you want within that framework. One way to do this is to put a full wave bridge after each zone output. That makes it DC instead of AC. The solenoids will still work fine on this rectified AC.
Now that you essentially have DC output from the zone controllers, do what you tried to do before, but this time do it right. The safe thing to do is to put a diode between each zone output and solenoid. If I understand your setup right, zone 1 would drive solenoid A thru a diode (after the full wave bridge, consider those part of the zone outputs now), zone 2 would drive B thru a diode, and zone 3 would drive A and B each thru separate diodes.
Added 2:
Here is a schematic of what I was referring to above:
Note the full wave bridge immediately after each zone controller output. That is D1, D2, D3, and D4 for the zone 1 output, for example. Each valve driven by each zone is then isolated with another diode. These are D9, D10, and D11 for zone 1, for example.
With this level of diode isolation, the same valve solenoid can be driven from multiple zones. For example, valve 1 could be connected both below D10 and D14 without those connections causing shorts or back driving one zone when the valve is driven by another zone output.
Best Answer
Since this sounde like a one-off and therefore a few $ more once shouldn't matter that much, two relays would solve this problem easily. Get two DPST (minimum) relays with 24 VAC coils. The existing drive circuitry turns on the relays instead of the valves, which will be indistinguishable to it. Each relay then turns on its zone valve with one pole and the existing common valve with the other pole. In other words, the common valve will be separately enabled if either of the new zone valves are on. All this will happen transparently to the controller if hooked up right. The controller will only see a additional zone.
This is not cheap or elegant, but just two relays and a few wires and you're done.