Stop and take a deep breath. You seem to have gone through a number of changes to your circuit, and it's hard to tell exactly what you've ended up with. Let's assume that you currently have

simulate this circuit – Schematic created using CircuitLab
Now. First things first. Get yourself a cheap DMM. No argurments, no backtalk, just do it. Without one you really have no way to tell what's happening in your circuit.
Next, get rid of your 9 volt battery. Again, no backtalk. Switch back to your 5 volt supply. The problem is that, with a 9 volt supply, both Q1 and Q3 will be at least partly turned on, no matter what the state of the control inputs. This means that when you turn on either NPN that side will be shorted to ground, and your battery voltage will drop very badly.
I recommend putting an LED and a 1k resistor from 5 volts to ground, just to help you see if the power supply is getting overloaded. If it dims, you know something is pulling down the battery. And if something is pulling down your battery, now that you have a DMM you can figure out what it is, rather than guessing.
If you want to try your 9 volt battery, you need to buffer your inputs so as to drive them to 9 volts when high. You can do this (for one half of the circuit, by

But if you try this make sure you replace both halves of the circuit. I only showed one side - the other should be identical.