06-06-2015, 11:25 AM
Yes, I was planning to come back to the loudspeaker driver section later.
I was going to suggest using the transistor as an emitter follower to get around some of the problem. But I was planning to take that a stage further, using a PNP transistor in addition to the NPN, to make a push-pull stage. Obviously no need for biasing into Class B, as it's being fed with a square wave; a class C job with no biasing would work just fine.
That's because this simple stage has a big problem - the cone is only being moved in one direction from its rest position. Or, to put that another way, the signal has a DC offset. That's a very wasteful way to use a loudspeaker.
Obviously, the junction of the two output transistors would connect to the loudspeaker via a capacitor, so it doesn't matter what "state" IC2 assumes when silenced. (I could knock up a diagram later, but I'm about to go out)
I was also going to say that although they are cheap and plentiful, the 2N3055 is OTT for this. Bearing in mind that the highest current delivered to an 8 ohm load will be comfortably less than 2A, I'd consider a TIP31 or similar. Smaller package, easier to deal with. The TIP32 is the PNP compliment. Of course, there are countless alternatives. In operation, the voltage drop across these will be nice and low, so they won't get hot, but I'd provide for a small heatsink on the PCB, just in case.
Assuming this is for a siren that you want to be as loud as possible, I'd carefully investigate the loudspeaker in question. Many have strong peaks at points in their frequency response range, and it makes sense to tune the oscillator so that you are on one of these peaks. If the peak is 10dB up relative to another, nearby frequency - which is not unlikely with a cheap loudspeaker - then that's a subjective doubling in loudness for no extra power. Along the same lines, it makes sense to be where the human hearing system is most sensitive - which is somewhere in the 3 to 4kHz region. So, I'd allow for a pre-set resistor to determine the frequency of IC2.
You might also want to investigate piezo sounders. These generally come with a tuned acoustic system that produces a strong peak at a certain frequency. And, they take almost no current in operation - you might even be able to do away with the driver transistors. They are very, very efficient compared to moving coil loudspeakers - you just need to have plenty of voltage available.
But ultimately, I guess it all depends on what you're trying to achieve. If the goal is to make a really loud siren for a particular application, then hopefully the above will help (although frankly, I'd probably just buy something ready-made). But if the real objective is to have some fun messing about with some electronics - and that's a highly worthwhile goal in my opinion - then I'd drop the transistor and just drive the loudspeaker from the 555, with a resistor in series to limit the sound level (say, 100 ohms), and a series capacitor (say, 10 to 100uF), and also use a high impedance loudspeaker - the sort that are sold for exactly this sort of fun
Finally, I took another look at the original post. The circuit is lacking a smoothing capacitor. The 555s are decoupled, but the feed to the loudspeaker is not. This means that the gated square wave will be modulated by 100Hz ripple. This may or may not be intentional, or indeed desirable (good reason to breadboard it first, of course).
I was going to suggest using the transistor as an emitter follower to get around some of the problem. But I was planning to take that a stage further, using a PNP transistor in addition to the NPN, to make a push-pull stage. Obviously no need for biasing into Class B, as it's being fed with a square wave; a class C job with no biasing would work just fine.
That's because this simple stage has a big problem - the cone is only being moved in one direction from its rest position. Or, to put that another way, the signal has a DC offset. That's a very wasteful way to use a loudspeaker.
Obviously, the junction of the two output transistors would connect to the loudspeaker via a capacitor, so it doesn't matter what "state" IC2 assumes when silenced. (I could knock up a diagram later, but I'm about to go out)
I was also going to say that although they are cheap and plentiful, the 2N3055 is OTT for this. Bearing in mind that the highest current delivered to an 8 ohm load will be comfortably less than 2A, I'd consider a TIP31 or similar. Smaller package, easier to deal with. The TIP32 is the PNP compliment. Of course, there are countless alternatives. In operation, the voltage drop across these will be nice and low, so they won't get hot, but I'd provide for a small heatsink on the PCB, just in case.
Assuming this is for a siren that you want to be as loud as possible, I'd carefully investigate the loudspeaker in question. Many have strong peaks at points in their frequency response range, and it makes sense to tune the oscillator so that you are on one of these peaks. If the peak is 10dB up relative to another, nearby frequency - which is not unlikely with a cheap loudspeaker - then that's a subjective doubling in loudness for no extra power. Along the same lines, it makes sense to be where the human hearing system is most sensitive - which is somewhere in the 3 to 4kHz region. So, I'd allow for a pre-set resistor to determine the frequency of IC2.
You might also want to investigate piezo sounders. These generally come with a tuned acoustic system that produces a strong peak at a certain frequency. And, they take almost no current in operation - you might even be able to do away with the driver transistors. They are very, very efficient compared to moving coil loudspeakers - you just need to have plenty of voltage available.
But ultimately, I guess it all depends on what you're trying to achieve. If the goal is to make a really loud siren for a particular application, then hopefully the above will help (although frankly, I'd probably just buy something ready-made). But if the real objective is to have some fun messing about with some electronics - and that's a highly worthwhile goal in my opinion - then I'd drop the transistor and just drive the loudspeaker from the 555, with a resistor in series to limit the sound level (say, 100 ohms), and a series capacitor (say, 10 to 100uF), and also use a high impedance loudspeaker - the sort that are sold for exactly this sort of fun
Finally, I took another look at the original post. The circuit is lacking a smoothing capacitor. The 555s are decoupled, but the feed to the loudspeaker is not. This means that the gated square wave will be modulated by 100Hz ripple. This may or may not be intentional, or indeed desirable (good reason to breadboard it first, of course).
