Could try connecting two speakers in parallel using a single mosfet. No relay required to shut OFF.

 

The circuit in post #10 will have the speaker powered all the time unless you have a series capacitor in line with the speaker, or add a relay to remove power from the amplifier when not sounding.
The other way to go is just buy an alarm horn that incorporates the tone generator and amplifier in the speaker horn then all that is needed it to apply power to it.
EDIT: OOPs. Post #20 notes the "off" problem needing a relay. That is a comment I missed.

 

The next thing I'm going to need to deal with is how to turn this off.

You can enable/disable the circuit using two small N-MOSFETs (below):

Note that the enable signal is 0V and the disable is +5V.

 

I'm not getting the full peak-to-peak output expected. Basically about 18V p-p instead of 24V. And I'm not sure why. The 12V source for testing is a SLA battery on 14 gauge wires to the board, and reads 12.3V during testing

Measure the supply voltage directly at the MOSFETs to the MOSFET ground.
Do you have a 0.5Ω resistor in series with the supply to the MOSFETs?

 

Measure the supply voltage directly at the MOSFETs to the MOSFET ground.
Do you have a 0.5Ω resistor in series with the supply to the MOSFETs?

Voltage on the MOSFETs is 11.3V to ground. As I probe around the breadboard I'm believing more and more that the loss is due to the tiny Dupont jumper wires.

I'm happy with the result and will be committing it to copper.

Also thank you for the "enable" addition. I see that your setup uses a 5V control signal. Will 3.3V be enough to turn off (enable) the 2N7000's?

 

Note that the enable signal is 0V and the disable is +5V.

I'm looking at the plots and it looks like 5V enables the system. But based on the MOSFETS it look like the 5V should clamp the Drains to ground.

 

Will 3.3V be enough to turn off (enable) the 2N7000's?

It should, but would be better to use a lower Vgs(th) MOSFET such as the BSS123.

I'm looking at the plots and it looks like 5V enables the system.

What are you looking at?
The plots from your system output?

 

What are you looking at?

The LTspice plot in your post #23.

 

The LTspice plot in your post #23.

That clearly shows the speaker output is active when the /enable is low and inactive when the /enable is +5V (?).

 

That clearly shows the speaker output is active when the /enable is low and inactive when the /enable is +5V (?).

You are totally correct, my R/G color blindness kicked in. What I saw was the white trace going from 0 to 5 back to 0 and thought that was the /enable trace. But in fact the trace is red and not really obvious to me on the computer screen - and it should have been because the legend at the top for the trace is red. Sorry for my confusion.

 

You are totally correct, my R/G color blindness kicked in. What I saw was the white trace going from 0 to 5 back to 0 and thought that was the /enable trace. But in fact the trace is red and not really obvious to me on the computer screen - and it should have been because the legend at the top for the trace is red. Sorry for my confusion.

No problem. So the circuit works for you needs?

Voltage on the MOSFETs is 11.3V to ground.

Did you measure to the ground right at the N-MOSFETs' source pin?

 

No problem. So the circuit works for you needs?

Definitely. Really appreciate the time you've spent getting this going for me.

Did you measure to the ground right at the N-MOSFETs' source pin?

Yes, from the n-channel Source to the p-channel source pins. I really think the issue is related to the breadboard and tiny wires (still some tiny wires between the MOSFETs. Since the MOSFETs don't even get warm, the losses have to be elsewhere. I'm going ahead with committing this to copper.

Again, thank you for all of your help!

 

The most reliable way to switch off an alarm sounder is to switch off the power with an actual switch, or a relay contact. AND, probably the reduced output power level is due to connection resistance. In higher powered circuits, just a little bit of resistance can provide a voltage drop that reduces the power quite a bit.

 

Just a quick update to close this out. The circuit has been committed to copper and the voltage levels no longer reduced. I went with the relay option to turn it on & off. Working well. Thank you to all.