I'm OK with digital electronics, fully capable of popping a 2n2222 and a suitable base resistor onto a GPIO pin but analogue is still quite a mystery to me, hence I'd appreciate some design advice with a much beefier output stage to drive a 5w 8ohm speaker. I already have a 3W 4ohm speaker driven from 12v through a TIP102 and it makes a reasonable din, but two things concern me:
a) putting too much current through the speaker coil. I think I may already be doing this - I checked a "speaker power" calculator site and it looks like I'm shoving 4.5W through it, given that there is no base resistor on the TIP102 (my signal source is already limited to 3mA)
b) being too cautious and limiting the drive so it's not as loud as it could be...it's for an alarm that I'd like to hear throughout my house.

Basic parameters: The signal is out of a UM3561 which is controlled by a Wemos D1. The spec sheet shows the max output current is 3mA and my toy 'scope shows it to be a square wave 0v - 1v with a varying duty cycle (digital FM? - great name for a radio station...). I have easy access to 3.3v 5v and 12v. My parts bin has a few TIP102s and several logic level MOSFETs and a bucket full of small-signal BJTs)

I have some notion that I need to bias the TIP102 to be in its best operating region somewhere around the midpoint (i.e. 6v and "adding" the signal to it so it fluctuates between e.g. 5.5 to 6.5 volts. Am I talking nonsense and/or is there a better way?

There's also a slight hitch: The UM3561 has a low side MOSFET switch whose gate is tied to a GPIO pin so that I can power it on and off under program control. BUT if I connect the output direct to the base of the TIP102 the Vdd find its way through to ground and thus the alarm is always on, albeit at much reduced volume. I put a 1uF cap in series to fix the problem but I don't know what the optimum value should be...very low pF values stop any sound at all and higher uF lead to unacceptably long delays as the cap charges...

So, what I'm after is circuit that will get the most sound out of the speaker without popping it and using the parts I already have, i.e. I don't wan't the complication of an op-amp etc - I'm not after fidelity or low distortion I want a loud noise to scare burglars.

Here's the broad outline: and the one final simple question: How do I improve my output stage to get the most racket out of a 5W 8ohm speaker?



Thanks in advance for any advice

 

Remove the series capacitor C1 and replace it with a series resistor. Experiment with the value of the resistor (10kΩ) for best sound quality.

If your volume is too high, insert a high wattage power resistor (2-10W) in series with the loudspeaker. Experiment with the value of the resistance (10-33Ω).

 

Won't this cause problems by restricting the already feeble 3mA output form the UM i.e. I end up with 0.1mA? Happy with the concept of the hefty series resistor ...but still need advice on how to safely get the max current through the coil...

 

UM3561 can output 3mA.
We are only taking 0.3mA into a 10kΩ resistor.
Your TIP102 NPN darlington will have a minimum current gain of 1000, i.e. putting out 0.3A to the speaker.
A 22Ω 3W resistor in series with the speaker will drop the volume and save your speaker from burnout.

 

I would remove your low-side switch at pin-2.
Connect pin-2 to GND.

Put a diode, anode to pin-3.
Ground the cathode to turn off the alarm.

 

You do not have an audio amplifier, instead your simple unbiased Darlington is switching DC into the loudspeaker that will burn it out and/or hammer it to death. A loudspeaker is driven with AC, not DC.

 

You do not have an audio amplifier, instead your simple unbiased Darlington is switching DC into the loudspeaker that will burn it out and/or hammer it to death. A loudspeaker is driven with AC, not DC.

I know. That's why I asked the question. It's why I said "I have some notion that I need to bias the TIP102 ". So other than pointing this out, do have a working solution, which is what I came here for?

 

@Mr Chips. Thanks for your helpful suggestions - I will try them out and let you know the result, but still after a cheap / easy audio output stage within my parameters.

 

Note to self: Must look up definition of 'audio output stage' - this thing sure does make noise ...and is identical in principle to the recommended circuit in the datasheet...apart from throwing out more power than a bog-standard 2904 or 2222."

As I also mentioned already, I don't want HiFi or zero distortion, just loud noise.

Helpful and constructive replies still needed

 

I suspect that your UM3561 siren generator is outputting square waves. There is no need for a linear low distortion audio amplifier.

To avoid over heating your speaker and wasting power you want to make sure that the current through the speaker is reduced to zero when the no sound is required.

 

I suspect that your UM3561 siren generator is outputting square waves. There is no need for a linear low distortion audio amplifier.

To avoid over heating your speaker and wasting power you want to make sure that the current through the speaker is reduced to zero when the no sound is required.

It is indeed "The spec sheet shows the max output current is 3mA and my toy 'scope shows it to be a square wave 0v - 1v with a varying duty cycle "

"you want to make sure that the current through the speaker is reduced to zero when the no sound is required."..that's kind of what I asking for a circuit example to show me how to do, as transistor biasing is a black art to me...

 

A loudspeaker is designed for an AC input so that its voice coil and cone moves back and forth (alternating). Your single output transistor drives the speaker with DC one-direction pulses which are not AC. Biasing the transistor will not help.
AC means Alternating current.
A loudspeaker is driven by an amplifier output that has at least two transistors that operate push-pull.

You are using a 12V power supply like used in a car. There are many car radio amplifier ICs available that actually have two amplifiers for a speaker, one amplifier output goes in the opposite directions of the other amplifier and each amplifier drives each wire of a speaker so that the voltage swing is doubled which causes the speaker current to also be doubled which gives almost 4 times the power of a single amplifier.

You get burglars but I get racoons. I wonder if your siren idea will drive away racoons or instead get the police at my door for disturbing the peace.

 

You get raccoons, I get deer, coypu and wild boar...hoping it will scare some of those. Not sure my nearest neighbour who is 1km up the road will be calling the gendarmes on me...I want it loud, but not THAT loud!

Are you saying I need F-V converter or filter network to convert the DC to AC first? I was hoping to avoid additional complication...

 

You cannot convert your DC pulses to AC. For AC you need two output transistors. One transistor pushes down and the other transistor pulls up. Lookup Push Pull Transistors in Google. A dual polarity power supply is not needed when the amplifier has an output capacitor driving the speaker.

 

How many watts to you want? You can use Google to find amplifiers over a wide range or output powers.

For example this is the first one I came up with a simple search for a 10 watt audio amplifier circuit. There are many more and in different configurations to suite your taste.

 

From the original post " output stage to drive a 5w 8ohm speaker "

also from the original post "So, what I'm after is circuit that will get the most sound out of the speaker without popping it and using the parts I already have, i.e. I don't wan't the complication of an op-amp etc "

 

The datasheet for the UM3561 shows a resistor from its pin 3 output to drive a transistor, not a capacitor.

 

What parts do you have that you want to make into an amplifier?

 

The datasheet of the siren IC shows a 3V supply and a little 2SC9013 transistor (driving the speaker) fed a very small base current through a 10k resistor. The datasheet for the 2SC9013 is from India and says it is used for "potable" radios. Then the transistor is safe to drink! Maybe the speaker is piezo like used in a greeting card.

 

Sounds like all he needs is one speaker lead connected through a capacitor to a complimentary pair driven by the loaded collector of the 2SC9013, and the other speaker lead grounded. He doesn't really need an audio amplifier, just something to drive the speakers.