small speaker/amp kit?

You can always use the venerable 386 amp chip, but you'd be reinventing
the wheel and the sound still wouldn't be any good. *Better sound comes
from speakers in an enclosure.* So, buy some $10 portable amplified
speakers. Wire your power supply to where the batteries are supposed to
go. Clip off the mini plug and solder directly to your circuit. You may
find a 386, wired in gain-of-50 configuration, to be a perfect pre-amp.
Use DC coupling between your circuit and the amp/preamp (a small 0.1
disc cap).

Here's one to start you off, and it's only $5. Such a deal.

http://www.allelectronics.com/cgi-bin/item/NGA-4/580400/POCKET_AUDIO_SPEAKER_.html

-- Gordon

True, but I have no room for an enclosure in this bot, nor do I care
much about the quality of the sound.  I just need it to be audible.


Hmm, that's an interesting idea.  I'll give that some thought.


Oi.  If I have to build a 386 amp circuit, I'd much rather just connect
that to the 1" speaker I already have.
 

That *is* a good deal -- even if I don't end up using it in a robot,
it's exactly what I've been looking for for listening to podcasts on my
Palm in the car.  Thanks for pointing it out!

Best,
- Joe

In that case you want to use a piezo speaker, though at 3.3 volts you
can't expect much. Drive it with as high a voltage as you can; you can
use the NPN transistor to level convert from 3.3 to 5 volts. A PM
speaker won't move much volume of air you need unless it's in an
enclosure. One can't change the laws of physics.

-- Gordon

Sounds like something Scotty would say.

OK, a transistor circuit would be a good exercise for me.  But, for the
sake of argument, would an optoisolator work?  Suppose I connect the
3.3V MCU output to the "control" side of the optoisolator, and run the
unregulated 9V (motor) supply through the "controlled" side (making up
terms here since I don't know the proper ones), and connect that to my
speaker?

It seems like an optoisolator should be able to operate even at
thousands of Hz, and they have the advantages of (1) helping to keep my
noisy power line separated from my MCU, and (2) being much easier for me
to understand.  :)

Thanks,
- Joe

An optoisolator has as its output an open collector transistor, so the
overall level-shifting effect is about the same. However, choose wisely
(as the knight said to Indiana Jones); some/most opto-isolators are
meant to be driven with digital signals, not analog. The phototransistor
is not biased linearly.

Many of the old Forrest Mims Radio Shack books show transistor-based
level shifting circuits. They are about the easiest things to wire up,
and no more involved (actually less so) than an opto-isolator.

-- Gordon

Good point.  I'm still not clear on whether the FREQOUT command on the
ARMmite actually generates a real sine wave... the docs claim it's so,
but everywhere else in the same docs, these pins are described as
digital outputs.

But Don has suggested that I'll get better results using PWM
(square-wave) output anyway, and that should work fine with an
optoisolator, if I understand correctly.


I'm sure that's true when you understand them, or are working from a
schematic made by someone who does.  I'm still at the "which of these is
the collector, and what's that do again?" stage.  But you've encouraged
me to give it a try -- it'll be good to have a specific goal for delving
(not for the first time!) into transistors.

Thanks again,
- Joe

Well, I'll wrap up this thread with a success story: I took a simple
1-transistor circuit (from David Cook's introduction to motor drivers!)
and used it to drive my speaker (switching the motor supply).  After
much fiddling and fixing of stupid mistakes, it worked!  The sound from
the speaker isn't what you'd call loud, but it's certainly audible
enough for my purposes.  (And the motor batteries are rather low --
about 7.5 volts, while they were 9.6V when full, so I imagine it'll be
louder on a full charge.)

Incidentally, I also tried it on the piezo element I had on hand.  This
turns out to not be a speaker, I believe -- it just emits a high-pitched
(and obnoxiously loud) tone.  Probably meant as an alarm element or some
such.

Cheers,
- Joe

Sounds like it is an "internal drive" type, which means it generates a
fixed tone. Sonalarms were originally built to emit a shriek loud
enough to get you out of bed in the middle of the night from the other
room. Driving a typical "external-drive" type piezo with a 5v mcu pin
will generally not emit an excessively loud tone.

There is no such thing as a "real" sinewave coming out of a
microcontroller chip, unless it has a DAC built-in, and I don't think
any mcus have this. So, they will use either the PWM method I
mentioned before, or possibly DDS ...

http://www.google.com/custom?q=direct+digital+synthesis

Don Lancaster also used to talk about somethign similar in magic
sinewaves ...

http://www.google.com/custom?&q=don+lancaster+magic+sinewaves

Hi Joe. If your speaker is a typical 8-ohm jobber, the mcu will not
drive it, since the typical output resistance of a CMOS mcu pin is on
the order of 150 to 200-ohms or so. OTOH, you may be able to find a
150-ohm speaker.

Also, the mcu will directly drive a common piezo tranducer [external
drive type], usually with adequate volume, and/or a magnetic
transducer [20-mA type]. In the latter case, it should be wired
through a series 10-uF or so electrolytic cap, and have a diode [wired
upsidedown] in parallel with the transducer coil to squelch inductive
spikes.

You will get less volume if the mcu is 3.3v. Also, if you're using the
Armexpress for this, you will get **significantly** better volume if
you use a stream of square pulses to the transducer, rather by using
the ARMbasic FREQOUT function.

See also ...

http://www.google.com/custom?q=lm386+kit

That's interesting.  It actually did drive it, just not very loud.  (And
on Dennis Clark's suggestion, I tried putting about 200 ohms in series
with it, and it still worked, but was even softer.)

The speaker I'm using has no useful markings on it, but I don't think
it's a piezo, because its backside is strongly magnetic (suggests a
regular speaker cone, no?).


That sounds good; I really like simple for this.  Something like one of
these?   http://www.mouser.com/kobitone/


Why is that?  Not that I don't believe you, I'm just curious.  FREQOUT
is supposed to generate sine waves, which I'd expect to sound better
than square waves.  If it's actually able to generate sine waves on a
digital output pin... the docs are a bit vague on that, and I haven't
got an oscilloscope.


That's very helpful, thank you!  But if I can accomplish my modest goal
with a directly connected piezo transducer, I'll be all the happier.

Thanks,
- Joe

Wrong direction. Had you driven it with an npn inverter you'd be
farther ahead. But just take your DMM and measure the dc resistance to
get an idea what your speaker is.




Yeah, without looking. I buy kobitones, and use both piezo and
magnetic as indicated last time..



They "sound" better than squarewaves, but that doesn't mean they have
as much energy, ie volume.

I didn't measure the FREQOUT signal on my scope, but just tried it
both ways with the Armexpress.

Usually they generate such sweet-sounding "sinewaves" by using PWM
[squarewaves] at a high enough frequency the speaker mechanics will
filter out the PWM fundamental, and then rapidly vary the PWM duty-
cycle to produce a "sinewave". I expect if you were to look at the
output using a spectrum analyzer, you'd see most of the actual energy
is at the PWM fundamental frequency, and is therefore wasted, and much
less energy is at the sinewave frequency peak.

Even simpler and smaller-A simple single transistor common emitter
amp,
(5 parts total) will have that thing more than audiable. cost less
than a buck,
and can be 3,5 or 12 volt powered and take up about an inch square.
Sure its re-inventing ,but not much.
Mark