What am I missing about driving this little computer speaker?

[GibsonGM]

Ok, now you're explaining it well! I was also worried that the idea was not coming across, ha ha..

Listen - you know what square wave sounds like, right?  Total crap...it IS distortion, the shape of distortion. So we have that.  It should buzzzz....at 1kHz of course.

One good experiment to do to see if your speaker is working is to play guitar or other somewhat 'pure' sine wave (ok artistic license here) into the 386, to that speaker. Be sure everything on the amp end is ok first.  Then you can delve into why they're not playing well together.   Sounds like maybe the 386 could be loading your 40106 or something, but I'm not sure.    Caps between things with DC offset is a great idea!

As far as your reported drifting of the signal...loading will do that, as would long leads, issues with coupling to itself.    Do you have a 555 oscillator around, or a simple 1 transistor sine wave oscillator like R.G.'s "quick and dirty oscillator"?  Would be good to test for comparison. 

side note, if the 386 DOES work, you can put 2 speakers in series to get the 'ideal' 8 ohms.  Will be louder, too.    Me, I'd try to use the already-there amp, I just have a thing about 386's as they tend to be pretty unstable at times.    I put the C3/R1 Zoble network right on the speaker, it seems to help my stuff better...YMMV....I put the big cap at the output before this. Won't affect your chip output here, tho.

[deadlyshart]

Ok, now you're explaining it well! I was also worried that the idea was not coming across, ha ha..

Listen - you know what square wave sounds like, right?  Total crap...it IS distortion, the shape of distortion. So we have that.  It should buzzzz....at 1kHz of course.

One good experiment to do to see if your speaker is working is to play guitar or other somewhat 'pure' sine wave (ok artistic license here) into the 386, to that speaker. Be sure everything on the amp end is ok first.  Then you can delve into why they're not playing well together.   Sounds like maybe the 386 could be loading your 40106 or something, but I'm not sure.    Caps between things with DC offset is a great idea!

As far as your reported drifting of the signal...loading will do that, as would long leads, issues with coupling to itself.    Do you have a 555 oscillator around, or a simple 1 transistor sine wave oscillator like R.G.'s "quick and dirty oscillator"?  Would be good to test for comparison. 

side note, if the 386 DOES work, you can put 2 speakers in series to get the 'ideal' 8 ohms.  Will be louder, too.    Me, I'd try to use the already-there amp, I just have a thing about 386's as they tend to be pretty unstable at times.    I put the C3/R1 Zoble network right on the speaker, it seems to help my stuff better...YMMV....I put the big cap at the output before this. Won't affect your chip output here, tho.

Hi, thanks for the reply again, I think I've finally got everything sorted out!

I googled around a bit for "CD40106 instability" and such, and found someone saying that you should always have a ~100uF cap across the power rails with it, because CMOS devices tend to be very sensitive to that stuff. So I did that and now I have a beautiful, noise-free square wave! Also, it sounds good going through the LM386 (though I have to first chop the signal down pretty far, divider with 2Mohm/10kohm resistors), so I'm guessing that problem was due to it just being a really messy signal going in...

thanks, I grabbed the other speaker so I'll probably put them in series like you said!

The last thing (not a big deal though) is that the pot for the 386 does this really strange thing where, when I touch it to change the level, it buzzes like crazy, and I can actually hear the radio on the speaker when my fingers are close to/touching the knob! So it'll sound fine, amplifying my square wave, I'll go to turn it up, it'll buzz like crazy, but when I take my hand away it's back to normal. It's weird to me because I tested for continuity and the knob part isn't actually connected to any of the pot leads... so I'm guessing that I'm creating a big capacitance somewhere by touching it? either way, not a big deal.

thanks for all the help!!

[GibsonGM]

Great to hear!  Yes, they can be buggy.   Extra decoupling (your 100u cap...often with a .1u in parallel with it) is a great idea.

The pot is working with your body to make a capacitor. If it's mounted in a metal shield that is grounded, that often goes away!