I have a new internet radio that has a a stereo headphone jack.
I want to drive a single speaker with both channels.
What is the impedance of the speaker? Headphones range from around 30
ohms to around 600 ohms, but loudspeakers can be way less. Is there a
published minimum impedance that can be driven by this device's
headphone jack?
Will I cause harm to the amplifiers if I just tie the outputs together
and use the that to drive the speaker?
It could.
If the amplifiers are playing nearly the same signal, there won't be an
issue because the output transistors will be at the same voltage. This
situation of tracking almost exactly the same voltage means that the
left output looks like a high impedance to the right one, and vice
versa.
The correct stress test for this is: what if the left and right signal
is quite different? For instance, complete silence in one channel and
loud music in the other. In this situation, the output device with
signal swings sees the silent one as a low impedance. This is like
asking it to drive a near-zero-ohm speaker. (Basic electronics: a
voltage source is a low impedance, and a perfect voltage source
is a zero impedance!)
So, in essence the common mode signal should mix to the speaker nicely,
but any differential signal faces a short circuit between the two
different voltages.
Even if this doesn't damage the output device, are you okay not hearing
much of the differential signal out of the speaker?
Do I need to add some resistors to isolate the amps from each other
before the speaker?
It's not a bad idea just from the consideration of the possible low
impedance of the speaker compared to headphones.
Both outputs should be separated from each other by at least the minimum
required load resistance so that means idnvidually coupling them to the
speaker through resistors which are at least half that resistance.
But remember since R will shunt any differential signal, but not common
mode signal, your differential signal is attenuated. The common mode
signal goes through R and the speaker, but the differential signal goes
through R, and then another R in parallel with the speaker.
This will work best if R is significantly larger than the speaker load,
in which case you take a lot of loss of volume across R.
The best approach is to feed the channels to another device which can
properly mix them, and re-amplify to a single speaker.