Impedance matching with emitter follower

Discussion in 'General Electronics Chat' started by Eduard Munteanu, Sep 9, 2007.

  1. Eduard Munteanu

    Thread Starter Active Member

    Sep 1, 2007
    Hello. Common-collector BJT amplifiers (aka emitter followers) are said to be good at matching impedance. But, as far as I can see, they do horrible when it comes to parameter spread.

    I have a common-emitter amplification stage, with Zout = Rc = 6k, which needs to drive a 4 or 8 ohm speaker. Adding an emitter follower immediately comes to mind, as it also provides some current gain and they're easy to make (as opposed to transformers).

    But theory says...
    Theory says that the impedances for an emitter follower are:
    Z_{in} \approx \beta_F R_E\\<br />
Z_{out} \approx {1 \over g_m} + {R_{source} \over \beta_F}
    (from Wikipedia)
    In the output impedance, the second term can be neglected if we approximate the power source as an ideal voltage source. So:
    Z_{out} \approx {1 \over g_m} = {V_{in} \over I_{out}} = {1 \over \beta_F} {V_{in} \over I_{in}} = {Z_{in} \over \beta_F} = R_E
    Looks like Z_{out} is independent of \beta_F. But Z_{in} is not!

    In the CE amplifier you had a way to select input and output impedances, as well as other parameters, to be almost independent of the transistor's characteristics. The emitter-follower seems to be awfully affected by \beta_F spread, which is very annoying if you want to match impedances for maximum power transfer. Is there anything there can be done, any hope?

    I tried adding a collector resistor. At the first glance, you might think it can limit \beta_F if you set it at Rc = {V_c \over I_{c(desired)}} = {V_{source} \over {\beta_{F(desired)} I_b}}. But that's not right, the output waveform just clips. That resistor, along with the voltage source, behaves like a (bad) current source and it won't operate the transistor linearly.

    What I think it would work might be some kind of negative feedback, as in the CE amplifier, so that you can limit the current gain, thus using enforcing an upper bound for beta, thus having a fixed input impedance.

    Any thought on this?
  2. Eduard Munteanu

    Thread Starter Active Member

    Sep 1, 2007
    Looks like Wikipedia has other means for g_m (I thought it was transconductance, as in g_m = {I_{out} \over V_{in}}). Somehow, the same problem might also arise for Z_{out}, provided they're correct, since g_m = {I_c \over V_T} = \beta {I_b \over V_T}.