Dxvideo said:
... However, I haven’t seen any "unity gain" current feedback amplifier... Are you sure that?
Yes current feedback amplifiers can be used unity gain - but you need to keep a Rf resistor in the circuit - not just short -in and out like a voltage feedback op amp
in a cfa the feedback resistance seen at -in determines the dominant pole of the loop, that is why they can be so fast "optimum decompensation" is built into the circuit by the reduction in feedback Z as the gain is increased with constant Rf
by using larger value Rf resistor cfa op amps can be "overcompensated" and work fine at unity gain
see the TPA6120 data sheet fig 27
from the identical chip THS6012 data sheet:
"As with all current feedback amplifiers, the bandwidth of the THS6012 is an inversely proportional function of the
value of the feedback resistor. This can be seen from Figure 17 through Figure 20. The recommended resistors
with a ±15 V power supply for the optimum frequency response with a 25 Ω load system are 680 Ω for a gain = 1
and 620 Ω for a gain = 2 or -1. Additionally, using a ±5 V power supply, it is recommended that a 1-k Ω feedback
resistor be used for a gain of 1 and a 820 Ω feedback resistor be used for a gain of 2 or -1. These should be
used as a starting point and once optimum values are found, 1% tolerance resistors should be used to maintain
frequency response characteristics."
This isn't just theory, the TPA6120 is excellent in unity gain application because of its large input common mode range which allowed me to use 1/2 of the TPA op amp in my Class A amp output stage in unity gain:
http://www.head-fi.org/forums/f6/ad8397-class-188758/index2.html#post2263321
and a few more jcx posts in that thread; one (#16) shows the voltage feedback version of the Class A bias circuit I built with the TPA and 1.5K Rf instead of the feedback short circuit shown in the sim's U2 Lt1128 vfa unity gain follower circuit
slight over compensation improves overshoot/stability into C loads but these op amps need a load isolating impedance between their output and any significant C - see the 10 Ω series Ro in the TPA eval circuit- I used a lossy bead core EMI suppressor/ inductor to get near zero output Z at audio frequency