here it is:
NOTES FROM THE TEST BENCH
By Bruce Hofer
An externally hosted image should be here but it was not working when we last tested it.
In the design of high-performance amplifiers and integrators, some analog designers are beginning to explore the virtues of cascading two op-amps running on different power supply rails within the same overall feedback loop. This architecture inherently eliminates two subtle forms of distortion: (1) input stage susceptibility to power supply interference caused by the non-linear load currents of the output stage; and (2) thermal modulation of input stage offset caused by power dissipation changes in output stage. These two forms of op-amp non-linearity are very real, and not the least bit insignificant.When properly designed, the dual op-amp architecture can deliver amazingly low residual THD, typically below -140 dB! Indeed, it is the design approach I took when developing many of the analog circuits in the Audio Precision APx555. However, placing two op-amps within the same overall feedback loop is extremely challenging and can easily lead to instability. A practical circuit design must deal with component tolerances and lot-to-lot variations in op-amp high frequency behavior (which tends to be poorly specified in the first place). I would have to characterize these designs as being among the most difficult I have ever developed in my 46-year career.
Large signal overload behavior and transient effects during power-on also had to be considered. During the evaluation phase of the first APx555 prototypes, we noticed a higher failure rate of the output op-amp within these dual op-amp architectures. It turned out the supply rails that powered the output op-amp were turning on slower than the rails powering the input op-amp. For a brief period of time at turn-on, the input op-amp was forward-biasing junctions inside the input pins of the second op-amp with its slower-responding power supplies. During this fault, the input op-amp was effectively sourcing the supply current for the output op-amp, including the charging currents for its decoupling capacitors! These currents peaked over 40 mA exceeding the 10 mA absolute maximum rating of the op-amp thus causing catastrophic failures.
The solution was obvious, but not easy...add a current-limiting resistor between the two op-amps. Unfortunately the added resistor caused some extra phase shift within the feedback loop that required additional iterations in the design. So it goes with state-of-the-art engineering....
-Bruce Hofer
Audio Precision co-founder and Chief Analog Engineer
.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.