What attracted me to Walt's Multi-loop feedback technique was! Just prior to Walt's land mark article, I was trying various methods of lowering the open loop gain of an op amp so as to obtain the virtues of a low feedback design like is possible with discreet components. I used several methods that were being tried at that time including but not limited to operating Op amps that had comp pins as the op amp output to drive a Buffer like the AD-744, AD-829 ect. this method allowed one to bypass the op amps internal output stage, and as such all the Distortions that it produced. While this was the primary reason for doing this technique, I thought by adding a resistor placed from this connection to ground then this would lower the gain of the VAS stage in the op-amp and thus the open loop gain of the device. This method is often used on high end discreet component designs. One disadvantage of lower open loop gain is higher errors and DC offset was one big error of a direct coupled (DC) Audio amplifier operating with a minimum of negative feedback. An alternative method I tried was to place this resistor from the Compensation pin to the inverting input. this provided local feedback to improve the linearity of the op amp while it was loading down the VAS stage to reduce the open loop gain. I was getting real happy with the sound in this configuration. No current source required on the op amp low overall negative feedback, a constant amount of negative feedback across the Audio bandwidth that produced a consistent amount of THD throughout the Audio bandwidth rather than extremely low numbers in the midrange and bass with THD rising to several times it's 1KHz value by the time it got to 20 KHz. as in Conventional design. I built several line stage preamps and a couple of headphone amps upon this Topology using several different buffers and op-amps. my Audiophile friends were real impressed with the sound and two out of three were shocked to find the units Op amp based.
then one day while eating lunch and reading my latest copy of EDN (Hard Copy version) I skipped right to the Walt's tools and tips section as I always do and lo and behold hear is part one, with Walt talking about linearity in op-amps. I could not wait until the next installment because I wanted to know how the master improved the linearity of op-Amps. was there a method of doing this with less complication, how was it going be done on op amps without compensation pins to be used as outputs. inquiring minds want to know. Since the series was just that a year long series it took a while to find out. when it was all finished I was presently surprised that by simply adding one resistor to the feedback loop I could convert My attempt into Walt's topology. This also allowed op amps with out compensation pins to be used.
The primary advantages of Walt's architecture is to get all the advantages of a low feedback design, Including Low TIM (Transient intermodulation distortion) with a simple circuit based around op-amps and Buffers. since the Op-Amp is receiving it's feedback from it's own output rather than from the output stage the Phase shift of the output stage will not impact Phase margin as much as with the conventional method of one overall feedback loop. In most cases this removes the requirement of using Phase lead capacitance around the op-amp and allows the use of modest bandwidth Output stages. this opens the door to battery operation and thus allows a ultra high quality Amp with the ability to drive headphones to go on the road yet still provide sound quality you expect to get at home. Other improvements of the Multi-loop topology are not as well known but still are important. One being the fact that the Op-amp is always provided with feedback even under abusive conditions. In a conventional overall feedback loop if the output is shorted the op-amp is not receiving any feedback and will Lock up to one of the supply rails. this places a large DC voltage upon the buffers input. since the output is shorted it is sitting at 0 VDC and will cause the Transistors of the buffers input stage to become reversed biased and cause the junctions to be damaged or Destroyed. semiconductor manufactures have included input to output protection diodes to prevent this damage. however this protection is short lived and the abusive condition must be corrected or the Diodes will overheat. This will happen regardless of whether the Buffer uses short circuit current limiting or not. Another consideration is clipping or slew limiting in which case will result in the same type of problem causing long settling times.
The multi-loop topology may not result in the lowest Distortion numbers because the distortions of the output stage are not reduced as much due to less available feedback. However as I pointed out while the numbers may not be as low as in one overall loop, thay are consistent over a wide band of frequencies. Consider that a typical Op-amp buffer combo with 100% overall feedback may have 0.0001% THD @ 1KHz this will rise with frequency because the Op-amps open loop gain is declining with frequency so by the time 20 KHz. is reached the THD could be 0.1%. this wide variation in THD will result in a Bright sounding Amp with grain and other nastiness. Now consider the multi-loop situation a typical THD at 1KHz. might be 0.1% However will still be 0.1% at 50-100 KHz. because the open loop gain is still the same as it was at 1KHz. This is of coarse a typical scenario and the actual numbers may be different however the concept is the same.
