The Right Half Plane ZeroA story from the field…IntroductionKevin [email protected]• Design Engineer for National Semiconductor in Longmont• 15 years in the industry, IBM/Hitachi, NSC• Just a “regular guy”…Some Review- an only somewhat related questionWhat is the current through the NMOS?Some ReviewSame Question-- What is the current through the NMOS?Some Review=Is everyone comfortable with this configuration ?Some Review-- You’ve seen it before…Q: So How Does it Work? How is the DC point determined at Out?A: Negative Feedback…1927 – Negative Feedback Amplifier InventedHarold Black (ATT)Interesting Articles:http://en.wikipedia.org/wiki/Harold_Stephen_Blackhttp://ieeexplore.ieee.org/iel1/37/5926/00229565.pdf?arnumber=229565&htry=31Negative Feedback…“It is the invention of negative feedback which makes highly linear amplifiers possible. Negative feedback basically works by sacrificing gain for higher linearity (or in other words, smaller distortion). By sacrificing gain, it also has an additional effect of increasing the bandwidth of the amplifier. However, a negative feedback amplifier can be unstable such that it may oscillate. “http://en.wikipedia.org/wiki/Harold_Stephen_BlackMore Review-- OpAmp Compensation• What is it?• Why do we need to do it?• Does the required compensation change depending on the actual circuit in which the opamp is used?More Review-- Compensation• We usually compensate IC Amps with a Capacitor, -- could it be done with an Inductor?Which node would you pick? Why?More Review-- CompensationOften times Miller Compensation is used…More Review-- Compensation• Something called a Right Half Plane Zero can show up in the overall amplifier transfer function when using Miller compensationThis can cause two primary problems1) Transient Response Oddities (large signal)2) Stability Issues (small signal)Complications arise…Transient OdditiesWhy??Stability Issues…(Phase Margin)Superposition can be used to show how a single RC network can contribute 180 degree’s worth of phase shift!Tame the FeedForwardPath...Numerous Methods…Nulling Resistor Compensation BufferMany More…The Compensation BufferOnly Two Additional Devices – NO feedforward path (RHPZ) to worry about! Right??Why did I choose to do it this way???Sometimes you choose a topology that minimizes your engineering effort-maybe…--process variation etcThe Compensation BufferThe Compensation BufferDC Simulation ResultsThe Compensation BufferLoopGain Simulation ResultsLoopGain can be defined different ways- Here I am using –(CU Class Loop Gain).-- I show the gain around the entire loop.BUT Trouble… Falling Edge Transient- Found in Hardware…The Compensation BufferImportant Engineering Skill – Learn how to handle your own mistakes!Other Chips showed this…The Compensation BufferIt is “good” when simulation predicts reality! The Compensation BufferBack to Simulation…The Compensation BufferWhy??? What might happen when we decide to drive Vout DownVoutVSolution• 20x Tail CurrentTo SF- keep follower ONThe Compensation