why is there an inductor on my class ab amp output?

I have been looking at a few amplifier schematics recently and noticed two things that seem very common on class AB amps. one is a series inductor on the output, typically 1 or 1.2 uh, sometimes in parallel with a 200 ohm resistor, sometimes not.....and the second is an rc circuit to ground, ,not usual to see like .047 uf cap in series with a 10 ohm resistor to ground?

what are the purpose of these?
 
Most amplifiers become unstable into a capacitive load of sufficient capacitance, interposing an inductor can prevent this. For audio amps an inductor in the 1 to 5µH range is common, having an impedance of a fraction of an ohm at most in the audio band, but significantly higher at the frequency range of possible oscillations.


The RC Zobel network keeps the amplifier loaded resistively at these high frequencies.


Both limit the phase shifts seen at the amp output around the frequencies where it matters most.
 
Amplifiers use feedback to set the gain and reduce distortion. But a capacitive load on that feedback can make that feedback unstable and send the amplifier into destructive oscillation. The L-R series network isolates the feedback from being loaded by the speaker wires and speaker at high frequencies while having little or no effect at audio frequencies.

The R-C load must be placed before the R-L build-out and is called a Zobel network. It is required to stabilize CFP (complimentary feedback pair aka Sziklai pair) output arrangements where the driver transistor and the output transistor constitute a short feedback loop by themselves. Zobel networks are sometimes used on other circuits, sometimes for good reasons and sometimes not.

Each stage of an amplifier has a dominant capacitance (aka "Miller capacitance") that diminishes the gain -6dB/octave above a frequency called a pole on the mathematical complex number plane gain graph. The phase shift from each pole approaches 90 degrees above that frequency. A multi stage amplifier with feedback will be unstable unless one of the stages dominant pole overwhelms the loop gain so that the loop gain is less than unity before the combined phase shift from other stages makes the feedback positive rather than negative.
 
So here is my 2 cents and my understanding... take it for what it is worth from an old electronics teacher :p

warning..technical gobbledygook ahead

the reasons for them consist of 2 things.
power correction for more output power to the speakers
reduction of high frequency feeding back into the system to prevent oscillations.

power correction:

in an ac wave going though a resistor, the current and voltage phase relationship remains unchanged.
inductors cause the current to lead the voltage (inductive ractance)
capacitors do the opposite
capacitors cause the voltage to lead the current (capacitive reactance)

an ideal load is resistive. An inductive or capacitive load causes inefficiencies and in power distribution those inefficiencies are described as the power factor PF and are the difference between real power (the power that actually put to work) and apparent power (the power that is needed to be produced to get the real power out for a given reactive load)

why is this important? out of phase ractance causes unused power so if a system had an 80% power factor, it means that out of 100 watts produced only 80 is actually used to create work / mechanical motion. so the closer to resistive you get, the more efficient your system is. 100% power factor would be a load that appears purely resistive and not inductive or capacitive. (ok techies, yes i know this is not 100% accurate.. but trying to keep it simple without getting into VA and phase angles etc)

the closer you can make a load to being resistive the more efficient it is. power companies and industries spend a lot of money doing power correction to make their loads appear more resistive.

the same theories are true for amplifiers and speakers.

the rc part of this circuit is also refereed to zobel netowrk ... though less commonly but perhaps more appropriately, as a Boucherot cell. ( worth googling )

It is Named after Paul Boucherot and was developed around 1900 Boucherot cell typically consists of a resistor and capacitor in series, that is usually placed across a load for power correction and also adds stability. Boucherot's theorem is named after Paul Boucherot who developed it to explain real and apparent power.

The Zobel netowrks poneered by otto zobell are centered around image impedance and were developed at bell labs for for line losses in telephone lines providing impedance matching and filtering. his theories can also explain how the boucherot cell functions in a loudspeaker application and his work is used in other audio applications..

Paul Boucherots work centered around power factor correction and in so,, the Bouchot cell corrects the power factor by countering inductive load by adding capacitive ractance therefor making the load appear less inductive and hopefully more purely restive making the amplifier more efficient ... though the correction could go to far and make it appear capacitive.


Stability - preventing oscillation:

reactance returns energy back into the source. So correcting the reactance to be as near restive as possible will reduce this.

any energy returning beck into the system or leaking into the system can result in it oscillation, due to both feedback circuit and the sensitivity of the in phase components. Due to the high bandwidth of the amplifier circuits this can happen at very high frequencies. so steps are taken to eliminate the high frequency from feeding back into the system.

The Boucherot cell enhances stability as it dampens high frequencies reducing the chance for oscillations by also acting as a low pass filter and classic rc snubber circuit.

The Coil and resistor part are also referred to as a parasitic choke. it consists of a coil, and a parallel resistor. in short it provides a high reactive load to potential high frequencies therefor reducing the chances of high frequency oscillations and dampening high frequency ringing.

the inductance value should be high enough to provide a high impedance load to any higher frequencies the amplifier was not intended to be used at, but low enough that it does not effect the high frequency response of the amplifier under its designed impedance load.

the resistor is to make the Q lower for the coil, which in this application is a good thing.

so the reason why some say the parasitic choke is for handling capacitive loading may be that the Boucherot cell will not.... because it will bring the reactance in the direction of less inductive or more capacitive with the goal of making it as close to resistive as possible. this is fine under an inductive load....however it will do noting positive if the load is restive or capacitive as it will make the load appear more capacitive and could introduce instabilities due to that. So there are conditions where it is not effective but for most situations with standard coil driven drivers, it is fine.

the parasitic choke is a catch all for any high frequency content that might promote oscillations that is not handled by the Boucherot cell.

well i hope this helps. it is nice when someone is inquisitive and asks why something unusual is in a circuit
 
Last edited:
  • Thank You
  • Like
Reactions: 1 users
Neville Thiele once came up with an alternative: an inductor with a resistor in parallel, the resistor having a value equal to the nominal load impedance, and then a capacitor straight from the output terminal to ground. With L = C * R^2, the whole thing forms a first-order series filter together with the loudspeaker. He claimed the advantage is that it filters off RF interference picked up by the loudspeaker cable, so it can't get rectified in the amplifier.
 
hmmm....if you wanted to look at it that way then....wouldn't the rc network being closest to the load provide the most dampening of parasitic caused by the load, and the LR circuit being closest to the amplifier provide the most protection of the amplifier from renaming parasitic making there way back into the system?

it isn't just Japanese amplifiers following that layoyt...i would point out that mcintosh and nad have followed that layout off the top of my head.

i've seen where the coil has been done away with like of the top of my head adcom didn't use it.

be cool if nelson pass chimed in and set us all straight :)
 
wouldn't the rc network being closest to the load provide the most dampening of parasitic caused by the load,
and the LR circuit being closest to the amplifier provide the most protection of the amplifier from renaming
parasitic making there way back into the system?

Moving the Zobel to the speaker could allow ringing at the amplifier output.

Many do use a 100R power resistor across the speaker terminals, high enough to not need a capacitor.
 
Last edited:
Is Scoobis correct about voltage and current? I learned ELI the ICEman, where Voltage, E, leads Current, I, in an inductive circuit and Current leads Voltage in a capacitive circuit. If you think about a capacitor charging, that's current leading voltage. Yes, no?

Craig
 
Moving the Zobel to the speaker could allow ringing at the amplifier output.

sorry what i intended was within the amplifier... just the order of the lc vs rc components ... i didn;t intend to mean to move it outside the amplifier. ... sorry for the confusion

yes i totally agree, putting it at the speaker wouldn't have the desired effect for the purposes of parasitics.

people do put zobel networks at the drivers, but that is totally for making the loads resistive in nature ....
 
Last edited:
here is how i understand the capacitive load thing...

I'll preface this by saying the correct way of "isolating" captive load is by adding a series resistance. I will explain why this is.

the capacitive load most refer to for handling "isoating" capacitive load is modeled after electrostatic speakers. this is modeled / emulated by using an 8 ohm resistive load and paralleling it with 2uf of capacitance.

WARNING: DO NOT EMULATE THIS CAPACTIVIE LOAD ON YOUR AMPLIFIER AS YOU WILL RISK IT GOING INTO OSCILLATION AND DESTROY ITSELF!!!

in a purely resistive load the noise gain slope would cross the open loop gain slope at a rate of closure less then 40db/dec. This makes an amplifier stable.

adding a significant capacitance at the load causes the noise gain slope to increase by 90 degrees before it crosses the open loop gain slope and there for causes a rate of closure greater or equal to 40db/dec, which is the point the amplifier becomes unstable.

The correct way of dealing with instabilities caused by capacitive loads is to add a series resistance on the output of the amplifier that will add an -90 degree node to the noise gain slope before it crosses the slope of the open loop gain slope. The rate of closure is then less then 40db/dec which is the point the system is stable.

the use of a coil and resistor is just adding filtering to the unstable frequencies .. filtering out the unstable components before they get to the amplifier. it also ads increasing reactance counter to the capacitive reactance in the frequency domain which can be factored in. however, remember we already added a zobel network to move the reactance move capacitive to reduce the inductive loading.

why is this a coil commonly used and not a resistor ... there is power loss across a series output resistor. a coil in comparison will be little resistance loss. The resistance of a coil is perhaps 10 mOhms were a series resistance for a given solution could be 100 mOhms or higher. This introduces power loss across the series resistance that is greater then the coil aproach.

i've seen where the inductive ractance was such that it effected frequency response at low impedance loads like 4 ohms. so the design and empirical testing becomes critical in choosing an effective coil value that is as low as it can be and be effective.

empirical testing of a series resistor value is also a wise thing to do.

There is a lot more to this but then id be getting into some heavy math, which id not even know how to write on a forum.

maybe someone can clue me into why a coil is otherwise a better solution other then the power loss savings over a resistor. At this point it is the only reason i know and why i think using the right series resistance for a design is the right way to do it.
 
Last edited:
A capacitive load on an amplifier with shunt feedback at the output adds an (asymptotic) -20 dB/decade roll-off and a -90 degrees phase shift to the loop gain above a frequency that depends on the open-loop output impedance of the amplifier and the capacitance of the load. Like scoobis wrote, this tends to destabilize the amplifier.

With a series resistor, you can change that to (asymptotically) 0 dB/decade and 0 degrees above the frequency where the capacitor's impedance becomes smaller than the series resistance.

With an inductor, it changes to +20 dB/decade and +90 degrees between the frequencies where the capacitor's impedance becomes smaller than the impedance of the inductor and the frequency where the impedance of the inductor becomes greater than the open-loop output impedance of the amplifier. You have to somehow damp the resonance of the inductor and capacitor, though, for example with a shunt resistor that again reduces the asymptotic slope to 0.

All in all, you can either use a resistor or an inductor or an inductor in parallel with a resistor. Disadvantages of using only a resistor are power loss and loss of damping of the fundamental resonance of the loudspeaker.