Lots and lots of people report that their homebuilt amplifiers self destructed when powered up with the speaker disconnected. Why? Because they left out the Zobel network that is clearly shown on the TDA datasheet.
Perhaps you can consider the Zobel to be an insurance policy that protects the amplifier if the loudspeaker voice coil, or one of the connectors, or one of the solder joints, goes open circuit. Perhaps during testing by an unfamiliar repairman.
Perhaps you can consider the Zobel to be an insurance policy that protects the amplifier if the loudspeaker voice coil, or one of the connectors, or one of the solder joints, goes open circuit. Perhaps during testing by an unfamiliar repairman.
Some amps are ran on the ragged edge of oscillation and not being connected to a speaker can exacerbate the problem.
I was building a new untested amp and fired it up without a zobel. It behaved odly - the bias was all over the place. I installed the usual 100nF and 10R and it heated up pretty good. Oscope confirmed oscillation at 400kHz and was due to P-2-P capacitive feedback. I added a 33pF NP0 compensation cap across feedback cap and all was cured. This was a case where a proper PCB probably would have been fine. But the zobel was needed here and first time I saw one heat up. So I think they are a good safety net on amps with feedback. On Class A amps and zero global feedback amps they tend not to be needed or used. Don't see them on any Pass Class A amp.
38 years ago I built an amplifier based on RCA application notes. In the first instance I used 2N6472 and 2N6247 output transistors but later replaced these with some MJE150XX types thinking that faster transistors would be an improvement.
Earlier I had built the Linsley-Hood Simple Class A amplifier which does without a zobel so I deleted this when I modified the RCA design. This amplifier survived the set up test so I brought this into our lounge and briefly tried it out with music - it sounded brighter but not unpleasant - something to get used to.
Later in the evening I prepared to listen to music I turned the amplifier on and started the turntable with a disc playing.
At the far end of the lounge the TV set facing my back needed to be turned off. When I turned around to do that what I saw on the screen told me it was the amplifier that needed to be turned off.
Earlier I had built the Linsley-Hood Simple Class A amplifier which does without a zobel so I deleted this when I modified the RCA design. This amplifier survived the set up test so I brought this into our lounge and briefly tried it out with music - it sounded brighter but not unpleasant - something to get used to.
Later in the evening I prepared to listen to music I turned the amplifier on and started the turntable with a disc playing.
At the far end of the lounge the TV set facing my back needed to be turned off. When I turned around to do that what I saw on the screen told me it was the amplifier that needed to be turned off.
+1
+ The damped output inductor is not only about RF, it is definatly needed for the amp to accept capacitive loads. It is easy to understand and see in simulation. A stray cap at the output can induce an oscillation ( typically 2.5 MHz) , the inductor isolates the stray cap from the amp output.
Without this inductor, oscillations can occur at some range of cap values, not only at high values like 2uF.
Then the inductor does need damping with the resistor. Too low a resistor can induce oscillation too, because it defeats the stray cap isolation done by inductor.
This inductor is just turns of enamel wire copper, there are tons of calculators available on the internet to easily diy such inductors ( air core ).
So cheap and easy, there no reason not to have it
Current feedback doesn't always need a zobel network
any further opinons on this statement?
in my vssa vssa squarewave response it is causing a problem.
any further opinons on this statement?
in my vssa vssa squarewave response it is causing a problem.
Zobel Circuit (Impedance Stabilization)
Although speakers are rated for a certain impedance (i.e.- 4 or 8 Ohms), the actual impedance curve may vary with frequency (speakers have inductance). To compensate for non-linearity of speakers (on mainly subwoofers), Zobel circuits are used.
Speaker Crossover Calculators by V-Cap - V-Cap capacitorwww.v-cap.com › speaker-crossover-calculator
Zobel networks are a type of filter section based on the image-impedance design principle. They are named after Otto Zobel of Bell Labs, who published a much-referenced paper on image filters in 1923.
The impedance of the loudspeaker is thus typically modelled as a series resistor and inductor. A parallel circuit of a series resistor and capacitor of the correct values will form a Zobel bridge.
Note that the circuit will work just as well if the capacitor and resistor are interchanged. In this case the circuit is no longer a Zobel balanced bridge but clearly the impedance has not changed. The same circuit could have been arrived at by designing from Boucherot's minimising reactive power point of view. From this design approach there is no difference in the order of the capacitor and the resistor and Boucherot cell might be considered a more accurate description.
When used to cancel out the reactive portion of loudspeaker impedance, the design is sometimes called a Boucherot cell. In this case, only half the network is implemented as fixed components, the other half being the real and imaginary components of the loudspeaker impedance. This network is more akin to the power factor correction circuits used in electrical power distribution, hence the association with Boucherot's name.
To compensate, consider the speaker impedance, transmission line (Wires) impedance, and amplifier output impedance together.
Although speakers are rated for a certain impedance (i.e.- 4 or 8 Ohms), the actual impedance curve may vary with frequency (speakers have inductance). To compensate for non-linearity of speakers (on mainly subwoofers), Zobel circuits are used.
Speaker Crossover Calculators by V-Cap - V-Cap capacitorwww.v-cap.com › speaker-crossover-calculator
Zobel networks are a type of filter section based on the image-impedance design principle. They are named after Otto Zobel of Bell Labs, who published a much-referenced paper on image filters in 1923.
The impedance of the loudspeaker is thus typically modelled as a series resistor and inductor. A parallel circuit of a series resistor and capacitor of the correct values will form a Zobel bridge.
Note that the circuit will work just as well if the capacitor and resistor are interchanged. In this case the circuit is no longer a Zobel balanced bridge but clearly the impedance has not changed. The same circuit could have been arrived at by designing from Boucherot's minimising reactive power point of view. From this design approach there is no difference in the order of the capacitor and the resistor and Boucherot cell might be considered a more accurate description.
When used to cancel out the reactive portion of loudspeaker impedance, the design is sometimes called a Boucherot cell. In this case, only half the network is implemented as fixed components, the other half being the real and imaginary components of the loudspeaker impedance. This network is more akin to the power factor correction circuits used in electrical power distribution, hence the association with Boucherot's name.
To compensate, consider the speaker impedance, transmission line (Wires) impedance, and amplifier output impedance together.
Last edited:
Hi all of you ... my two cents experience :
Nigel7557 is right and went in the circle , an higher capacitance in the Vas will decrease bandwidth but IMPROVE HF margin as the 180° phase shift will only remain stable in the audioband... I modified many many amps (from Sansui AU101 to BOSE1800...) and finally obtained the best stability results WITH ZOBEL (any values from 22nF to 100nF and 8 to 25 ohms) but higher compensated VAS /predriver ... Don't forget : an amp looks always as a controlled - blocked oscillator . Anything you'll modify on it will dramatically MOVE phase margin poles and oscillation (any frequency) will go on . Zload (2 to 8ohms) , length of wires, are parts of the oscillator . And output inductor acts as a HFlowpass flter that will tune itself the oscillator .
The lower the BW, the better the stability ...
Hard to understand ? HAM experience ! 73
Nigel7557 is right and went in the circle , an higher capacitance in the Vas will decrease bandwidth but IMPROVE HF margin as the 180° phase shift will only remain stable in the audioband... I modified many many amps (from Sansui AU101 to BOSE1800...) and finally obtained the best stability results WITH ZOBEL (any values from 22nF to 100nF and 8 to 25 ohms) but higher compensated VAS /predriver ... Don't forget : an amp looks always as a controlled - blocked oscillator . Anything you'll modify on it will dramatically MOVE phase margin poles and oscillation (any frequency) will go on . Zload (2 to 8ohms) , length of wires, are parts of the oscillator . And output inductor acts as a HFlowpass flter that will tune itself the oscillator .
The lower the BW, the better the stability ...
Hard to understand ? HAM experience ! 73
Last edited:
This reminds me of a project I built 30 years ago based on an RCA application note which had an EF2 output stage with 5MHz output devices.
Later I replaced these with 30 MHz types thinking that would allow the zobel to be removed. I happened to have this running briefly before turning the televison on. At that point I saw interference with the reception which was not present prior to the change.
Simulations don't show the effects of radiated fields due to abrupt output device switching. While in simulations I have looked at some circuits which don't change if a zobel is removed I would not omit one if it came to an actual build.
i build quite a few nelson's amps, and none of them have zobel, as far as I remember
Most of his designs work in Class A as does the JLH simple Class A design which also lacks a zobel. Most of his designs work in Class A as does the JLH simple Class A design which also lacks a zobel. I am running the 1996 version of the latter so I appreciate what you are saying.
My reply to knutn was about BJT circuits operating in Class AB. I not aware of any circuits of this type which do not include a zobel network.
My reply to knutn was about BJT circuits operating in Class AB. I not aware of any circuits of this type which do not include a zobel network.
What has the class of operation got to do with the necessity for a Zobel? Perhaps the absence of a global nfb has much more to do with it.
An obvious example for a class AB circuit not needing a Zobel is the diamond buffer used by Dartzeel and Audio Research.
Last edited:
Zobels are wise
The series R-C network to ground from the output node of the power amplifier that we refer to as the Zobel is wise to have, costs little, and does no harm. It's primary job is to assure that there is at least some resistive load at high frequencies regardless of the speaker/cable load, or even essentially no load. The resistive component of the Zobel network, often on the order of 5-10 ohms, damps HF resonances that can be local to the output stage or that occur in the loading wiring. There are dozens of ways that an output stage can develop a parasitic oscillation, many depending on the details of wiring, and inductances, and parasitic capacitances. Too many to predict or simulate. At frequencies where oscillations may occur (thay may only be bursts on a certain portion of the waveform) the speaker cable will look like an improperly terminated transmission line. This adds another huge variable to the mix. While it is true that some amplifiers can get away without a Zobel at the output, this, in my view, is not good design practice. Just because an amplifier does not oscillate on the bench with a reasonably friendly load (including capacitive loads), gives little assurance that it will not break into parasitic oscillation bursts under some real-world conditions. In some cases, I believe that this can be a contributing factor to amplifiers sounding different when they measure about the same. I tend to believe that sometimes otherwise-good amplifiers sound different because they misbehave differently in the real world.
My recommendation is that the Zobel be very close to the output node where the emitters of the output EFs are joined by the low-value emitter resistors, and should go to a ground that has low net inductance to the collectors of the output stage. The resistive damping provided by the Zobel should be "tight" and local to the output stage. Although here it is described in the context of an EF output stage, the same precautions are probably applicable to a common-emitter or common-source output stage. You want to do all you can to minimize resonances in an output stage.
I often use distributed Zobels in my output stages. Instead of having a single Zobel that is perhaps 0.05uF and 8 ohms, and using a moderate-wattage non-inductive wirewould resistor, I may use two Zobels at physically different locations on the output line, each composed of 0.022uF and a 16-ohm metal oxide film 3W resistor, which is naturally non-inductive.
I also like to use power supply rail decoupling networks between the output stage and the drivers to filter and damp the rails at the same time. A series R of, say 10 ohms, and a shunt C of, say 1 uF (paralleled with a larger electorlytic) creates a lowpass filter at high frequencies. That is a good thing. However, in addition, it looks like a Zobel hanging off the main rail, damping resonances there.
Cheers,
Bob
The series R-C network to ground from the output node of the power amplifier that we refer to as the Zobel is wise to have, costs little, and does no harm. It's primary job is to assure that there is at least some resistive load at high frequencies regardless of the speaker/cable load, or even essentially no load. The resistive component of the Zobel network, often on the order of 5-10 ohms, damps HF resonances that can be local to the output stage or that occur in the loading wiring. There are dozens of ways that an output stage can develop a parasitic oscillation, many depending on the details of wiring, and inductances, and parasitic capacitances. Too many to predict or simulate. At frequencies where oscillations may occur (thay may only be bursts on a certain portion of the waveform) the speaker cable will look like an improperly terminated transmission line. This adds another huge variable to the mix. While it is true that some amplifiers can get away without a Zobel at the output, this, in my view, is not good design practice. Just because an amplifier does not oscillate on the bench with a reasonably friendly load (including capacitive loads), gives little assurance that it will not break into parasitic oscillation bursts under some real-world conditions. In some cases, I believe that this can be a contributing factor to amplifiers sounding different when they measure about the same. I tend to believe that sometimes otherwise-good amplifiers sound different because they misbehave differently in the real world.
My recommendation is that the Zobel be very close to the output node where the emitters of the output EFs are joined by the low-value emitter resistors, and should go to a ground that has low net inductance to the collectors of the output stage. The resistive damping provided by the Zobel should be "tight" and local to the output stage. Although here it is described in the context of an EF output stage, the same precautions are probably applicable to a common-emitter or common-source output stage. You want to do all you can to minimize resonances in an output stage.
I often use distributed Zobels in my output stages. Instead of having a single Zobel that is perhaps 0.05uF and 8 ohms, and using a moderate-wattage non-inductive wirewould resistor, I may use two Zobels at physically different locations on the output line, each composed of 0.022uF and a 16-ohm metal oxide film 3W resistor, which is naturally non-inductive.
I also like to use power supply rail decoupling networks between the output stage and the drivers to filter and damp the rails at the same time. A series R of, say 10 ohms, and a shunt C of, say 1 uF (paralleled with a larger electorlytic) creates a lowpass filter at high frequencies. That is a good thing. However, in addition, it looks like a Zobel hanging off the main rail, damping resonances there.
Cheers,
Bob
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.