Zobel Or No Zobel ?

[sakis]

ZOBEL OR NO ZOBEL ?????

AND WHY ????

some have exactly after the output of the amlifier the well known 0.1 mfd cap plus 100 R resitor to ground and then followed by one resistor in parallel with a coil .....

some others have first the network coil resistor then the 0.1mfd cap and resistor to ground

once i ve seen coil and resistor network and then from the out of the RC a 0.1 mfd to ground ....

coments about the best technic and also the best arangement between all that is very welcome ......

been working with a mosfet and seemed to play better with 0.1mfd +100R resistor to ground followed by a coil but with no resistor in paralel with it tryied 1R,2R,10R,100R..... but after all sounded better withoutany resistor ...just a coil ...any comments ????

[richie00boy]

Zobel (not zombel) is usually always required to load the output stage at HF to damp oscillatiory tendencies. Those that leave it out are either bad designers or lucky.

[sakis]

its correct zobel !!!!!

since some of my amps have zombies inside i got carried away and wrote zombel ...... thanks !!!

[Bonsai]

Even Self seems not to be able to explain precisely why the universally optimum values are 10 Ohms and 0.1uF for the Zobel network .

However, I can tell you from practical experience that you should always use a Zobel network.

It should run from the output node of the amp - so directly at the junction of the emitter degeneration resistors in a typical EF bipolar output stage to the main filter capacitor 0V junction.

Don't forget the out put inductor either - 2-5uH with 2.2 Ohm 3W resistor in parallel.

If you leave out the Zobel network, you are likely to see parasitic HF oscillations in the 2-5MHz range. If you omit the output inductor, you can expect oscillations of 200Khz to 500KHz with a capacitive load - and real world capacitive loads come from typical speaker cables and speakers themselves.

The two simple techniques - Zobel + output inductor networks - will go a long way to creating a stable, reliable amp.

[sakis]

bonsai .....

can you comment and what goes first and what after ???

like zobel first and inductor after ???

and then can you comment about inductor but with out resitor in parallel ???

[jagwap]

Yes it is good advice to add the Zobel, and the output choke.

However some designs do not need one or the other (yet to see one with neither)

Current feedback doesn't always need a zobel network, and amplifiers can be stable without a choke. Don't forget the parallel resistor on the choke if you have one, and it dampens the ringing caused by the choke into reactive loads.

Also a zobel style circuit before and after the choke is sometimes needed on MOSFET output stages as they can need additional control at high frequencies due to their speed and tendancy to parasitic oscillation.

Less is more when it come to series componenets in the signal path. But Zobels should be transparent-ish. Remember to make sure the resistor can take the power of the highest frequency the power amp input filter will pass.

[jagwap]

I'll jump in with an opinion:

The zobel can go before the choke. It can go after in the form of just a capacitor across the output if the chokes output resistor is the right value for the zobel.

How ever if you're playing with MOSFETs, slap it right next to the devices.

Without a damping resistor the choke will resonate with large capacitive loads, with huge overshoots on squarewaves. Not good if you're using electrostatics, or esoteric speaker cables.

Try a 10kHz squarewave and attach 8ohm load in parallel to 100n, then 2uF and watch the output with a choke with and without the damping resistor. The resonance gets much less, but still present.

[jagwap]

Yes it is good advice to add the Zobel, and the output choke.

However some designs do not need one or the other (yet to see one with neither)

Current feedback doesn't always need a zobel network, and amplifiers can be stable without a choke. Don't forget the parallel resistor on the choke if you have one, and it dampens the ringing caused by the choke into reactive loads.

Also a zobel style circuit before and after the choke is sometimes needed on MOSFET output stages as they can need additional control at high frequencies due to their speed and tendancy to parasitic oscillation.

Less is more when it come to series componenets in the signal path. But Zobels should be transparent-ish. Remember to make sure the resistor can take the power of the highest frequency the power amp input filter will pass.

[timpert]

Hi,

A zobel is required because the gain of your output stage is very much dependent of the load connected to it. Normally, you dimension your output stage to give appropriate gain at 8 ohms load, but in reality, the load of an actual loudspeaker (plus cable) usually is inductive at ultrasonic frequencies. Unless you have a very limited bandwidth in the output stage, the gain of the output stage will typically rise with rising frequency due to this inductivity, and as a result, the open-loop gain of your amplifier is out of control. This compromises the gain and phase margin, and may make your amp unstable.

Therefore, a zobel is included to make sure the output sees a resistive load at high frequencies, keeping the open loop gain within bounds.

The output inductor is simply there to guarantee your load going inductive at HF. If you drive (long) cables, certain filter topologies, or a piezo tweeter, the load doesn't turn inductive at high frequencies, but capacitive. Here, the zobel scheme falls apart, so you include a series inductor in the amplifier after the zobel, to keep the scheme working properly.

Finally, the resistor parallel to the L is there because, effectively, a series-resonance circuit is formed by the combination of the output inductor and a capacitive load, which forms a short at resonance. The resistor in parallel with the inductor is there to reduce the Q of this circuit, and as a result prevent the short from occurring.

The optimal values for these components require careful dimensioning, and this is not a trivial excercise. But if the values you come up with experimentally work for you, and if the amp doesn't get hotter than normal, and if you don't see an oscillation with the scope, I'd just say leave them in, close the case and have a beer...

Jurgen

[sakis]

thank you all so much about all these