I've seen a number of Zobel networks [RC] in the output
of solid state amplifiers shunted to ground
I was wondering how the values are determined.
Elliot Sound Website almost always uses them with 10ohm/100nF.
The chip amps seem to use smaller resistors but the same
100nF cap. Seems like it would produce high impedance at most
frequencies and drop to close to 10ohms at some high frequency.
There also seems to be an RL in series with the output. . .seems
like it would create more output impedance (generally bad?) at
higher frequency.
Would these circuits prevent oscillation at high frequency. . .especially
if using high amounts of negative feedback?
of solid state amplifiers shunted to ground
I was wondering how the values are determined.
Elliot Sound Website almost always uses them with 10ohm/100nF.
The chip amps seem to use smaller resistors but the same
100nF cap. Seems like it would produce high impedance at most
frequencies and drop to close to 10ohms at some high frequency.
There also seems to be an RL in series with the output. . .seems
like it would create more output impedance (generally bad?) at
higher frequency.
Would these circuits prevent oscillation at high frequency. . .especially
if using high amounts of negative feedback?
Hi,
the Thiele network is there to provide a load for the amplifier at very high frequency.
Consider the component values at low frequencies.
The Zobel cap (~100nF) is effectively an opencircuit.
the series L (~2uF) is effectively a short circuit.
The result is that at audio frequencies the output of the amp passes straight through the L to the speaker and the Thiele network has little effect.
Now consider high frequencies above the audio band.
The C of the Zobel is now effectively a short circuit and this leaves the series R (~10r) as a load to ground.
The Series L is now a high impedance and the parallel R (~4r) bypasses the L.
The amplifier now has two parallel loads connected to it's output
1. the 10r + a cap.
2. the 5r feeding the speaker cables and the speaker (effectively in parallel).
The resulting High Frequency load is about 3ohms and gives the amp some lossy load to dissipate any VHF oscillations that might try to start. But this load damps them out, if the components are correctly chosen to suit the amplifier circuitry. That's the tricky bit that most builders avoid. They choose 10r+100nF and omit the inductor.
the Thiele network is there to provide a load for the amplifier at very high frequency.
Consider the component values at low frequencies.
The Zobel cap (~100nF) is effectively an opencircuit.
the series L (~2uF) is effectively a short circuit.
The result is that at audio frequencies the output of the amp passes straight through the L to the speaker and the Thiele network has little effect.
Now consider high frequencies above the audio band.
The C of the Zobel is now effectively a short circuit and this leaves the series R (~10r) as a load to ground.
The Series L is now a high impedance and the parallel R (~4r) bypasses the L.
The amplifier now has two parallel loads connected to it's output
1. the 10r + a cap.
2. the 5r feeding the speaker cables and the speaker (effectively in parallel).
The resulting High Frequency load is about 3ohms and gives the amp some lossy load to dissipate any VHF oscillations that might try to start. But this load damps them out, if the components are correctly chosen to suit the amplifier circuitry. That's the tricky bit that most builders avoid. They choose 10r+100nF and omit the inductor.
Best cap to use is self-healing polyprop. It can take a thrashing and remain reliable, like motorstart caps. I use Arcotronics, 100nF 160V.
Dielectric is not important sonically, however.
The 10R resistor in series with the cap should be a minimum of 1W, and preferably flame proof.
This network also keeps the amp loaded down at all frequencies, which is important to the phase margin, since with the inductance of voice coils their impedance increases with frequency, unloading the amp.
Cheers,
Hugh
Dielectric is not important sonically, however.
The 10R resistor in series with the cap should be a minimum of 1W, and preferably flame proof.
This network also keeps the amp loaded down at all frequencies, which is important to the phase margin, since with the inductance of voice coils their impedance increases with frequency, unloading the amp.
Cheers,
Hugh
As stated the role of the Zobel network is to supress instability, idealy the value of the resistor should equal the nominal value of the load, the cap is normaly 100n for no other reason than it seems to work best. Care should be taken with the W value of the R consider that an amplifier output of 20vRMS at 20kHz (approx 50W into 8 Ω) will dissapate .62 Watts and most instability occurs higher than 20kHz thus a 1 Watt resistor is only to be used on small amplifiers. In fact I think the reason 10Ω is popular is that it is easier to find 10Ω 5Watts than 8.2s. In the event of high frequency instability the first sign is often a smoking Zobel resistor and if it fails destruction of the output stage often follows.
Leaving out the series inductor:- You will get away with this untill you meet poorly designed crossovers the result is demise of the crossover caps and a serious increase in the heat dissipated by the amp output stage.
Leaving out the series inductor:- You will get away with this untill you meet poorly designed crossovers the result is demise of the crossover caps and a serious increase in the heat dissipated by the amp output stage.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.