Zobel network

[Valvebloke]

I think part of the problem here is that the original precise definition of what a Zobel network is has been blurred over the years as people have taken to using the phrase to describe almost any circuit designed 'to help a bit' with an awkward load impedance. A true Zobel network is a complex thing designed to present a fixed input impedance to whatever is driving it. Of course in practice the meaning of the word 'fixed' is open to debate. Some people will only regard the impedance as fixed if the variations in it are very small indeed. At the other extreme there are people who are happy to see any reduction in the impedance variation. But for the circuit to be a true Zobel network I would argue that the variation has to be reduced a great deal, and over a very wide frequency range. You may feel differently of course. Without any agreement on how well the job needs to be done I don't think we can give a usefully general answer to the above questions about component values etc.

VB

[SergeAuckland]

What he said! So the resistor would just go across the terminals, pos to neg?
If I use a ribbon, I would xover no lower than 3kHz, (24dB/octave, active) but the Shackman could go down to 500Hz, allegedly, so a different cap for each. I haven't yet worked out yet what value I would use in each case to block LF, it would have to be low enough to not interfere with the xover slope. Ribbons, or planars, tend to be about 6-7 ohms and fairly flat, but the Shackman is transformer coupled. Pretty poorly IMO, so if it was worth persevering with I would probably buy something better, maybe Sowter, as the originals drop to about 1 ohm at 20kHz! I would put a couple of ohms in series with it to start with.

I think part of the problem here is that the original precise definition of what a Zobel network is has been blurred over the years as people have taken to using the phrase to describe almost any circuit designed 'to help a bit' with an awkward load impedance. A true Zobel network is a complex thing designed to present a fixed input impedance to whatever is driving it. Of course in practice the meaning of the word 'fixed' is open to debate. Some people will only regard the impedance as fixed if the variations in it are very small indeed. At the other extreme there are people who are happy to see any reduction in the impedance variation. But for the circuit to be a true Zobel network I would argue that the variation has to be reduced a great deal, and over a very wide frequency range. You may feel differently of course. Without any agreement on how well the job needs to be done I don't think we can give a usefully general answer to the above questions about component values etc.

VB

I would be very wary of using just a single series capacitor to block the LF from a ribbon tweeter. These things are pretty fragile, and I would much prefer at least a 12dB/octave rolloff, better still 24dB/octave. Ditto with the electrostatic tweeter.

I also completely agree about the misuse of the term "zobel network". This has come to mean anything attached to the output of an amplifier to provide a load at HF. As VB says, proper zobel networks are far more complicated. KEF had something like a proper one on their 104.2 loudspeakers, which maintained a flat 4 ohm impedance with no more than 0.5 ohm variation from LF right through to extreme HF. KEF called it Conjugate Load Matching, but in effects it's a "proper" zobel network.

S.

[awkwardbydesign]

I would be very wary of using just a single series capacitor to block the LF from a ribbon tweeter. These things are pretty fragile, and I would much prefer at least a 12dB/octave rolloff, better still 24dB/octave. Ditto with the electrostatic tweeter.

I am using a 24dB/octave active crossover; the series cap would just be to block any residual LF rubbish from the valve amp. The 'static panels are designed(?) to work down to 500Hz anyway; I wouldn't use them anywhere near that low, I just mentioned it as an indication that they shouldn't need as small a cap as ribbons/planars.
BTW Serge, although I might sometimes make tongue-in-cheek jokes about your fondness for measurements, I do appreciate the time you take to help. With good nature, too.

[SergeAuckland]

I am using a 24dB/octave active crossover; the series cap would just be to block any residual LF rubbish from the valve amp. The 'static panels are designed(?) to work down to 500Hz anyway; I wouldn't use them anywhere near that low, I just mentioned it as an indication that they shouldn't need as small a cap as ribbons/planars.
BTW Serge, although I might sometimes make tongue-in-cheek jokes about your fondness for measurements, I do appreciate the time you take to help. With good nature, too.

Perfect! No issues there at all.

Thanks also for the kind words. I try my best to be helpful.

S.