A Zobel or Thiele RC to isolate the output from capacitive loads ?
It is usually said that its purpose is to provide a sufficient load at high frequencies
to the output stage, a driver being able to reach an impedance of 1 kOhm at around 1 MHz, then avoiding or reducing EMI intrusion into the amp circuit.
Sorry, I meant the output inductor, not the Zobel.
1uH with 100pF would resonate at 3.5MHz, not very far from the 5MHz reported oscillations. Which lets me wonder if the oscillations are not local to the output stage, rather than related to the global feedback loop. Under certain circumstances (here's an example), emitter or source followers may exhibit a slightly negative output impedance (the real part), which is what 1uH +1nF would need to transform the output stage in a LC oscillator.
If you use a triple bipolar output stage, adding a BC cap to the drivers may help. For MOS output stages, increasing the gate resistors could also help.
No typo.
Without the usual Zobel and Thiele (L//R) network, you have little hope of stability except on the crudest circuits.
But even with a Zobel & Thiele (eg as recommended by Self), 1 - 10n is still the most likely to incite instability. Your compensation optimisation & stability tests, whether SPICE world (.FREQ Bode, Nyquist & also .TRANS) or 'real life', need to include testing with these loads .. and also real speakers.
Try a big guitar speaker at various levels & frequencies and also when the amp is clipped both slightly & heavily.
A Zobel network IS part of the HF gain structure for the amp. Just look at the Return Ratio, with & without it, to see the effect with the LTspice Tian probe.
Physical position of the Zobel & Thiele networks is also important. Both need to be close to the output devices. But that puts the Thiele inductor close to the hugely distorted currents on the power lines.
________________________
Arthur, you might want to make your C2 = 22n at the Output Terminals.PHOENIX said:
Last edited:
Hi Arthur,
Sorry to be so slow in getting back to you on this. I've been stuffing myself with turkey and other goodies today 🙂.
You've already got some good answers. Your basic topology, using an R-C Zobel and an L-R series network, looks good and typical. However, not all amplifier circuits are completely OK with only 1uH as the inductor. Also, when using an inductor as small as 1uH, I would use a parallel resistance quite a bit smaller than 10 ohms. Keep in mind that, absent that resistor, the inductor and the load capacitance will form a series-resonant circuit that may drive the global feedback loop nuts. The shunt resistor across the inductor helps kill the Q of the series resonance. Take a look at the resonant frequency and Q of your worst-case capacitance (most oscillatory tendency) with 1uH and 10 ohms.
Try 2uH and 4.7 ohms.
The kind of output networks you can get away with depends a lot on the open-loop output impedance of the amplifier at high frequencies and how conservative the overall feedback compensation is. It also depends on the phase angle of the open-loop output impedance.
Apart from all of this, if there is a grounding or decoupling problem at HF, this sort of thing can happen with a capacitive load.
Cheers,
Bob
the sch does not show the decoupling capacitors nor where they connect to each other and crucially it does not show how the Zobel connects to the decoupling common.
The route from the Collector of the output device through the decoupling to the Zobel, through the Zobel to the output node and through the output device MUST be a low impedance route to pass MHz signals.
The "clue" connected to the ground plane may explain much if the plane is not low impedance, i.e. broken.
Hello guys,
I am going to do a couple of things and then report back. Firstly I will play with the values of the inductance and R, a I will see where that gets me.
Regards
Arthur
I am going to do a couple of things and then report back. Firstly I will play with the values of the inductance and R, a I will see where that gets me.
Regards
Arthur
That's what I thought you had in mind, I never doubted you perfectly mastered
this kind of things.
I just made the point because it is often stated that a Zobel RC helps stability with capactive loads; but it does not and it is quite difficult to understand how it should do.
The Zobel RC seems to be mandatory to power devices in follower configuration (with gain being very close to 1), even for no-NFB amps.
But it may be avoided in the few amplifiers which have an output stage with some voltage gain or from collectors/drains (not Sziklai followers).
Grounding the zobel to the ground plane is dangerous. You don't know what it spews into the ground plane and hence into an input node.
Better to star-ground the zobel to the power supply return and keep it out of anything related to the signal.
jan
The Zobel is an HF route that must return from the output node to the collector/drain of the output device/s.
This, for lowest impedance, MUST go through the local Decoupling.
Connecting to the power supply return may include many centimetres of unnecessary route length.
This, for lowest impedance, MUST go through the local Decoupling.
Connecting to the power supply return may include many centimetres of unnecessary route length.
It’s grounding the input into a ground plane that is dangerous, for the reason you state above. A ground plane has the lowest HF impedance and is perfect for ground decoupling and zobels. It’s still possible to have a ground plane and decide which point in your circuit you are going to have as your “audio ground reference” that is free of contamination. What’s important is that your feedback ground reference, input ground reference, and speaker out ground reference, are all connected to the one true “audio ground reference” in one place and one place only - i.e. star these grounds, and have a ground plane for the nasty HF stuff.
Well-stated. On my typical amplifier board, I have a couple of large electrolytics for the rails of the output transistors, at least 1000uF each. These are bypassed by smaller capacitors. Their connected center is a defined ground. It is to this ground that I connect the Zobels. I also connect the speaker return and the power supply ground to this local "star".
Cheers,
Bob
Very often the ground plane is used to 'collect' any signal ground that appears somewhere in the circuit. That means that depending on the geometry there's not a well defined 'signal ground'. If you ground a zobel to this plane, you have no control over ground currents and where 'grounds' can pick up signals from the currents.
Jan
With a careful layout, it is easy to limit signal ground to very localised areas such that a plane is not required for signal ground and it can be starred. The plane can then be used for HF-critical things like decoupling and zobels. Star grounding is not suitable for decoupling and zobels as it has high impedance at HF.
I like to pick a point on the ground plane to be the "one true audio signal ground” and to here connect speaker return and a twisted pair carrying the feedback signal and signal ground. The other end of the feedback twisted pair then connects to a small plane around the input stage, to which the input signal & ground is connected.
The only thing I haven’t figured out yet is the best way to route the ground signal for loop-gain compensation (if a loop-gain compensation configuration that involves ground is being used).
I was thinking two-pole compensation where the input stage has been cascoded. In this configuration, maximum power-supply rejection is achieved by grounding the resistor in the two-pole network (the feedback network from TIS/VAS out to TIS/VAS in, the “in” side of which is actually taken to the emitters of the input-stage cascodes).
Set it up so you can move the comp gnd to different points then pic the best one.
Why guess.
I don't believe your claim.
I believe there are ways to achieve low impedance at HF and VHF using physical traces and/or wires, rather than planes.
Andrew, with discrete wires you are always up to the unaviodable L. How can you circumvent that then?
Jan
Jan