Yes, conjugate networks perforce should be tuned for every loudspeaker. They also must be placed at the transducer end, otherwise the driver will get the same signal as before because most amplifiers [1] are voltage sources, and one would miss the point of nulling out impedance swings at the end of a non-ideal cable.
[1] except for a few of Papa Nelson's more amusing offerings
Agreed on 30 degrees typically being where the load is highest for subwoofers. However, having a 60 degree load phase angle means there will be a 30 degree angle at a lower impedance magnitude; if the maximum phase is 30 degrees then the worst-case EDPR will occur at a 15 degree load phase, more or less, at less stress on the amplifier. Large phase angles are an indication problems may or may not occur elsewhere.
Outside SOA considerations [2] a plate amp/subwoofer system won't really need a conjugate network because the cable length is so short.
[2] which don't apply to class-D
[1] except for a few of Papa Nelson's more amusing offerings
Agreed on 30 degrees typically being where the load is highest for subwoofers. However, having a 60 degree load phase angle means there will be a 30 degree angle at a lower impedance magnitude; if the maximum phase is 30 degrees then the worst-case EDPR will occur at a 15 degree load phase, more or less, at less stress on the amplifier. Large phase angles are an indication problems may or may not occur elsewhere.
Outside SOA considerations [2] a plate amp/subwoofer system won't really need a conjugate network because the cable length is so short.
[2] which don't apply to class-D
Class D subwoofer amplifiers often don’t have any filtering AT ALL, running the PWM straight into the voice coil.
If you want to use a conjugate impedance network inside the loudspeaker box to improve the stability of the amplifier (per the subject of this thread), you will either need to build the amplifier into the loudspeaker box or use a loudspeaker cable with a high-frequency characteristic impedance close to the loudspeaker box impedance. Loudspeakers are usually around 8 ohm, it might be difficult to get the cable characteristic impedance that low. With a coaxial construction, you would need a thick inner conductor with a thin insulation around it.
The ringing (at square-wave) shows primarily raising of the high frequency response, which can sign of instability (but not only because of). For example, see the well compensated amplifiers with output LR-filter at parallel RC-load. The RL-component and RC load make an damped LC-tank, but there isn't any instability (but there is a lot of ringing)
Last edited by a moderator:
You mean the voice-coil is the filter. With a built-in amp the class D output wires are short which reduces the EMC they can produce. For a remote amp you'd need proper filtering in the amp to prevent emissions.
In that case there is ringing at the load but not at the amp output.
I documented that in an article in Audio Amateur somewhere early 1980'-ies 😎
Jan
That is right.
A well damped zobel doesn’t ring at load either, at least for non-capacitive load.
This is my go-to solution for Zobel.
When keeping, R20 = 1/2 R22; R22 = Rload (Nominal 8 Ohm); C3 = L1 / (R22)^2, the network is critical damped.
What happens for non-resistive loads such as real speakers, where there's often a slope upwards in the highs and a considerable impedance peak somewhere near the crossover frequency?
That would be a broken amplifier. The run away high end often associated with marginal stability should only be occurring above 20 KHz, not down there where the tweeter is crossed over. If there are any frequency response anomalies near typical crossover frequencies (100 to 5 KHz) there is a serious problem with it. A little series resistance (low damping factor) will only cause mild peaking and no stability issue near the crossover frequency. The impedances at 5k have no bearing on what’s happening 50-100k. The impedances up there might.
Another thing to consider when you look at published networks is that few people seem to make an actual measurement of the inductor and losses. I remember those old Dynaco amps where the output inductor was wound around an aluminum can capacitor. Rest assured that winding a coil around an aluminum can give unpredictable results. The usual instructions of N coils of wire around a 2W resistor may have worked well in the circuit, but question whatever value they wrote on the schematic. I've built amps where the network required for stability was substantially different than the typical recommendations. Be sure the check peaks of the waveforms at all different levels, loads and frequencies before declaring it stable!
@jan.didden
I documented that in an article in Audio Amateur somewhere early 1980'-ies 😎
This post I made was a quote from another member > I did not write it. I don't know why the 'Authors detail' didn't show ???
It has something to do with transferring from one Thread to another. [ my 'navigation skills' are not tip-top ]
But I am glad it stimulated more discussion about Stability Networks 🙂
In that case there is ringing at the load but not at the amp output.
I documented that in an article in Audio Amateur somewhere early 1980'-ies 😎
This post I made was a quote from another member > I did not write it. I don't know why the 'Authors detail' didn't show ???
It has something to do with transferring from one Thread to another. [ my 'navigation skills' are not tip-top ]
But I am glad it stimulated more discussion about Stability Networks 🙂
Last edited:
Ah yes, I see it now. PM if you want adjustments.
The post box can be fiddly that way. This is part of the reason why I use the BBcode mode like the old forum did for everyone, as it gives me full editing control.
I was looking at this 'stability network' you provided, and the value of R20 @ 4.7ohms seems way too high > even high enough to effect
audio-range frequency response.
In fact, 4.7ohms would effectively 'destroy' the low frequency Damping Factor of most high-quality amplifiers 😕
I think damping only applies to low frequencies, so the 4.7 ohms doesn't matter. The DC resistance of the series choke should be very low- they're typically wound with heavy wire, so damping should be unaffected. I've never looked at higher frequency output impedance of amplifiers, but I bet it's not as low as one might think.
Hi @Audio>X , in the audio frequency range, the dominated part is the inductor. To analyze the output impedance below 20KHz, you could remove all the resistors and capacitor.
The conventional wisdom is to use any inductor between 1uH to 4.7uH. 4.7uH is about the largest value before you see any negative effects. In the case of 4.7uH, the impedance at 20KHz is about 0.6 Ohm. The damping factor would be 13 assuming the load is 8 Ohm. At 1KHz, the impedance of the inductor would be 0.03 Ohm. The damping factor would be 260 if we ignore the output impedance of the amp.
In reality, the output impedance of the amp would not be zero. Thus, the damping factor usually falls between 100 to 200 at 1KHz.
If you opt to use smaller inductor, you could keep all resistor value the same and reduce the capacitor value proportionally.
The conventional wisdom is to use any inductor between 1uH to 4.7uH. 4.7uH is about the largest value before you see any negative effects. In the case of 4.7uH, the impedance at 20KHz is about 0.6 Ohm. The damping factor would be 13 assuming the load is 8 Ohm. At 1KHz, the impedance of the inductor would be 0.03 Ohm. The damping factor would be 260 if we ignore the output impedance of the amp.
In reality, the output impedance of the amp would not be zero. Thus, the damping factor usually falls between 100 to 200 at 1KHz.
If you opt to use smaller inductor, you could keep all resistor value the same and reduce the capacitor value proportionally.
WOOPS > I must have had 'odd brain freeze' > I saw L1 as a 4.7uF capacitor 😵 [ how embarrassing ]
SORRY.
SORRY.