The speaker crowd have been discussing the merits of avoiding a load like this, and wondering why so many have done this blindly and apparently not had a problem.
What problems may happen and how would the uninitiated tell? What impedance threshold/frequency range should be considered risky? Are some amplifier types more susceptible?
What problems may happen and how would the uninitiated tell? What impedance threshold/frequency range should be considered risky? Are some amplifier types more susceptible?
Last edited:
A capacitor across the output of an amplifier will degrade the phase margin, possibly by enough to turn the amplifier into an oscillator. This will burn out the amplifier and the tweeter.
Amplifiers with high negative feedback are especially vulnerable. Most amplifiers have an output inductor to prevent a capacitive load from degrading the phase margin by too much.
As a hobbyist amplifier designer, I feel that a speaker's impedance should never drop below 4 ohms at any frequency. Lower impedances will not be handled well, especially if the load is reactive (feeding the amplifier's energy back into the amplifier).
Ed
Amplifiers with high negative feedback are especially vulnerable. Most amplifiers have an output inductor to prevent a capacitive load from degrading the phase margin by too much.
As a hobbyist amplifier designer, I feel that a speaker's impedance should never drop below 4 ohms at any frequency. Lower impedances will not be handled well, especially if the load is reactive (feeding the amplifier's energy back into the amplifier).
Ed
The issue is easy to explain: in an amplifier with NFB, as the frequency increases, the open loop gain decreases. Thus the closed loop internal resistance looking back from the load, increases. Just as an inductor does. Ergo, with a range of certain capacitive loads; the amp may turn into an oscillator. If the output stage can't support the power at the frequency of oscillation, simply it explodes.
NO.
Just use the proper parallel R-L network inline with the load. One does not need the extra 4 ohms at lower frequency, just up high.
Just use the proper parallel R-L network inline with the load. One does not need the extra 4 ohms at lower frequency, just up high.
An electromagnetic speaker's voice coil has enough DC resistance to avoid problems. The designer needs to avoid introducing a low-impedance path in the crossover (the 330nF capacitor in post #1).
Electrostatic speakers are capacitors and need an amplifier designed to drive a capacitor. This is when the output inductor becomes mandatory.
There is an unfortunate school of thought in speaker cables that try to match speaker impedances by being extremely high in capacitance. That is unnecessary and impossible. Don't buy such cables.
Ed
Electrostatic speakers are capacitors and need an amplifier designed to drive a capacitor. This is when the output inductor becomes mandatory.
There is an unfortunate school of thought in speaker cables that try to match speaker impedances by being extremely high in capacitance. That is unnecessary and impossible. Don't buy such cables.
Ed
Perhaps @hautparlurker was suggesting putting a resistor in series with the 330nF capacitor? That's not a problem surely?
Shunting the series inductor with a small cap is usually done in order to trap a breakup peak, and if you use a resistor to control the depth of the notch, you'd get a side benefit of avoiding the supersonic capacitive load. No? Here's an example from Troels. (Although the resistor isn't actually necessary from the point of view of the load, in this example, as the only shunt capacitor is in the compensation network, so there's a resistor in series with it already).
Last edited:
Adding R to the notch cap will widen the frequency band over which it notches, and reduces the depth of the notch. Which may or may not be what you want to flatten the woofer.
Using the proper R-L at the amp’s output generally solves this problem. The impedance might still be fairly low, but as long as there is enough real part to stabilize the amp and low energy at the upper frequency range it should be able to deal with it.
Using the proper R-L at the amp’s output generally solves this problem. The impedance might still be fairly low, but as long as there is enough real part to stabilize the amp and low energy at the upper frequency range it should be able to deal with it.
The .22uF cap tunes the inductor and rejects the frequency at which they resonates (10KHz aprox) and some adjacent ones depending on Q. It is a high impedance in series with tweeter and thus blocks this frequency from appearing at tweeter's coil.
The resistors in post #9 are large enough to avoid creating a low-impedance path at high frequencies.
Ed
Ed
So, if I throw a resistor large enough, it should be alright then?
Couldn't really achieve the response I was after, it created a knee. I'll try different designs with using a 3rd order filter.
The resistor will preclude low impedance, but there are better ways to design the crossover.hautparlurker said:
Ed
That might be the inductor resistances I used (0.5 for the large and almost nothing for the small). The point is the same.. there are different ways to get the same response even when it gives a different impedance.
I can't really get the response I want. Looks like I am going to add another notch filter for the woofer break up, and build a very very crowded crossover. I still don't understand how one of the most famous builds doesn't create any amp burning problems but anyway...
For a speaker builder who doesn't understand the electrical implications for the amp (i.e., who would have though that the filter in post #1 was perfectly fine)...what do you mean by "proper R-L"
But it is not just that example...I can post the schematics of dozens and dozens of designs that use the filter in post #1.
Okay EdGr, but what if someone wanted to take the easy way out and just add the resistor in series anyways (like ianbo's post #9)? I found a thread back in 2014-ish that suggested using a resistor about the magnitude of the impedance of the driver, so typically a 4- or 8-ohm resistor in series. Any suggestion on a rule of thumb that would, or would not, work?