Computing Peak Currents into Loudspeakers

I believe these notes are intended for engineering, not end users. The engineers will design amps to handle these transients. I'd stick with the rule of matching amp impedance handling to speaker impedance, 4=4

 

If it was published that the 4 ohm minimum amp was stable down to 2 ohms for transients the average user (90%) would try hooking up a 2 ohm load because they said it would do it. If you can't understand engineering principals, don't read them they are not meant for you.

 

Try the Java Applet "A/C Response of Inductor"

http://www.falstad.com/circuit

A speaker in a box is a much more complicated circuit but this should give you a basic idea of AC circuits

 

The statements are more about the configuration of a system and the ratings of amplifiers. This all an effort to give as much headroom and safety margin as possible to the system so that failures are kept to an absolute minimum. I really wouldn't read very much more into it than that. As the author stated, "This is an extreme requirement, and not many systems have been designed to satisfy it.”

 

Papers like this are intended to formalize the design process and define the conditions that their products should be used under. Some interpret these boundaries as only a guideline, like a suggested speed limit posted on a highway. Others throw them out the window in favor of pushing the limits and squeezing all the tone out of a system that they can. There's a balance between product reliability and a long service life or a different type of sound. The guys in the white coats tend to be conservative but they can't always imagine how products will be used.

When a musical instrument speaker cabinet with a specified nominal impedance is connected to a matching tap on a tube amp, they are matched at a steady or idle state with no input. Apply an input and the operating point shifts plus and minus about that point. That is the standard that is used. When an AC input is applied, all bets are off. So a speaker is selected that can take peaks that the amps will deliver, and the amp's output will be able to cope with the impedance swings of the speaker. That change in impedance also affects and steady state operating point of the amp. As does the swings of the power supply. Everything interacts. The operating limits or design guidelines are needed as a reference point. But it's a complex system.

There's also a lot a variation. Some amps and speaker cabinets perform right at their limits, others are more tolerant to extending their intended limits. For example, some amps that are rated for 8 ohms can operate comfortably and sound good with a 4 ohm cab connected. Some can't, it depends on the design. When you connect a 2 ohm cab onto a 4 ohm tap on that same amp, it might not perform as well because the 2 ohm cab can swing dangerously low with frequency. That's why general rules of thumb don't always apply when mismatching cabs with amps. People that know the amps (components and design specs) and speakers can predict how far the boundaries can be pushed. Papers from manufacturers, set those boundaries and warn that they shouldn't be exceeded.

 

Whoa, a little bit of information can be very, very misleading when it comes to real world applications.

There are several different mechanisms being addressed, but not necessarily correctly identified in the detail necessary for understanding.

1. There is peak current limits, which in the old days with fairly smallish amps could be an issue with some semiconductors and some designs. For example, there were some popular power transistors with decent power dissipation specs of ~150 watts but limited to ~7 amps Ic (collector current). In an amp with say 100 watts into 2 ohms, you may be able to get by with 2 parallel pairs for power dissipation (actually SOA or safe operating areas) but be pretty darn shy on Ic at 14 amps as that would be a power output of (14 x .707)**2/2 or 50 watts so clearly we are short in the Ic department. In this case, as a designer, we would in fact need more semiconductor for current's sake.

2. There is a dissipation limit as well, which is the voltage drop across the power device (rail - audio) which becomes a huge problem as rail voltage gets higher, so the number of semiconductors required to meet the SOA requirements might be 5 or 6 per rail, the Ic has plenty of margin. As this rated power becomes even greater, it makes sense to use class G or H which limits dissipation and also decreases the number of devices that share the current but this is fine sinec the more modern designs use modern semis that have an Ic of maybe 15-20 amps each and 3 of them would be adequate to support (45 x .707)**2/2 = 500 watts and 4 of them would be good for 900 watts.

Since current is not much of a problem to deliver these days, especially with class D (which does away with the SOA issue completely), being wary of JBL's "sub-nominal impedance" notes is not as big of a deal now as it was then. Generally, with quality amp designs, all of this is taken into account within the design, and specification of output stage. Where it does rear it's ugly head is with products that are designed to such a low price point that there is no budget to add silicon (semiconductors) so it becomes "convenient" to slide the current limiting down so that the sub-nominal impedance areas are limited... which prevents magic smoke BUT (and this is a big, critical "but") since current and voltage are not in phase with a reactive speaker load, when the design senses on current but acts on voltage (typical VI limiter), the clamping action creates a nasty artifact called "current clipping" where the clipping occurs on the side of the waveform, roughtly 90 degrees lagging to the voltage. This may be one reason for the presumed negativity of driving older, and less expensive solid state amps hard and into VI limiting. Good designs provide adequate margin between traditional voltage clipping into a reactive load and the threshold of VI limiting.

For a refresher, I just looked at the output stage of the GBE-1200, there are 6 output transistors per swing at 16 amps each, so the output stage is capable of (96 x .707)**2/2 or 2300 watts even though it's rated at 1200 watts "RMS" at the audio out. Each device is also rated at 200 watts dissipation, 1200 watts x 2 (2400 watts) available which fits nicely into the SOA margin. These devices are used in a great many high powered pro audio amps (which is the area I came from) and represent pretty much the state of the art bipolar devices. It's unlikely that we will see much more development in this type of device as the large power amps are becoming almost exclusively class D in nature (including those that I have been working with at this power level)

Note that the 1100 watt margin is for a variety of factors such as thermal derating, sub-nominal impedance derating, VI limiter margin, etc. Also note that I have taken the Ic and converted from peak to RMS to reflect comparisons to rated power in RMS units.

I have greatly simplified the math, specifically omitting the SOA calculations as thay are confusing, difficult and time consuming. The GBE-1200 is also a bit unusual in that it has a 2 ohm impedance mode switch which specifically increases the available SOA for driving sub-nominal 2 ohm rated loads.