JG, if you take the impedance curve of an actual loudspeaker, it will contain within it all of the reactance, kickback, what-have-you. It's then pretty easy to put together an equivalent circuit with the same reactive characteristics. You can even simulate things like voice coil heating and cabinet resonances, though the effects of the latter are pretty darn small compared to the fundamental resonances.
A few minutes with some experimental impedance curves and SPICE will give you all the dummy loads you could ever want. Some speaker software packages will do this modeling for you.
A few minutes with some experimental impedance curves and SPICE will give you all the dummy loads you could ever want. Some speaker software packages will do this modeling for you.
My WATT 1's are the hardest load that I have. .5 ohm at 2KHz. Try that sometime!
The next worse load is the big electrostatic panels with 5uF capacitance loading or more. That is what the Parasound JC-1 is designed to drive, and it does it very well, except that Charles Hansen's amp drives them just as well or better. Maybe some here should take note.
The next worse load is the big electrostatic panels with 5uF capacitance loading or more. That is what the Parasound JC-1 is designed to drive, and it does it very well, except that Charles Hansen's amp drives them just as well or better. Maybe some here should take note.
john curl said:Much of the research has been done originally by Otala, dismissed by Cordell, and shunned by the AES. You are too late Joshua_G, you have to look back at the OLD knowledge of 30 years ago.
Though I didn't see the research by Otala, I'm quite sure I didn't "invent" anything here …
Sure I can. So can anyone with an impedance curve in hand and a computer.
My speakers were large panel ESLs, very easy to model- mostly a big ol' cap as John said. My current speakers have a very flat impedance curve so the model isn't very interesting.
99% of closed box speakers can be closely simulated with an LCR parallel circuit in series with a small resistor and an inductor. You can add more components to fine tune it (e.g., for the tweeter rise which isn't purely inductive), but this simple one will get you 90% of the way there.
My speakers were large panel ESLs, very easy to model- mostly a big ol' cap as John said. My current speakers have a very flat impedance curve so the model isn't very interesting.
99% of closed box speakers can be closely simulated with an LCR parallel circuit in series with a small resistor and an inductor. You can add more components to fine tune it (e.g., for the tweeter rise which isn't purely inductive), but this simple one will get you 90% of the way there.
Joshua_G, you are going to run up against a 'dead end'. This is because the 'models' are only partial, and they won't show you much. Of course, a PURE inductance or capacitance will make almost any amp cry for help, but you can't model everything in between. We, real amp designers, have addressed this for decades. The biggest problems are protection circuitry firing too often, current limiting, and breakdown because of operation in a forbidden region.
john curl said:
Hi John,
You are quite correct about the loudspeaker and how it acts. Indeed, I covered this in detail in my IIM paper long ago - using SPICE to model the loudspeaker, no less. The high currents that can be generated under certain conditions are real. We don't disagree on this.
Where we do disagree is in regard to the correction signal that is required to keep the output node from wiggling in response. That signal going back does not create a problem, at least insofar as Otala's definition and measurement method for IIM. I have measured real amplifiers for IIM and also shown theoretically that IIM does not increase when the NFB is responsible for the low closed-loop output impedance. I think your intuition is just getting the best of you here.
Perhaps you are referring to some other kind of measurable anomoly that occurs under these conditions as a result of that small correction signal. If so, it may not conform to Otala's measurement definition of IIM, but I would still be interested in learning about it.
What measurements have you done to confirm your assertion that that correction signal causes a problem.
Do you agree that IIM should be measurable, as Otala asserted?
BTW, I am sure your amplifier has very little IIM, whether you have measured it or not. There is nothing wrong with having a low open-loop output impedance; it is usually sufficient, but not necessary for low IIM.
Cheers,
Bob
- Status
- Not open for further replies.