This system requires very fine tuning, this is not achievable by the user. So the amp must be located into the loudspeaker cabinet, and temperature compensation of the voice coil resistance is mandatory. Resulting overdamping leads to lack of low frequencies, and this also must be compensated for.
Final results are impressive, but in fact the overall result is totally ruined by acoustical resonances of the listening room...
Finally I have decided to not further investigate this way, because normal damping give correct results if the speaker is made with sufficiently strong magnet.
Pierre.
Last edited:
Ironically, optimal speaker performance is not contingent on maximum damping factor, but instead it is dependent on a properly matched damping factor.
A power amp with for example significant negative output impedance, which is not most speakers are designed for, can substantially harm sound quality.
We are in agreement. The rail voltage and internal pass element dissipation are of no "interest" to the speaker, how the amp does it's thing is hidden from view.
The concepts are the important thing here, and the VI space is key to some of the questions. (V being vertical axis, I being horiz.
Many here already know the following...
Q1. What exactly is happening when damping factor is measured via external stimulus at the output node?
A1. The output node is being pushed horizontally in it's VI space. The output node remains at zero, but the pass elements are reacting to the node current to keep it at zero volts.
Q2. Is this consistent with control of a resistive load?
A2. No. A resistive load exercises the VI space with a line through 0,0. This line has a of slope R.
Q3. Is this consistent with a reactive load?
A3. No. When the output node zero crosses voltage, the output current is NOT zero. The pass elements are not at rest, but there is a current which is attempting to keep the node at zero voltage despite being required to dischage an energy storage mechanism at the load. This is a phase lag (ELI) or lead (ICE), as in ELI the ICE man..
edit: as a quick aside.. the term "back EMF" applies to all reactive loads. It is not a consequence of non linearity. It is taught in circuit theory either first or second semester. It is primarily thought of as inductive in nature, but it is more consistent to think of it as reactance. That way, capacitive loads which present ellipses on the VI space are considered as well. Remember, the phase measure of a speaker is also indicative of the capacitive or inductive nature of the load. At resonance, a zero crossing of the phase response indicates a transition from a reactance of inductive to a reactance of capacitance.
Also of note is, what the amp sees at the terminals is a consequence of the speaker coil resistance, inductance, as well as all the acoustic storage mechanisms, from sealed baffle, horn, out to 4th or 6th order bandpass. Speaker measurements which are of a steady state nature do NOT indicate what will really happen when non periodic signals are applied. One such example is the impedance peak of the woof at LF. That resonance requires multiple cycles to reach that specific impedance.
Q4. What does a reactive load look like in VI space.
A4. An ellipse.
Now for the real questions. (yes, there was a point)
Q5. When an amplifier is providing lots of low frequency drive to a reactive load, the output VI space runs through all four quadrants. Is the damping factor identical within all four quadrants?
A5. Who knows? I do not know if it has ever been tested.
Q6. When an amplifier is providing lots of LF into reactance, will the damping factor presented to a secondary hf signal external drive signal be the same as that of a singular signal from the amp? Remember, the simple measure of damping factor is a horizontal exercising of the VI space, and the ellipse forces damping factor to be measured at different locations in the VI space.
A6. Who knows?
Q7. In all quadrants of the VI space, will the damping factor be the same for a hf reactive load superimposed on the ellipse of a lf reactive load.
A7. Who knows?
The last question is actually a descriptor of a simple two way speaker.
I gave Jon Risch the test setup design for all of this back on March 29, 2007 for use by Peavy if he so desired. I do not know if anything was done with the information.
Concur on both points. I've also embellished so that all can follow, not just a few.
John
Last edited:
John is that really true? If I 'push' a current into the output of an amp, I can measure a voltage deviating from zero. In fact that is how I measure Zout. So the amp output node is exercising both I and V space when damping a speaker, no?
Jan
The output node is being pushed horizontally by the stimulus.
You are measuring how much vertical results.
It is of course a chicken/egg sentence.
For good amps, the stimulus is pretty much horizontal, but you are correct in your thinking.edit: my statement was more of a global one to distinguish it from what happens when the amp is outputting a voltage.
edit: The amp is trying to maintain the voltage despite what the current is doing. For a pure resistance, it's easy, as the pass element forcing the voltage is also the only one handling current. Reactive loads bring the other side into play.
John
Last edited:
It took a certain voltage to make current flow into the UUT, no?
Damping factor involves one or more impedances, which implies changes in both V & I.
Both the stimulus and the response are in the real world, diagonal lines.
Damping factor is commonly defined using only real numbers, and it would take redefining it in complex numbers to answer the question. If we do that, we will find that if the circuits are linear, then quotients of circuit parameters such as damping factor will be constant.
That all said, we know from power circuit engineering that circuits with reactive elements, currents, and voltages all hurt efficiency. They are therefore avoided as much as possible.
Again, it takes an understanding of linear and nonlinear systems. If the speaker has constant parameters as a function of frequency (almost never true) and is linear at all frequencies (a little more true, but still untrue) then the damping factor will be the same.
Please see the answer to Q6.
That is documented here: http://www.diyaudio.com/forums/solid-state/8752-reactive-load-tests.html#post2792395
I don't think that a Walt Jung "Sign Changer" is what is needed here
http://waltjung.org/PDFs/Some_Timeless_IFDs_EDAAI_062298.pdf
I suspect that something more along the line of a variable high pass might be a better solution.
If the damping factor is sufficiently small then it doesn't matter if it varies a bit. I would expect that it does vary, but not by much unless the amp is approaching a voltage or current limit. I suspect that naive current limiter design might be a problem for some amps, as some people can't help assuming (at least subconsciously) an almost resistive load.jneutron said:
Does it matter? Damping factor is mainly about LF. However, forgetting about damping and instead thinking of frequency response this would seem to be a good mechanism for frequency-dependent IM to occur. It matters.
It provides a specific function, so is required for that specific function.
That does not control the signals entirely. The design is for flexibility. There are various ways to skin this cat, I provided a universal one with the most independent handles for the test engineer. If some functions are not desired, it's easy enough to change.
In reality, using actual inductances and resistances is trivial enough, it just requires changing out parts to exercise various locations in VI space. It changes the nulling out of the LF content for HF damping measure, but can be done.
Curious, G mail says march 29, your link has me saying march 26. Weird. AHA, the 29th was when I sent the jpg, 26th was the first date of the e-mail correspondence...duh..my error back then.
You've hit the crux of one of my points. Output stage protection is not a simple current limit, and foldback protection can be an issue if a 3 way speaker with 3 independent reactances is pushing the envelope in the reactive quadrants. Foldback being the primary method of protecting pass elements from reactive loads.
The assumption that damping factor is independent of the quadrant of operation is not a supported one by test AFAIK. Amplifier measurements using a resistive load does not duplicate the four quadrants of operation. Remember, Q2 and 4 are sinking signal into the rail opposite the output voltage level. That is not a configuration typically assessed during distortion tests that I am aware of. So yes, frequency dependent IM during four quadrant operation is a concern.
The assumption that damping factor is only useful for LF is also not a good one. I would suspect that some researcher may have done work to that end, I've no idea.
Achieving a higher damping factor will actually push the limits of the outputs even more, I could see a designer compromising SOA even more to get high, if unrealistic, damping factors.
John
That is quite difficult for a class B audio amp driving some generic 3 way speaker.
Easy with power systems.
You just stated that given A is the same (but almost never) and B is the same, then C is the same.. Logically, you just said that damping factor will almost never be the same. Is that what you meant to say?
The assumption that a generic damping factor as a result of current driven measurement is the same as that during 4 quadrant operation is not supported by any linearity argument, especially given that the 4 quads alternate sink and source paths. Perhaps a class A amp, but not what has been discussed in this thread. Nobody's made that distinction yet..That can no longer be said..
John
another decades old topic treated in the linear power amp articles of JAES 70's-80's
"tug-o-war" puts another power amp at the ground end of the load R, feed it with any amplitude, relative phase you want to draw any ellipse on the test amp output I-V
you could think of the far end amp as the source of the "back emf"
there is a small difference from a actual complex load in the load effect on loop gain - very low for high global feedback amps over audio frequency
in sim I've also used a relatively prime frequency to tug the output of the tested amp over a larger region
the sim isn't real world but the results seem to follow Cherry - the "interface distortion" distortions are generally no higher than that of the same amp delivering the same current into a resistive load
"tug-o-war" puts another power amp at the ground end of the load R, feed it with any amplitude, relative phase you want to draw any ellipse on the test amp output I-V
you could think of the far end amp as the source of the "back emf"
there is a small difference from a actual complex load in the load effect on loop gain - very low for high global feedback amps over audio frequency
in sim I've also used a relatively prime frequency to tug the output of the tested amp over a larger region
the sim isn't real world but the results seem to follow Cherry - the "interface distortion" distortions are generally no higher than that of the same amp delivering the same current into a resistive load
Last edited:
That is not the sign changer I mentioned. As I stated in the jpeg, it is in the third edition of his IC opamp cookbook, page 281.
John
There is no room for irony... Speakers are designed by the makers for the average damping factor of the most common amp on the market
But superior performance can be attained with overdamping, providing that adequate amplitude/frequency correction is introduced at the input of the amp. Otherwise some lack of low frequencies is noticeable. All of this is well known, and very easy to compute.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.