Joe Rasmussen said:
Hi Joe,
The trouble I have with the Olson hard amp concept is that if there is clipping or other non-linear behaviour like switching from clA to clAB or B, the output impedance will also be undefined. For instance, at clipping, the output level cannot increase so the output impedance as seen be the speaker looks like infinite. This is no different than with an amp with say .1 ohms Zout. So what's all the fuss about?
Jan Didden
Definitely !!!!
But Olson's proposal is also an extreme case either. Me personally, I prefer reasonably balanced approaches. I.E. neither the amp with 0.000000001 Ohms output impedance nor the one with 6 Ohms.
An amp that has an output impedance that is consistent (i.e. independant of frequency, output voltage and output current) would be my choice.
It's output impedance can easily be taken into account when designing the box for a given target Qtc, as mentioned by Joe.
Regards
Charles
roddyama said:
Hi Rodd
That largely puts the finger on it: The ability of the amp to absorb back EMF is what is supposed to be behind the supposed logic of damping factor. But this thinking has a serious flaw:
The fact is that ALL amplifiers are poor in this respect because of the DCR or Re of the voice coil. Think about it. This severely limits peak current that can be absorbed back into the amp's output stage since it has go through anything from 3-4 Ohm (of a typical 4 Ohm speaker) or more. In effect, there is always a large value DC resistor beween speaker and amp!
Really, and this is an oft misunderstood concept, it is primarily the Qe of the driver, no matter what its Re is, that is the source of damping, not the amp. What the amp must do, or really not do, is erode that Qe. So ultimately it's the driver itself and the box alignment that dertermines damping, transient response etc. I know this because I've put in a lot of hard yards on this topic, also been able to consult some fairly enlighten sources on this. I might also add that there was a time when I didn't quite get it myself.
Now I am by no means relegating the importance of the amp. The amp must still cope with considerable back EMF, not so much for damping purposes but at least some. But the amp must be designed to have good bandwidth and not rely on feedback to get low Z out. Make the amp as open-loop linear as possible and the bass of a well designed speaker will be just fine.
We haven't even mentioned the potential danger of back EMF getting into the feedback loop of the amp... ?
Joe R.
Hi,Joe Rasmussen said:
I don't think so. IMHO, almost ALL speakers are poor because have coils with relatively high and temperature dependent resistance.
Someone can find very good and short "DF problem" explanation on Thrue Audio site
http://www.trueaudio.com/post_013.htm
Author is John L. Murphy.
Regards
So for someone just getting into the technical stuff (only general college physics many moons ago), do I understand this all to mean that; if the impedence of the speaker system is fairly constant with respect to frequency and the output impedence of the amp is constant with respect to frequency, and the Q of the speaker system with a given amp is where it should be, then the absolute impedence of amp don't much matter? Or, put another way, it's the system as a whole that determines the outcome?
FWIW my pp ultralinear amp has an output impedence of 6 ohms but sounds just fine with my speakers. Very tight solid base to my ears.
Sheldon
FWIW my pp ultralinear amp has an output impedence of 6 ohms but sounds just fine with my speakers. Very tight solid base to my ears.
Sheldon
Sheldon said:
Hi Sheldon.
Basically a pretty good summation. Congratulations, you seem to have gotten something that took me years to appreciate.
If you, as you do in your case, have 6 Ohm output Z and a bass alignment that copes with it, then the so-called DF means nothing. What the higher Z does is erode the Qe of the bass driver (not the total Qtc as some think), shifting the bass alignment to a different point, but one that may still sound audibly acceptable. In your case DF means nothing... your speaker still works, right, in fact your DF may well be less than 1 or no DF at all!!! What more proof is needed?
I still want to make clear the point that one of the two most imminent researchers, one of the founding fathers and pioneers of Thiele-Small parameters, the one and only Richard Small makes it abundantly clear that Qe (which is the dominant component of Qts) can be tuned by varying Re.
That meant DF was of no consequence to him. I argued with him, only later to realise how right he was!!!
Now how can you tune Qe? Add series resistance. That was what Small said !!! Just include output Z as part of the addition.
This also demonstrates that output Z now becomes part of Re, indeed, this may be a difficult concept to handle, but Re is really a loop. Low output Z is part of Re and vice versa, they are inseparable. The only thing to add is that output Z closes the loop but still has a value, but it can be several Ohms. DF at several Ohms, what kind of DF is that?
------------------------------------------------------------------------
Let us look at maths that don't lie, Let us set up a series of values and analyse the results of changing Re in its various incarnations.
8 Ohm (nominal) driver
6 Ohm DC Resistance or Re of voice coil
Qe = .4 (this is the key value that gets altered)
Qm = 5
Qts is calculated 1/(1/Qe+1/Qm) = 0.37
This value 0.37 is based on there being no added series resistance or Z. In fact it is the same as if the terminal were shortened. Re must become a loop or part of it.
Now what happens if we have an amp:
Output Z = 3 Ohm
This changes Qe proportionally, so:
New Qe = (Re+3)/Re times Qe = 0.60
Yes, Re has increased 50% from 6 to 9, and also Qe has increased 50% from 0.4 to 0.6 .
There is a proportional relationship between Qe and Re. If Re is increased by 50% then Qe also get increased by 50%.
Now we can recalculate Qts using formula 1/(1/Qe+1/Qm) = 0.54
That is a 46% increase in Qts.
BTW, We can deduce from this that speakers having low Qm (or high mechanical damping) are less affected by added Re than a high Qm driver.
Now these maths don't lie. What you have to ask is this: What effect does it have on the speaker that you have changed Qts from 0.37 to 0.54 - in case of some vented alignments, that could be awful, but a sealed box much less so. Also a Bessel like vented alignment should cope quite well.
There really is nothing missing in this analysis. This explains why some amps, especially zero feedback types like SET, with typically 3 Ohm output Z and thus hopeless DF and often lower than 1, can still sound OK.
So forget DF, it's the bigger picture that matters!!!
Joe R.
Joe,Joe Rasmussen said:
Nice analysis. It shows that as long as the output Z of an amp is >0, it will have an affect (mathematically) on the Qts of the speaker in its box (ie. the "damping" of the the system). As the amps output Z becomes larger (>1 - 2 ohms), it begains to have a significant affect on Qts and by extension again, the overall damping of the speaker.
So, semantically speaking, "Damping Factor" would appear to be the correct term to give the ratio Zsp/Zout.
However, it has been the experience of most here that well controlled bass can still be had using an amp with relatively high output Z. So that would mean, as you have shown, that dramatic changes in Qts will have little affect on the perceived "tightness" (cone control) of the bass speaker. Where does that leave us?
Why does one speaker/amp combination sound tight yet by just changing the amp, the same speaker will sound loose and flabby? What is the mechanism that imparts the character of "controlled bass"?
roddyama said:
That is a very good question. Yes indeed!
I don't think it is down to one or two factors, but there are at least some I can think off. The first that comes to mind, and ain't a new thought, that the open-loop characteristics must be good and especially linear. These have more stable and reliable output Z, even if they are higher. The other is dynamic stability. Have you ever seen Fletcher-Munson curves? They really also reveal that in the bass we have very little dynamic compression when compared to mids. So we seem sensitive to distortion in the mids but sensitive to dynamic/transient behaviour in the bass (and also, but not quite to the same degree in the treble). I believe this is the key to understanding why some amp really do the trick in the bass. I have heard an amplifier with 2-3 Ohm output Z and only 18 Watt, that absolutely killed 100 Watt plus SS amps, not only in bass quality but also seeming a lot more extended. You might want to read the review of the amp on enjoythemusic.com - here is a part quote:
"The bass was tight and strong... and was the tightest I had ever heard from my system. Remember, I am not measuring the bass reproduction of this amplifier to another tubed one, but my Distech Monoblock 140 watt per channel solid-state amplifiers. Then the tympani kicked in and I heard the most authentic reproduction of tympani I'd ever witnessed. So they did do the bass better than any amplifier here."
In fact this amplifier got one of their Best of 2003 Awards. Oh, I must also confess, I had a hand in this design.
Now you may have an idea what I'm listening to. So how did we get this truly great bass quality. By enhancing transient stability. The key was the way the power supply interfaces with the power supply, actually by fixing and stabilising the surrent. Under transient conditions the current doesn't change. Yes, it is Class A, but with a difference, the current is regulated! That's right, not the voltage but the current regulated. You get this huge power supply isolation.
I think I've said enough....
The complete review:
http://www.enjoythemusic.com/magazine/viewpoint/0104/aachapter52.htm
Joe R.
Hello,
I have some informations about the Damping factor but i still need more.
If i have 6 Class d Amps and a pair of 3 way speaker. I use active crossover in the system. For the bass, i prefer the High DF what the D-class amp can produce.
But if i like to make the midrange and the treble make more softer, how can i make it?
Can i attach a stepped attenuator after the amp and before the midrange driver?
I am not an electronic exper yet, i am kind of a student.
Please give me advice how can i reduce damping factor fro the midrange and the treble from 1000 to 20-30 or lower?
I have some informations about the Damping factor but i still need more.
If i have 6 Class d Amps and a pair of 3 way speaker. I use active crossover in the system. For the bass, i prefer the High DF what the D-class amp can produce.
But if i like to make the midrange and the treble make more softer, how can i make it?
Can i attach a stepped attenuator after the amp and before the midrange driver?
I am not an electronic exper yet, i am kind of a student.
Please give me advice how can i reduce damping factor fro the midrange and the treble from 1000 to 20-30 or lower?
Hello,
I have some informations about the Damping factor but i still need more.
If i have 6 Class d Amps and a pair of 3 way speaker. I use active crossover in the system. For the bass, i prefer the High DF what the D-class amp can produce...
Please give me advice how can i reduce damping factor fro the midrange and the treble from 1000 to 20-30 or lower?
Damping Factor is a myth. In 1975 in a discussion with Richard Small, he made it very clear he wouldn't have a bar of it. In fact I was silly enough to argue that point, only to realise what his position was so simple. The output impedance of an amplifier, in Ohm, is basically in series with DC resistance (DCR) of the speaker. If the output impedance of the amplifier is 10% of the DCR, then the electrical Q of the speaker/driver goes up by 10%. What then happens is that you have changed the alignment of the speaker.
BTW, as far as the speaker system is concerned, that 10% increase makes for 110% - that is what the speaker sees, and as they are in series, how does it know the breakdown ratio is 10% + 100% ??? Yet that is what the calculation of DF is based on. It is completely unfathomable that the speaker should know as it has no way of separating external DCR and internal DCR - yet the proponents of DF just ignore this.
Maybe because they are amplifier designers and not speaker designers. You know, there are very few who designs both - I am one who does.
Then finally we come down to the idea that the speaker when it overshoots produces some kind of back-EMF which is then absorbed by the much lower output impedance of the amplifier. Again this thinking is flawed, why? For one thing, the driver's own DCR is in the way, limiting that possibility. The back-EMF(?) is absorbed mostly by the speaker's own DCR.
Don't look for output impedance to judge the sound, not even the bass. There are some amplifiers with high output impedances that sound magnificient.
I have a transconductance amplifier here that has an output impedance of 270 Ohm and the bass into a compatible speaker is just great - because the alignment is designed to be driven by such an amp. That amp has a negative DF.
Look for amplifiers that are linear dynamically into low frequencies. We are very sensitive to dynamic compression as the frequency descends. Look for amps that don't compress, match the alignment of the speaker - and you are now in the ballpark you need to be.
The Vacuum State DPA-300B has an output impedance of 5 Ohm and yet phenomenal bass - especially with the right speaker.
Vacuum State DPA-300B mono-bloc by fjhuangjun
Cheers, Joe
Last edited:
I'd encourage a more care in developing the circuit model; Re is in the way but not quite in the way being advocated. The notion underlying damping factor is an error voltage develops across the speaker due to contention between two voltage sources---one the power amplifier with its source impedance, the other the back EMF with a source impedance of Re. The definition of damping factor as Zload / Zsource correctly describes the error's scaling with the ratio of these two source impedances. Two other items of note are 1) drivers' efficiency in generating back EMF is the same as their forward efficiency and 2) the threshold of audibility for error terms is usually 50 or 60dB down.
Naïve application of the above indicates a damping factor of 1000 is required to push error 60dB down. This is correct if the speaker has an instantaneous 0dB return loss such that the back EMF delivered by rear waves returning to the drivers' cones/domes/whatever is of the same amplitude as that originally imposed by the power amplifier. In practice, such a case does not occur. Box speakers often manage around 20dB return loss---meaning damping factors of 30 to 100 suffice for inaudibility---and more carefully executed boxes exceed 40dB. Open designs (such as dipoles or variable resistance enclosures) tend to be lower still. So a damping factor of two is potentially subjectively equivalent to one of 80 depending on circumstances. Equivalent circuit models and DF discussions which fail to control for return loss are pretty much statistically insignificant due to the ginormous size of the error bars. This is rarely acknowledged even though audibility of return loss is a topic of extensive discussion by way of enclosure stuffing.
You've touched on voicing shifts at lower DFs, another variable often not controlled for. An additional consideration rarely controlled for is delay and reverb are almost always subjectively pleasing at low to moderate levels. The lower the DF the more the amp+speaker combination develops such behaviour. Typical back wave roundtrip propagation times of a few milliseconds are short compared to delay and reverb effects usually used by musicians and recording engineers. They're quite long compared to mastering techniques commonly used to give a recording a fatter, more pleasing sound such as delaying the entire track by a sample or two and adding the delayed copy back in. I'm not aware of any rigorous studies in this direction. But it's clearly all part of a continuum and an abundant body of evidence indicates delays begin to influence perception of imaging around two or three milliseconds.
These factors combine most significantly in the bass as enclosures are larger and low return losses more difficult to achieve due to box resonances and longer acoustic wavelengths. Hence the commonality of discussion about amplifier and speaker pairing for subjectively desirable thumpa thumpa type results, such as seen in this thread. From a circuit modeling standpoint it's inaccurate to approximate the back EMF source impedance as Re in such cases; additional Thiele-Small parameters and considerations such as speaker compliance volumes need be included too.
Last edited:
Hello,
I have some informations about the Damping factor but i still need more.
If i have 6 Class d Amps and a pair of 3 way speaker. I use active crossover in the system. For the bass, i prefer the High DF what the D-class amp can produce.
But if i like to make the midrange and the treble make more softer, how can i make it?
Can i attach a stepped attenuator after the amp and before the midrange driver?
I am not an electronic exper yet, i am kind of a student.
Please give me advice how can i reduce damping factor fro the midrange and the treble from 1000 to 20-30 or lower?
Hi,
Reposting the same topic of a current thread in a very old thread is very tedious.
rgds, sreten.
Last edited:
Yeah, you are right - I didn't even realise and had to go back to previous pages. Somehow the 'system' remembered and I got email notification - you set me right.
Talk about computers having long memories !!!
.
Well, yes and no.
You are basically just changing the electrical Q (Qes) proportionally relative to the Re of the voice coil - if series resistor adds 10% resistance, then a driver's Qes goes up 10%, say from 0.30 to 0.33. When that driver is put in a box, the alignment has changed. The maths has been in place for more than four decades now.
To apply the idea of a voltage divider, while there is some truth to it, is too simplistic. One has to look for the causes of that peak and that it is a complex resonance and has Motional EMF - we want to apply the "voltage" model because it is simple, alas the "current" model way of looking at things is not easy to grasp. Downright devilish.
The truth as I see it, there is still a lot to learnt in this field.
Cheers, Joe
.
sure, so the electric properitys of the network change as Re has been artificially increased. Still, -realively easily- it can be modelled but all the driver specs now ould have to be re-calculated for every step. still not unmanagable.
different Q factor, different Fs, different "lot of things".
it still yields a "dumb-eq" .
surely that is a verry rough way of handling it,
still it is pretty close.
different Q factor, different Fs, different "lot of things".
it still yields a "dumb-eq" .
surely that is a verry rough way of handling it,
still it is pretty close.
OK then, I give you a riddle:
I obtain a driver, say 8" and got it T-S Parameter. As the Total Q proves to be well below 0.707, that means I can calculate a sealed box that will give me Qc = 0.707 and thus I have made a 2nd order Butterworth alignment where the F3 and Fb coincide and drops off 12db/Octave below that. Should not be difficult, right?
What if I told you that I would be able to make such a box/alignment and be able to drive it from any source impedance and it will still stay a 2nd order Butterworth - in effect have no damping factor at all as the source impedance could be infinite and still be a 2nd order Butterworth alignment.
Is that possible?
If so, what damping factor? None!
Cheers, Joe
.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.