Audibility of Low Damping Factor? - Benchmark Myth-Busting White Paper

[BikeSmith60]

What is still leaving me perplexed at this stage is amplifier manufacturer's use of DF as a marketing argument. Hegel for example extolls on the virtues of DF values of 4000 in their higher-end products. They claim and users often agree that Hegel amplifiers generally "control" woofers and therefore bass response most admirably. Naim, on the other hand, is or was notorious for not caring a bit about this and selling truckloads of amplification with measured DF between 10 and 50.
I am indeed quite curious if an amplifier with a very high DF, used with cables of average, non-detrimental values, could better control the bass response of my current speakers which I would call the most unruly, ill-behaved boxes of boom, doom and gloom ever conceived (independent of room acoustics, of course). These speakers are currently driven by a Naim Supernait 2 and frankly, they despise each other!

 

[tvrgeek]

Fancy papers. Well, by listening to amplifiers for 50 years, I concluded a long time ago that something over 100 was about where to start. I'll buy the 200 number. ( with my 16 gauge zip cord of course.

And of course, any number that can be advertised "bigger is better" will drive design, epically boutique markets.

Bike... If your speakers are what you describe, I suggest it is neither the amplifier nor cables, but your speakers. Magic electronics are never going to solve poorly executed driver alignments. Yes, I have heard some very expensive speakers that were total do-do. If you read the paper, you will see your basic premise is totally incorrect.

 

[BikeSmith60]

@tvrgeek Yes, thanks for the thoughts. It's interesting to note that typically and historically Naim have felt that not building a zobel network in the output stage was/is better. This must be compensated by the speaker cables, according to Naim with very specific and somewhat rare LCR values and with lengths between 3.5 and 5 meters per side. Yep, this is a fact along with using their own DIN connectors as well...So anyway, it's a bit peculiar with Naim firstly and secondly it is no mystery that some speaker/amp associations make happy marriages and some do not. My question and interrogation is would an amplifier with a high DF make any difference or not. No, haven't tried one yet...I have tested with Jadis tube electronics, probably low DF and same effect. Also with British-made Musical Fidelity SS/AB amplification, bit better but unsure of DF difference. Curious to see if a powerful Class D properly implemented could be interesting or not. In the meantime, ripping my hair out like I generally do...
BTW, thanks for not asking which pair of evil speakers I have!

 

[tvrgeek]

No output filter seems like a good way to cause instability!

I spent about a year looking into the happy marriage issue. Great help from John Curl. He was taken aback when I mentioned my wife preferred Rotels over his HCA on our speakers. ( Paradigm at the time) I dug deep into design and with his "tough love lead" came to the conclusions I have on VAS output loading compensation vs TMC. Once I upgraded my speakers, ( my own) the Parasound was "clearly" (pun intended) better. I then modified a Rotel and a Hafler to TMS and the same attributes carried. Cheap speakers, VAS loading. Good speakers, Miller or TMC. It has to do with HD being lower across the board but can excite cheap tweeters, or higher through mids but decreasing so not upsetting cheap tweeters. Conformation the designers use the appropriate design for their market.

 

[tvrgeek]

Add on:
I almost went for a March amp as it seems to qualify as the "well implemented". I chickened out as even with excellent returns, I really don't like to do that to companies. I like to be sure first. You may have noticed my posts here in trying to get a good DAC. Had to return an SMSL and a Topping. Keeping my Schiit. I knew the Parasound 2125 would be "good enough". If it were not for COVID, I would have traveled this summer to MG car club events around the country and side-trip to high end shops and maybe heard a Purify based amp. Not here in the Triangle!

Don't know about Jadis. JoLida factory was around the corner from where I worked and I listened to Carys a bit. ( Ironic, close to where I live now) I am not a tube guy. I don't want a Taylor to sound like a Fender.

 

[witwald]

I am indeed quite curious if an amplifier with a very high DF, used with cables of average, non-detrimental values, could better control the bass response of my current speakers which I would call the most unruly, ill-behaved boxes of boom, doom and gloom ever conceived...

Possibly this article from 1975 by Floyd E. Toole, published in AudioScene Canada, may provide some insights: Damping, Damping Factor, and Damn Nonsense.

 

[witwald]

What is still leaving me perplexed at this stage is amplifier manufacturer's use of DF as a marketing argument. Hegel for example extolls on the virtues of DF values of 4000 in their higher-end products. They claim and users often agree that Hegel amplifiers generally "control" woofers and therefore bass response most admirably. Naim, on the other hand, is or was notorious for not caring a bit about this and selling truckloads of amplification with measured DF between 10 and 50.

Amplifiers with low damping factor values, say in the range 10 to 20, will have a large effect on the frequency response of typical loudspeaker systems. The frequency response of the amplifier is modified by the variation in the impedance of the loudspeaker as a function of frequency. Think of the peaks in the impedance located at the two resonance frequencies of a vented enclosure, or the single peak of a sealed enclosure. Such peaks are usually multiples of the resistance of the woofer's DC resistance. In addition, there will usually be an impedance bump/peak through the crossover region, followed by an impedance rise at high frequencies due to the voice-coil inductance of the tweeter (if a dome/cone tweeter is used). The effect of this type of impedance behaviour on amplifiers with various values of damping factor was illustrated in the graphs found in Post #19 in this thread.

 

[Head_Unit]

Net result was that the frequency response variations were very small as to not be worth continuing to test [amps with simulated speaker load].

Well, that's pretty true for solid-state designs, but for tube amps there is significant variance.

To expand the topic, what about capacitive/inductive character of output impedances? Not just of amps, but of sources...and input impedances of preamps and power amps? I've never seen this tested. Are they all really resistive? Since many devices change input or output impedance versus frequency my dim recollections of electrical engineering school are whispering in my head that cannot be resistive.

 

[KSTR]

My findings of the DF issues:
Once you make the frequency response of a speaker the same regardless of (moderate, say, single-digit ohms) drive impedance, the differences mostly vanish. Sometimes I was able to ABX minor difference in the "fine print" of the sound signature but most of the times I was not. Only when drive impedance is extreme and does significantly affect "back-emf" damping, differences tend to be obvious and can easily explained by transducer theory: While the transducer will give the same linear output, the nonlinear residue is not the same and especially the overdrive recovery (from even the slightest excursion overdrive when force factor starts to collapse) can be remarkedly different, as can be the response to external exitation.
Bottom line: obsession with DF is moot, given that the actual damping is dominated by the transducer's own DC resistance (plus any in the XO). A minor freq response change that is fully reversible, that's it

 

[tvrgeek]

OK, here is a question. In a system when running a sub, so mains crossed over somewhere, 60-100 whatever. Does DF matter at all on the mains?

I say this as I have been working up my next few speaker builds and have been looking for a very good midrange to cover from 1000 to 4000 or so. There are some good candidates, but all of them are significantly lower efficiency than the 7 or 8 inch mid-woofers I am looking at. So, what effect would a pad on the mid-woofer have? Adding an Ohm means DF goes to zip, but in this case, does it matter? Crossed way above F3.

 

[DonH56]

OK, here is a question. In a system when running a sub, so mains crossed over somewhere, 60-100 whatever. Does DF matter at all on the mains?

I say this as I have been working up my next few speaker builds and have been looking for a very good midrange to cover from 1000 to 4000 or so. There are some good candidates, but all of them are significantly lower efficiency than the 7 or 8 inch mid-woofers I am looking at. So, what effect would a pad on the mid-woofer have? Adding an Ohm means DF goes to zip, but in this case, does it matter? Crossed way above F3.

"At all", yes; in practice, maybe... It depends upon the output impedance of the amplifier (1/DF), wire impedance, and speaker impedance (looking into them so typically crossover + drivers + resonances d=ue to the cabinet). Amplifier output impedance goes up (DF goes down) with frequency because feedback decreases (less loop gain at higher frequencies). But, for SS amps, it is usually still pretty high at 20 kHz. Wire impedance change over audio frequencies can usually be neglected; just use the resistance and it will probably be close enough. That leaves speaker impedance, which can be all over the map and is a function of the speaker itself. If the midrange impedance is fairly flat over the frequency range you are using it, then the pad won't matter much. You can calculate the voltage division over frequency to see if throwing in a pad will create any significant response variations. Many speakers include midrange and/or tweeter pads any to match the woofer; in fact, a lot of bookshelf designs pad the mid/tweeter so they can spec deeper bass by trading sensitivity for bass extension. You may give up a little in measured transient response and distortion with higher driving impedance but chances are it is audibly negligible.

IMO - Don

 

[tvrgeek]

I thought we tended to get lower distortion with higher impedance loads as BJTs are more linear voltage amplifiers than current amplifiers, hence justification for MOSFET outputs. I see a very large trend for lower impedance speakers, and that makes me wonder especially as most AVRs can't safely drive 4 Ohm loads. The more I think about it, as the load is outside the feedback loop, it can only effect LF extension which is irrelevant in this case. Got to be careful with the resisters as a 3 dB pad may take a lot of power.

 

[witwald]

OK, here is a question. In a system when running a sub, so mains crossed over somewhere, 60-100 whatever. Does DF matter at all on the mains?

DF most definitely matters, and most especially if it is low in value. See the plot below that shows the effects of DF on the frequency response of the amplifier.

I say this as I have been working up my next few speaker builds and have been looking for a very good midrange to cover from 1000 to 4000 or so. There are some good candidates, but all of them are significantly lower efficiency than the 7 or 8 inch mid-woofers I am looking at. So, what effect would a pad on the mid-woofer have? Adding an Ohm means DF goes to zip, but in this case, does it matter? Crossed way above F3.

I'd expect that the Ohm you refer to is part of the loudspeaker impedance, by virtue of it being part of the equalisation being brought into play by the crossover network. On its own, which it really can't be, a pad wouldn't affect the frequency response of an amplifier with a very high DF all that much, with the main effect seeming to be a change in the high-frequency –3 dB cut-off point. If your hypothetical loudspeaker had conjugate impedance networks placed appropriately, it would simply produce a flat (resistive) impedance curve. Then your padding resistor wouldn't have any significant effect at all on the frequency response of a typical amplifier in its pass-band! And neither would the impedance of the loudspeaker itself.

The changes in amplifier frequency response when the DF is a small value come about as a result of the variation in the total loudspeaker impedance, not any particular component in the crossover network. That's why an amplifier+cable combination that still produces a high DF will ensure that the frequency response of said amplifier will be relatively load insensitive.

 

[andymok]

Crown has the Studio Reference amplifier that claims DF of 20,000 back then

 

[witwald]

I've gone and simulated an impedance curve for a hypothetical loudspeaker that uses a sealed enclosure and has a conjugate impedance network fitted so as to equalise the impedance at mid-frequencies and high-frequencies. For this loudspeaker, the woofer resonance is at about 70 Hz. Above 200 Hz the impedance is completely flat, with a value of 7.28 ohms. Hence, above 200 Hz or so, the equalised impedance would be purely resistive. Of course, in other speakers the value of this resistance might be less than what I've used here; it would depend on the loudspeaker in question.

The diagram below shows the simulated conjugate-impedance equalised loudspeaker impedance in red. The frequency response of amplifiers with various damping factors are shown as well. For DF = 10, there is a peak in the amplifier's response at 70 Hz that is about 0.63 dB above the flat part of the amplifier's response (above 200 Hz). The relative boost in low-frequency output is quite noticeable. By the time we have a DF = 40, the boost at 70 Hz is only about 0.16 dB, which is much less noticeable.

I think that the above example shows that tube amplifiers, with their relatively low damping factors, would benefit from the connected loudspeakers having their impedance equalised using a conjugate impedance network. It's surprising that I haven't heard of this being done by anyone, as there would be quite noticeable sonic differences coming into play. The conjugate impedance networks aren't that difficult to design, although they would likely require some large component values for the inductors and capacitors, which might make them a little expensive to construct.

 

[sergeauckland]

KEF did it in their 104.2 and 107 loudspeakers in the 1980s. In fact, they even got rid of the impedance peak at LF, having an essentially resistive 4 ohms all the way across the audio band. However, they gave it up in later loudspeakers presumably because of cost and that nobody cared enough about the issue for it to be a marketing story worth pursuing.

S.

 

[ElNino]

I think that the above example shows that tube amplifiers, with their relatively low damping factors, would benefit from the connected loudspeakers having their impedance equalised using a conjugate impedance network. It's surprising that I haven't heard of this being done by anyone, as there would be quite noticeable sonic differences coming into play. The conjugate impedance networks aren't that difficult to design, although they would likely require some large component values for the inductors and capacitors, which might make them a little expensive to construct.

These used to be fairly popular decades ago... a lot of loudspeaker designers used to refer to them as "Zobel networks", even though technically they might not be Zobels. Usually designers would only worry about impedance peaks in the midrange and upwards as the component values could get large otherwise. More recently, Zaph included optional impedance compensation networks in some of his DIY designs.

 

[Ongii]

According to the manufacturer of my loudspeakers (Zu Audio) I shall look for amplifiers with a low DP.
This has been confirmed by tests in newspapers.

 

[Pdxwayne]

I may have found amp + speakers combo that exhibits audible difference when changing speaker cable awg.

That would be Parasound A21 stereo amp paired with either Paradigm Signature S2 speakers or Paradigm Persona B speakers.

I have placed my 10 ft long10 awg speakers cables in storage for a couple of years because it sounded too warm with my Parasound A21 and Paradigm Signature S2. Using generic 12 AWG cables sounded more balanced to me.

Earlier this week I took out the 10awg cables from storage and used them with my Parasound A21 and Paradigm Persona B speakers. Again, as compared to 12 awg cables (one about 7 ft, the other about 14 ft), the 10 awg cables made my speakers sounded too warm. I ended up had to adjust my speakers toe in angle to directly facing my ears (previously facing outside of my shoulders) to get a more balanced sound.

The differences between 12 awg and 10 awg should be too small to be audibled, but in my case it is so obvious that I had to adjust my speakers toe in angles to restore balance.

Paradigm Signature S2 impedence and phase measurements are at
https://www.soundstagenetwork.com/measurements/paradigm_signature_s2/

Paradigm Persona B impedance and phase measurements are at
https://www.soundstagenetwork.com/i...oudspeakers&catid=77:loudspeaker-measurements

These are my personal observations. Sorry, not going to spend time defending myself doing abx. ; P

 

[SIY]

I may have found amp + speakers combo that exhibits audible difference when changing speaker cable awg.

That would be Parasound A21 stereo amp paired with either Paradigm Signature S2 speakers or Paradigm Persona B speakers.

I have placed my 10 ft long10 awg speakers cables in storage for a couple of years because it sounded too warm with my Parasound A21 and Paradigm Signature S2. Using generic 12 AWG cables sounded more balanced to me.

Earlier this week I took out the 10awg cables from storage and used them with my Parasound A21 and Paradigm Persona B speakers. Again, as compared to 12 awg cables (one about 7 ft, the other about 14 ft), the 10 awg cables made my speakers sounded too warm. I ended up had to adjust my speakers toe in angle to directly facing my ears (previously facing outside of my shoulders) to get a more balanced sound.

The differences between 12 awg and 10 awg should be too small to be audibled, but in my case it is so obvious that I had to adjust my speakers toe in angles to restore balance.

Paradigm Signature S2 impedence and phase measurements are at
https://www.soundstagenetwork.com/measurements/paradigm_signature_s2/

Paradigm Persona B impedance and phase measurements are at
https://www.soundstagenetwork.com/i...oudspeakers&catid=77:loudspeaker-measurements

These are my personal observations. Sorry, not going to spend time defending myself doing abx. ; P

Is this the appropriate place to work out your issues?