Amplifier Output Impedance (Damping Factor) and Speakers

[DonH56]

It was mentioned here, that amplifiers with lower DF tend to sound warmer. I do not believe that DF (above 20) has anything to do with the sound, but the way it is obtained does. The easiest way to improve anything, including output impedance is to increase NFB. 40dB NFB will reduce output impedance 100x, but can also introduce some TIM distortions - added higher order odd harmonics, that will make sound brighter.

I do not know if that is true. I mean that literally; I do not know. Most of the amplifiers with very low DF I have experience with are tube-based, and the "warmer" ones tended to have rolled-off highs particularly into the relatively low-impedance tweeters I have tended to over the years (ribbon and electrostatic).

 

[restorer-john]

@DonH56 Have you ever had experience with electrostatic speakers and SS amplifiers? I haven't owned any (e-stats) so can't really comment. And no, there's none in my storeroom.

There's always been a lot of discussion regrading amplifier output impedance and the capacitive nature of stats at HF. I'm expecting a stability issue if anything or increased THD at high frequencies/levels.

 

[DonH56]

@DonH56 Have you ever had experience with electrostatic speakers and SS amplifiers? I haven't owned any so can't really comment. And no, there's none in my storeroom.

There's always been a lot of discussion regrading amplifier output impedance and the capacitive nature of stats at HF. I'm expecting a stability issue if anything or increased THD at high frequencies/levels.

Yah, a fair amount. Always expected I would own a pair of ESLs but never did, though I have serviced, sold, and auditioned many in my own and others' systems.

As you know, the ESL panel is basically a big capacitor, and can look like several uF at HF. The actual load to the amp is not always capacitive, since it is usually transformer-coupled, but the impedance is very, very low (often 1 ohm or less at 20 kHz) and often with a large phase angle. So, your gut is right, distortion tends to go through the roof and stability can be a major problem. The saving grace for many amps (and speakers) is that the HF content is low enough in amplitude that they can handle it.

I have seen one ARC tube amp smoke the output transformer and output tubes (an older D-150 IIRC; my D-79 actually handled the same load) and a number of SS amps do anything from just shut down to oscillate and self-destruct, sometimes before anyone had time to react. As in, plug in, turn on, and listen to one quick squeal as the amp died. The big Krell and Mark Levinson did OK, as did a Perreaux, but some well-regarded amps like HK and Spectral did not like it (e.g. the "ultrawideband" leaders back then). Amazingly enough, my Phase Linear 700 did not blow up, and a Crown D-150A overheated (surprised the heck out of me) but was otherwise OK. My Hafler DH-200 was fine with it though ran hot. The Infinity ServoStatik (?) amp shut down immediately.

Can't remember them all now, but must have tried a couple of dozen, and most worked "ok" but most all struggled with high THD and ringing. The ringing was usually related to the output stage of the amp and the ESL's impedance so tended to spray HF energy not harmonically related to the signal. That led to "harsh" high end with some amps.

But, no ESLs, John? Not even a little pair of Janzens? For shame...

 

[Doodski]

SS amps do anything from just shut down to oscillate and self-destruct, sometimes before anyone had time to react. As in, plug in, turn on, and listen to one quick squeal as the amp died.

To my knowledge I have never experienced a oscillating amp. What is happening with the speaker amp interaction when this occurs? What causes it?

 

[DonH56]

To my knowledge I have never experienced a oscillating amp. What is happening with the speaker amp interaction when this occurs? What causes it?

Load the amp with a very low HF load with a high phase angle. If there is not enough loop margin (gain and/or phase) you will make an oscillator. Take a loop with well under 45 degrees of phase margin at the HF limit and stick a big cap on the output and it will sing. Unfortunately, I have seen many oscillating amplifiers over the years (though been out of the repair biz for a couple of decades now), and most were not driving ESLs but some other load. A lot of conventional speakers do not present a "nice" load to the amp, a trend that actually seems to have gotten worse over the years IME/IMO.

 

[restorer-john]

But, no ESLs, John? Not even a little pair of Janzens?

Heard plenty, but never owned any. The Quads with their crowbar protection (amp killer) put me off, plus their legendary fragility.

Only panel speakers I ever owned were some Maggies (a few pairs, I think all I kept was the terminal plates after corrosion got to them) and some unusual JVC/Victor Sawafuji Dynapleat based things.

 

[restorer-john]

and stick a big cap on the output and it will sing

It will sing, but the tweeters won't. Not for long anyway.

 

[pogo]

T&A are doing something else like adding audible distortion or changing the frequency response, both easily measurable.

These measurement methods will not be sufficient and certainly not on a static load that does not take the mechanical moment into account. The different DF should be recognizable via such a measurement:
Link

 

[SIY]

These measurement methods will not be sufficient...

...to help people who refuse to use controls chase fairies in the garden.

not on a static load that does not take the mechanical moment into account

So you've never actually read Olsen or any of the other basic texts dealing with electromechanical equivalents?

 

[Suffolkhifinut]

Have a look at Accuphase's first integrated amplifier, the E-202 in 1974.

They implemented switchable (>30, 5 and 1) 'damping' and it works very well (resistors-0v-NFB loop). A significantly aubible difference and beneficial for many sealed speaker systems, particularly large 12" and larger of the time, that were in realively small cabinets with an otherwise limited and 'tight' bass. Or electrostatics.

View attachment 176012

The article on the Accuphare E-202 is not in anyway related to a loudspeaker load. Using a square wave to test power amp protection is a nonsense, then to compound it using a 2 micro Farad capacitor as a tweeter load what did they expect? Then to extrapolate it across to electrostatic loudspeakers? Readily admit to being a ‘Subjectavist’, joined this website to get measured proof of audio performance. Yet on this thread we get ‘Objectavists’ bandying numbers around without any research to prove their point!

 

[restorer-john]

The article on the Accuphare E-202 is not in anyway related to a loudspeaker load. Using a square wave to test power amp protection is a nonsense, then to compound it using a 2 micro Farad capacitor as a tweeter load what did they expect? Then to extrapolate it across to electrostatic loudspeakers? Readily admit to being a ‘Subjectavist’, joined this website to get measured proof of audio performance. Yet on this thread we get ‘Objectavists’ bandying numbers around without any research to prove their point!

Unfortunately, I can't help you with you complete and utter misunderstanding of the subject we are discussing. Have a lovely day.

 

[Suffolkhifinut]

Unfortunately, I can't help you with you complete and utter misunderstanding of the subject we are discussing. Have a lovely day.

Unless I’m completely wrong the Accuphase E-202 is an amplifier and it’s purpose is to drive loudspeaker loads and is meant for audio use? With regards to the tests they aren‘t valid regarding the amplifier’s function. Science is only valid if it’s pertinent to what it’s trying to validate. If the purpose is invalid the Science proves nothing.
Too often we see tests being carried out to validate what someone believes, rather than trying to disprove it. You only know how strong a wall is if you try to pull it apart.

 

[SIY]

Unless I’m completely wrong the Accuphase E-202 is an amplifier and it’s purpose is to drive loudspeaker loads and is meant for audio use? With regards to the tests they aren‘t valid regarding the amplifier’s function. Science is only valid if it’s pertinent to what it’s trying to validate. If the purpose is invalid the Science proves nothing.
Too often we see tests being carried out to validate what someone believes, rather than trying to disprove it. You only know how strong a wall is if you try to pull it apart.

Here's your sign.

 

[DonH56]

An ESL panel can present a load of several uF at 20 kHz -- the big ones are pushing 10 uF, maybe more. Many conventional speakers have fairly high (negative) phase angles at various points in the response, often due to the crossover network. Square-wave testing, using appropriate band-limiting, has long been used to assess amplifier stability (audio and other).

 

[Suffolkhifinut]

An ESL panel can present a load of several uF at 20 kHz -- the big ones are pushing 10 uF, maybe more. Many conventional speakers have fairly high (negative) phase angles at various points in the response, often due to the crossover network. Square-wave testing, using appropriate band-limiting, has long been used to assess amplifier stability (audio and other).

Agreed years ago used to do it myself, but not to test an Amplifiers’s protection. Not arguing over the validity of using Square-Wave testing providing it’s pertinent. How would you calculate damping factor using an Electrostatic load? An Electrostatic panel is a reactive load with a leading power factor. The opposite of the load presented by a conventional speaker.

 

[sergeauckland]

An ESL panel can present a load of several uF at 20 kHz -- the big ones are pushing 10 uF, maybe more. Many conventional speakers have fairly high (negative) phase angles at various points in the response, often due to the crossover network. Square-wave testing, using appropriate band-limiting, has long been used to assess amplifier stability (audio and other).

Years ago, when magazines had technical reviews, not just flowery words, the standard for testing amplifiers was 8ohms in parallel with 2uF. Square waves into that load had to look square, with minimal ringing, and any deviation was pointed out as a fault. A few amplifier manufacturers even specced the amount of overshoot into 8ohms/2uF loads. An amplifier that wasn't stable into such a load would be rejected as broken.

S.

 

[SIY]

How would you calculate damping factor using an Electrostatic load?

The same way you do it for any other load. Of course, it's a meaningless number except for marketing (where 8R is used as a default numerator), the important measure is source impedance vs frequency.

 

[sergeauckland]

Agreed years ago used to do it myself, but not to test an Amplifiers’s protection. Not arguing over the validity of using Square-Wave testing providing it’s pertinent. How would you calculate damping factor using an Electrostatic load? An Electrostatic panel is a reactive load with a leading power factor. The opposite of the load presented by a conventional speaker.

You measure the output impedance. There's no point in calling it Damping Factor as it doesn't damp anything much.
S.

 

[Suffolkhifinut]

You measure the output impedance. There's no point in calling it Damping Factor as it doesn't damp anything much.
S.

Agreed however there is a World of difference between a primarily inductive load and a capacitive load.
The Armstrong 625 used to measure very well using the 8 Ohm / 2 microfarad load test method.

 

[pogo]

So you've never actually read Olsen or any of the other basic texts dealing with electromechanical equivalents?

And what measurement incl. setup currently takes this into account to show the behavior with different DFs on complex loads?
With transient signals, each turn of the voice coil couples individually and proportionately, so the resistance is distributed. I also always miss this in the chains of explanation presented here.