Amplifier Output Impedance (Damping Factor) and Speakers

[Tim Link]

I took the Denon amp and some other items back to the shop to partially unclutter my house . The last thing I did with it was hook it up to Revel M16s. I didn't perceive any special synergy with that combination. At some point when I can get a blind testing setup figured out I may get the Denon back again and determine if I can really tell the difference.

 

[fpitas]

Alright, the measurements have been taken comparing the Denon PME900HNE and the Camgridge Azur 540R. The measured difference is minimal, and in the opposite direction of what I expected. The Denon actually has a little more output above about 7k. This was measured with just 1 speaker (Sony SSCS5) hooked up to 1 channel and both receivers used their optical inputs and built-in dacs. The llight blue is the Denon, hot pink is the Cambridge. Taken at 1 meter on axis with tweeter. One thing that seems to be true looking at the impulse response is that the polarities are inverted from each other.

So why the heck do I think the Denon sound softer in the highs? One possibility is that I was using the line level subwoofer outputs while listening to both amps, and maybe the output is higher on the Denon than the Cambridge. The thing is I didn't always listen with the sub. I guess I'm just going to have to listen for a while more to both of them without the sub. Now that I've seen this result I wonder if I'll still perceive a difference.

There's another potential complication. I've been listening with my 3 speaker matrix mixed setup, which has the center channel playing left plus right channels mixed, and the sides play left minus right mixed. The side channels were wired in parallel and out of phase from each other so one amp channel was seeing a lower impedance load than the other. Maybe the Denon could handle that a bit better than the Cambridge?
I guess I'd better measure the response with 2 speakers wired in parallel and out of phase with each other.

View attachment 275322


So here's the results of both amps pushing 2 SSCS5s wired in parellel off of one channel, with one speaker out of phase of the other and spaced about 3 feet apart - microphone in the middle. Orange is the Cambridge. Again seeing a little more roll off up top compared ot the Denon. Some slight differences here and there but I wouldn't expect to hear much looking at that graph. Maybe those differences are more significant than they look.
View attachment 275335

Sometimes the perception of "highs" is really material around 1 - 2kHz. Our ears are pretty sensitive there.

 

[DonH56]

Our hearing sensitivity peaks around 3 kHz.

 

[fpitas]

Our hearing sensitivity peaks around 3 kHz.

Yes, and falls from there. It has just been my experience that the 1 - 2kHz region can be perceived as "brightness".

 

[Tim Link]

Sometimes the perception of "highs" is really material around 1 - 2kHz. Our ears are pretty sensitive there.

I have noticed that can sometimes be it. It's interesting that the two amps didn't differ much at all in that range, only up above 7k. I've now listened to 4 amps with those speakers, a 7.2 Denon AVR, a 5.2 Denon AVR, the Cambridge 540R, and the Denon PMA-900HNE. It seems they all sound a little different. I can rank them in order of preference, with the 900 being first, the 5.2 Denon being second, the 7.2 Denon being third, and the Cambridge being my least favorite. Last night I listened for a few hours to the Denon 5.2 and messed with the EQ quite a bit, including turning down the 1 - 2kHz range. I could change the sound, I think even improve it. But it doesn't create the same effect as I heard on the PMA-900. I accidentally EQ'd that unit too by leaving my EQ attempts at fixing the Cambridge on and that surprisingly didn't wreck the perceived effect. It's something subtle but very nice about snares, cymbals, tinkly metallic sounds. Massed strings also come across as particularly pretty sounding. I wouldn't call it particularly realistic but just very nice. It might be possible to create it with the right EQ. I'm hoping so but I haven't figured out the trick yet. I was able to make some pretty nasty sibilance by adding gain around 7k, so that's a good target zone if I ever hear excess sibilance. My past experience with building and EQing speakers is that I occasionally hit upon these kinds of pretty effects and I think it's a broadband thing. It comes across as nice highs but really involves balances over at least several octaves.

 

[Trdat]

So some simple questions that might and should help me understand the post better.

If we get half the power at 8ohm compared to 4ohm than if there is a peak of 25ohm in a sub or a midwoofer is the amplifier producing very little power between those frequencies of the high ohms?

Also, do I understand correctly that the damping factor offsets one part of the impedance peak in SS amplifiers? The post conclusion was that the two SS amps were probably inaudible when it came to the variation in how the amp controled the impedance peaks is that from the damping factor?

Ultimately, I get we match the 4ohm part of the impedance with a 4ohm spec of an amp and the 8ohm part of the impedance with the 8ohm spec of the amp and I get that it varies but how do we match say a sub with major peak in the main part of the impedance curve between 30 to 100hz when its up to 30ohm?

 

[NTK]

So some simple questions that might and should help me understand the post better.

If we get half the power at 8ohm compared to 4ohm than if there is a peak of 25ohm in a sub or a midwoofer is the amplifier producing very little power between those frequencies of the high ohms?

Also, do I understand correctly that the damping factor offsets one part of the impedance peak in SS amplifiers? The post conclusion was that the two SS amps were probably inaudible when it came to the variation in how the amp controled the impedance peaks is that from the damping factor?

Ultimately, I get we match the 4ohm part of the impedance with a 4ohm spec of an amp and the 8ohm part of the impedance with the 8ohm spec of the amp and I get that it varies but how do we match say a sub with major peak in the main part of the impedance curve between 30 to 100hz when its up to 30ohm?

Don't think in terms of power. Loudspeakers are designed to respond to voltages, i.e. if the loudspeaker has a flat frequency response, 1 Vrms at 80 Hz will produce the same dBSPL as 1 Vrms at 1000 Hz (or at other frequencies in its operating frequency range). And amplifier amplifies voltage, i.e. output voltage = input voltage * gain.

The important part for the amplifier is it can deliver the current to your loudspeaker as dictated by Ohm's law, (current = voltage / impedance). If it can't, output voltage will sag and clipping and lots of distortions will result.

Power is mostly meaningless. It ultimately only means how much heat is dumped into the environment.

 

[DVDdoug]

Also, do I understand correctly that the damping factor offsets one part of the impedance peak in SS amplifiers? The post conclusion was that the two SS amps were probably inaudible when it came to the variation in how the amp controled the impedance peaks is that from the damping factor?

A high damping factor means the amplifier's output impedance is very low compared to the load (speaker) impedance. And it makes the amplifier act as a "voltage source" where the voltage is independent of the load. This is what you want and this is how speakers are spec'd and tested.

Most solid state amplifiers are very good at this. It's a natural characteristic of transistors & MOSFETs and it can be improved with feedback. Negative feedback ("corrective feedback") can improve LOTs of things. If I was looking at a tube amp I'd check the damping factor. With a solid state amp, I'm not going to even think about it.

Ultimately, I get we match the 4ohm part of the impedance with a 4ohm spec of an amp and the 8ohm part of the impedance with the 8ohm spec of the amp and I get that it varies.

We don't "match" impedance. The output impedance is normally less than 1-Ohm (that's the damping factor) and if you match the impedance by connecting a 1-Ohm speaker (or several speakers in parallel) you'll "pull" excess current (which also translates onto excess power) and possibly burn it out, or it may go into thermal shutdown, etc.

but how do we match say a sub with major peak in the main part of the impedance curve between 30 to 100hz when its up to 30ohm?

The frequency response will be "as-specified" but less current will flow at those frequencies so you are putting less stress on the amplifier, and you are consuming less power and generating less heat in the speaker's voice coil.

On the other hand, if you have an impedance dip, you'll pull more current from the amplifier, putting more stress on it. But with regular program material that's usually OK because you only get maximum power at a certain frequency very-temporarily.

 

[mhardy6647]

Power is mostly meaningless. It ultimately only means how much heat is dumped into the environment.

Indeed.

As Col. Paul Wilbur Klipsch put it*:

After all, bragging about how much power your speaker can absorb is like bragging about how much fuel your vehicle can burn.

___________
* and interestingly, he wrote this in the April 1974 issue of Klipsch's newsletter Dope from Hope. Perhaps some of you remember 1974. How much fuel one's vehicle burned was a matter of great interest in those days.

 

[Trdat]

A high damping factor means the amplifier's output impedance is very low compared to the load (speaker) impedance. And it makes the amplifier act as a "voltage source" where the voltage is independent of the load. This is what you want and this is how speakers are spec'd and tested.

So what I get from this is that regardless of load the voltage is independant, so does that mean the amplifier can get enough or the correct amount of power regrdless if its 25ohm or 2ohm? I suppose this is a laymens understanding.

Most solid state amplifiers are very good at this. It's a natural characteristic of transistors & MOSFETs and it can be improved with feedback. Negative feedback ("corrective feedback") can improve LOTs of things. If I was looking at a tube amp I'd check the damping factor. With a solid state amp, I'm not going to even think about it.

So you are recommending that there is absoultey no point in even thinking about what happens at 30hms or 2ohms in a speaker impedance if I got a moden Solid state amp?

We don't "match" impedance. The output impedance is normally less than 1-Ohm (that's the damping factor) and if you match the impedance by connecting a 1-Ohm speaker (or several speakers in parallel) you'll "pull" excess current (which also translates onto excess power) and possibly burn it out, or it may go into thermal shutdown, etc.

So if we don't match impedance then what do we match? or there is no need?

The frequency response will be "as-specified" but less current will flow at those frequencies so you are putting less stress on the amplifier, and you are consuming less power and generating less heat in the speaker's voice coil.

On the other hand, if you have an impedance dip, you'll pull more current from the amplifier, putting more stress on it. But with regular program material that's usually OK because you only get maximum power at a certain frequency very-temporarily.

I understand this but I can't picture what happens at the response with very high ohms. Less, current is passing through but is enough volts passed through to provide enough power? Or does this go back to the above that irrespective of load the amplifier does manage to provide the correct current?

 

[antcollinet]

So what I get from this is that regardless of load the voltage is independant, so does that mean the amplifier can get enough or the correct amount of power regrdless if its 25ohm or 2ohm? I suppose this is a laymens understanding.

IF the amp is able to output sufficient current into the 2ohm part of the impedance characteristic.

So you are recommending that there is absoultey no point in even thinking about what happens at 30hms or 2ohms in a speaker impedance if I got a moden Solid state amp?

High parts of the impedance charcteristic are never a problem. The low parts demand more current (and therfore more power) from the amp at those frequencies. Once the amp runs out of current capability, the voltage will dip and influence the system frequency response.

o if we don't match impedance then what do we match? or there is no need?

We don't need to match - other than sufficient power capability to drive the low impedance parts of the curve at the volume we want to listen to.

But bear in mind that in a complex wide band signal such as music, the amount of power needed at a particular frequency dip will (normally) be a relatively small part of the overall power. This might go wrong (as an example) if you have a 2ohm dip at low frequency, and are playing organ music with a particular pipe at that same frequency. In that situation the amp might have to deliver double the power compared to the nominal 4ohm.

Bear in mind though that a 2ohm dip is pretty extreme. Most speakers (even 4 ohm) are fairly benign and don't go much below 3ohm.

I understand this but I can't picture what happens at the response with very high ohms. Less, current is passing through but is enough volts passed through to provide enough power? Or does this go back to the above that irrespective of load the amplifier does manage to provide the correct current?

A speaker (voltage driven remember) with flat FR will output the same SPL at the same voltage - regardless of frequency. Even if the power dips due to high impedance at a particular band. You can think of that as the speaker being more sensitive (to power) at that frequency.

 

[DonH56]

So what I get from this is that regardless of load the voltage is independant, so does that mean the amplifier can get enough or the correct amount of power regrdless if its 25ohm or 2ohm? I suppose this is a laymens understanding.

Look at the first post or two to see how frequency response might vary with load. The amplifier is completely independent of the load (speaker) only if the output impedance is 0 (and thus damping factor is infinite). With real amplifiers, there will be some change with load, but in practice once you get above a fairly low damping factor there is insignificant variation (opinions vary, but anything over 100 will be essentially independent of most speakers).

Note the amplifier does not "get" power; it delivers power to the speaker that "gets" the power. How much power depends upon the speaker's impedance and where you set the volume (lower impedance, louder sound, means more power is required from the amplifier).

So you are recommending that there is absoultey no point in even thinking about what happens at 30hms or 2ohms in a speaker impedance if I got a moden Solid state amp?

I rarely state absolutes. Two ohms nominal impedance is very low for a speaker and many amplifiers may current limit if required to deliver lots of power into a 2-ohm load. That is a separate issue from damping factor or output impedance, the topic of this thread. But the vast majority of SS amps will handle the vast majority of speakers with no problems, assuming you buy an amp that can deliver the power you need. See e.g. http://myhometheater.homestead.com/splcalculator.html to get a gross estimate of the power you may need.

So if we don't match impedance then what do we match? or there is no need?

Impedance matching to maximize power transfer is something common in high-speed systems like Ethernet, PCIe, USB, RF devices, and so forth. It also minimizes other issues present in very high speed signals, all well (decades) above audio. Audio frequencies are so low, and thus wavelengths so long, that it is not a concern. A perfect voltage source, having 0-ohm output, would resist any problems from mismatched impedances. But for audio, speakers are usually designed to be driven by an amplifier having low output impedance that is close to an ideal voltage source. Long way of saying "there is no need".

I understand this but I can't picture what happens at the response with very high ohms. Less, current is passing through but is enough volts passed through to provide enough power? Or does this go back to the above that irrespective of load the amplifier does manage to provide the correct current?

Amplifiers typically have power supplies that set the maximum voltage. Current is limited by the output devices can handle, what the power supply (and/or wall outlet) can deliver, and thermal management (heat). Assuming the amplifier is not clipping, then it will output a voltage based upon the input voltage, determined by the gain of the amplifier.

If you apply 1 V from your source (preamp, DAC< whatever), then perhaps the amplifier puts out 30 V. It does not matter the load, again assuming the amplifier does not clip or go into thermal shutdown; it will deliver 100 V at the output no matter the speaker's impedance. Power is voltage squared divided by the impedance of the load. Below is a table showing the power output for various input and output voltages to an example amplifier. In the real world power will likely be reduced at the high end as clipping or protection circuits kick in. The table is for a "perfect" amplifier, and impedance (resistance, R) is a single number. For a real speaker, R will vary over frequency, and power will adjust accordingly whilst voltage from the amplifier (Vout) stays essentially the same.

Amplifier gain = 30 (29.5 dB)
Power = Vout^2 / R (W)
Input Voltage	Output Voltage	2-Ohm Power	4-ohm Power	8-ohm Power	16-ohm Power	32-ohm Power
0.01​	0.3​	0.045​	0.023​	0.011​	0.006​	0.003​
0.05​	1.5​	1.125​	0.563​	0.281​	0.141​	0.070​
0.1​	3​	4.500​	2.250​	1.125​	0.563​	0.281​
0.5​	15​	112.500​	56.250​	28.125​	14.063​	7.031​
1​	30​	450.000​	225.000​	112.500​	56.250​	28.125​
2​	60​	1800.000​	900.000​	450.000​	225.000​	112.500​
4​	120​	7200.000​	3600.000​	1800.000​	900.000​	450.000​

 

[KSTR]

So if we don't match impedance then what do we match? or there is no need?

If there are dedicated outputs or a switch to select speaker impedance on SS or Class-D amps then one should use them. For tube amps this also mandatory.
Because if you run a 2 Ohms speaker on an "8 Ohm" output the amp will clip or shut off due to too much current asked from the amp too early.
OTOH, if you attach a 16 Ohms speaker to a "4 Ohm" output the amp will voltage-clip too early, not reaching full power again.

 

[Deleted member 48726]

So some simple questions that might and should help me understand the post better.

If we get half the power at 8ohm compared to 4ohm than if there is a peak of 25ohm in a sub or a midwoofer is the amplifier producing very little power between those frequencies of the high ohms?

Also, do I understand correctly that the damping factor offsets one part of the impedance peak in SS amplifiers? The post conclusion was that the two SS amps were probably inaudible when it came to the variation in how the amp controled the impedance peaks is that from the damping factor?

Ultimately, I get we match the 4ohm part of the impedance with a 4ohm spec of an amp and the 8ohm part of the impedance with the 8ohm spec of the amp and I get that it varies but how do we match say a sub with major peak in the main part of the impedance curve between 30 to 100hz when its up to 30ohm?

Let's say a speaker response is measured at 2.83 Volts. Then we connect an amp and adjust the volume so that the output = 2.83 Volts.
The speakers impedance varies across the frequency spectrum and the Current (A) is thus determined by this and will be provided by the amp. A low impedance at e.g. 60 Hz will demand more current from the amp.
Then let's say that we increase the volume so that the output voltage is = 30 Volts. The low impedance at 60 Hz from before is the same as this is inherent in the speaker design. The voltage from the amp is adjusted to 30 Volts. If the amp can't supply the needed current the only thing that can happen if that the voltage will drop because
U = R x I and remember that we ask the amp to supply 30 Volts and that the R (dip at 60 Hz) is set by the speaker and because I that is needed drops because the amp is at its limit, If it can't supply the current that is needed, voltage will drop, clip and distort.

 

[nutzandvoltz]

If we get half the power at 8ohm compared to 4ohm than if there is a peak of 25ohm in a sub or a midwoofer is the amplifier producing very little power between those frequencies of the high ohms?

Speaker loading effects on an amplifier varies due to the speaker's construction. In a voice coil type speaker, the impedance curve can be plotted by using the inductive reactance formula. The speaker's normal operating range is the in phase response, when the impedance goes greater than 3 times its nominal or average impedance, the phase response of the speaker changes as the impedance increases and does not pull as much current compared to its in phase operating range. The power is the same power coming out of the amp, its just the speaker doesn't pull very much current in those frequencies because its out of the proper operating range.

 

[pogo]

It's a shame that an A200 or an M200 still hasn't made the trip to ASR. From my point of view, this would trigger a rethinking, as these amplifiers are ideally designed for a test (DF HI > 800 / DF LO > 70, @8ohms).

 

[mhardy6647]

It always comes down to T & A, doesn't it?

Trigger warning: A Chorus Line content.

Spoiler

 

[pogo]

It always comes down to T & A, doesn't it?

The advantage of these T+A amps is that only the output impedance can be changed and nothing else. I can't imagine a better UUT for this topic.

 

[kchap]

The advantage of these T+A amps is that only the output impedance can be changed and nothing else. I can't imagine a better UUT for this topic.

Just seems like a marketing gimmick, like interchangeable OP amps.

 

[fpitas]

Just seems like a marketing gimmick, like interchangeable OP amps.

Yeah. I mean, I can add a series resistor if that's what I want.