Can anyone explain to me what is damping factor? I read in handbook that it has to do with output impedance/internal impedance. But how can internal impedance of a power amplifier be measured (or be calculated?).
At first I think it has to do with how many output transistors we use. More output transistors gives more damping factor. But I'm quite surprised when I look at a certain professional studio amp (that has damping factor >1000), it has a very few output transistor. So my assumption is wrong.
I tried to design an ordinary amp with 3 stages (differential, VAS, current amp), but it's goal is to pursue damping factor >1000. What is the key to pursue high damping factor in 3 stages power amp? Please someone share the knowledge.
At first I think it has to do with how many output transistors we use. More output transistors gives more damping factor. But I'm quite surprised when I look at a certain professional studio amp (that has damping factor >1000), it has a very few output transistor. So my assumption is wrong.
I tried to design an ordinary amp with 3 stages (differential, VAS, current amp), but it's goal is to pursue damping factor >1000. What is the key to pursue high damping factor in 3 stages power amp? Please someone share the knowledge.
lumanauw said:
Damping factor is simply the ratio of the nominal load impedance divided by the amplifier's output impedance.
So if you're looking at a nominal 8 ohm load and the output impedance of the amplifier is 0.1 ohm, the damping factor is 8/0.1 or 80. For a damping factor of 1,000, the amplifier's output impedance would have to be 0.008 ohms.
No, you're on the right track. You can reduce output impedance by paralleling a whole bunch of output transistors. You can also reduce output impedance by way of negative feedback. Or both. So while the amp with the high damping factor may have just a few output transistors, it would likely be incorporating a high amount of negative feedback.
Is the goal of a damping factor greater than 1,000 purely for the sake of doing it? If not, then I wouldn't bother.
se
kiwi_abroad said:What's the point of a stupendious damping factor, implying very low output impedance, when your speaker cables will add around 0.05 to 0.1 ohms.
Yeah. A bit pointless for your amplifier's output impedance to be an order of magnitude lower than the resistance of the speaker cable.
But it's not the voice coil's DC resistance that really matters. It's the loudspeaker's impedance which for typical dynamic type loudspeakes will vary considerably. So if the amplifier's output impedance is high enough, it can result in some rather significant variations in the loudspeaker's frequency response.
Here's an example of how wild it can get using a SET tube amp driving Stereophile's dummy loudspeaker load which is 2uF in parallel with 8 ohms:
An externally hosted image should be here but it was not working when we last tested it.
Though the more fundamental context of "damping factor" is about the amplifier's "control" over the loudspeaker at resonance. Dick Pierce *wince* wrote a nice article about this in Speaker Builder some years back.
There's copy of it over on diyspeakers.net. Just look under Articles.
se
interesting but get this......
I overheard a teenager salesperson at Frys explaining to a middle aged couple that damping (he called it dampening) factor is from the utmost importance.
it can only be achieved in expensive stuff he said, it is the difference in travelspeed between high freq and low freq inside the system and must therefor be the same hence one uses damping factor to mark it. To bring the high freq speed down to the low freq speed is a complex issue and is very expensive!!!!!
I saw him selling a Sansui (if my memory serves me wright) with some spkrs ( i think Polk) and other stuff, probably around 3k ...... within a few minutes....
Jean-Pierre
I overheard a teenager salesperson at Frys explaining to a middle aged couple that damping (he called it dampening) factor is from the utmost importance.
it can only be achieved in expensive stuff he said, it is the difference in travelspeed between high freq and low freq inside the system and must therefor be the same hence one uses damping factor to mark it. To bring the high freq speed down to the low freq speed is a complex issue and is very expensive!!!!!
I saw him selling a Sansui (if my memory serves me wright) with some spkrs ( i think Polk) and other stuff, probably around 3k ...... within a few minutes....
Jean-Pierre
A loudspeaker box designer can take the DC resistance of the voice coil and cross-over inductors into account, such that the quality factor or factors of the low-frequency resonance(s) have the right values and a nice response results when the box is purely voltage driven.
Any resistance from the amplifier is not taken into account by the box designer and increases the quality factors above the optimal values.
An equally valid way to look at it is to assume the loudspeaker to be designed to have a more or less flat response from the voltage between its input terminals to the sound pressure, and to calculate the response deviations in loudspeaker terminal voltage due to frequency-dependent voltage division between the loudspeaker on one side and the amplifier and cable on the other side.
Anyway, the difference in response between a damping factor of 100 and a damping factor of 1000 is negligible, unless there are very deep troughs in the impedance curve of the loudspeaker.
Any resistance from the amplifier is not taken into account by the box designer and increases the quality factors above the optimal values.
An equally valid way to look at it is to assume the loudspeaker to be designed to have a more or less flat response from the voltage between its input terminals to the sound pressure, and to calculate the response deviations in loudspeaker terminal voltage due to frequency-dependent voltage division between the loudspeaker on one side and the amplifier and cable on the other side.
Anyway, the difference in response between a damping factor of 100 and a damping factor of 1000 is negligible, unless there are very deep troughs in the impedance curve of the loudspeaker.
damping factor of 200?
In connection with some discussion on this topic , someone mentioned a link to a test report on damping factors. I can't remember where I read this ( on the Net of course).
It said that at damping factors above 15 or so it becomes very hard if not impossible to hear any improvements. I would also guess that they assume that the real world impedance of the speaker system is reasonably uniform over its bandwidth.
( If we talk of the damping factor with reference to bass response only , the factor 15 makes more sense).
For systems with a very large variation in impedance ( over frequency ) a reasonably high damping factor will ensure that the ( electrical ) frequency response at the speaker terminals is really flat. With a low damping factor you will get a pretty variable response as can be seen in the wavy graph in the earlier post.
So the two factors are kind of tied to each other.
Can you imagine what must be happening at the speaker terminals with SE amps and similar circuits !
Most of the time we say something is good because we like it that way. That's good because it leaves the field wide open for DIY projects. We will never get to the end !
And all the time we are having fun !
Cheers.
ashok.
In connection with some discussion on this topic , someone mentioned a link to a test report on damping factors. I can't remember where I read this ( on the Net of course).
It said that at damping factors above 15 or so it becomes very hard if not impossible to hear any improvements. I would also guess that they assume that the real world impedance of the speaker system is reasonably uniform over its bandwidth.
( If we talk of the damping factor with reference to bass response only , the factor 15 makes more sense).
For systems with a very large variation in impedance ( over frequency ) a reasonably high damping factor will ensure that the ( electrical ) frequency response at the speaker terminals is really flat. With a low damping factor you will get a pretty variable response as can be seen in the wavy graph in the earlier post.
So the two factors are kind of tied to each other.
Can you imagine what must be happening at the speaker terminals with SE amps and similar circuits !
Most of the time we say something is good because we like it that way. That's good because it leaves the field wide open for DIY projects. We will never get to the end !
And all the time we are having fun !
Cheers.
ashok.
what about a zero damping factor...
What about current drive to a speaker? An ideal current-drive amp would have an infinite output impedance, and therefore a damping factor of zero.
I have no experience in this regard but remember reading somewhere that good results are to be had with the exception of controlling woofer resonance. My questions are:
a) Does anyone have experience with/knowledge of current-drive of speakers?
b) If this can indeed be successfully done, how is it possible with a zero (or near-zero) damping factor?
What about current drive to a speaker? An ideal current-drive amp would have an infinite output impedance, and therefore a damping factor of zero.
I have no experience in this regard but remember reading somewhere that good results are to be had with the exception of controlling woofer resonance. My questions are:
a) Does anyone have experience with/knowledge of current-drive of speakers?
b) If this can indeed be successfully done, how is it possible with a zero (or near-zero) damping factor?
Resonance damping
With a current source the electronic unit would not be able to control (damp) the resonance. In this case you will probably have to depend on the mechanical damping of the driver itself. This would mean that all drivers would not be suitable for current drive. You would need specially designed drivers with very high mechanical damping . Currently ,the electrical damping of speakers seems to be higher than the mechanical damping.
Cheers.
I once asked some speaker designers at Hitachi about why they did not use current drive when it was better than voltage drive. They seemed to be floored and obviously did not follow up on it .
( That was about 20 years ago )! Great guys in any case !
With a current source the electronic unit would not be able to control (damp) the resonance. In this case you will probably have to depend on the mechanical damping of the driver itself. This would mean that all drivers would not be suitable for current drive. You would need specially designed drivers with very high mechanical damping . Currently ,the electrical damping of speakers seems to be higher than the mechanical damping.
Cheers.
I once asked some speaker designers at Hitachi about why they did not use current drive when it was better than voltage drive. They seemed to be floored and obviously did not follow up on it .
( That was about 20 years ago )! Great guys in any case !
> how can internal impedance of a power amplifier be measured
Feed a constant signal, at less than maximum output. Load the output with no-load and with rated load. Measure the voltages you get.
For an easy measurement, try a very simple tube-triode amp. With no-load, crank the output up to 3 volts. Then apply an 8Ω load; the output will drop to about 2 volts. Be sure it is not distorting significantly in either case. Now a little math will tell you that the output Z is about 4 ohms, because 8/(4+8)= 2/3. If the Zout is 4 and the load is 8, the Damping Factor (I never saw DampENing Factor) is 8/4= 2.
With a modern "good" sand-state amplifier, you will read something like 3 volts no-load and 2.997 volts at 8Ω. Output Z is about ((3.000-2.997)/2.997)*8= 0.008Ω, DF=1,000. (At these absurd levels, the measurement may be limited by meter accuracy rather than the amp's actual performance.)
Some amps do not like no-load. Also it is probably not a real-world condition when driving a speaker. 50Ω and 8Ω (or same ratio for other design loads) is normally safe, though the exact math is harder.
Output Z is often measured around 1 volt, but usually varies with signal level. And output Z always varies with frequency.
Why do we care? If the speaker was always 8Ω, we might not care. But real speakers are 8Ω at 300Hz and very-high Ω at bass resonance. If the peak at resonance were infinity, then with DF=2 the voltage across the speaker terminals is 20*LOG(3/2)= 3.5dB higher at bass resonance than at 300Hz. If the bass resonance impedance were 50Ω, the rise is 2.85dB.
It is possible to make a speaker acoustic response "flat" from a hi-Z source. Most home radios of the 1930s-1950s were tuned this way. The undamped electrical rise at bass resonance was balanced against an open-back cabinet's and undersized transformer's bass droop. Some of these sound very good. But when you want a little better, it gets very difficult to design.
Most modern speakers are tuned to be flat with a constant voltage input despite speaker impedance variations. If the voltage is not perfectly flat: With a damping factor of 10, the worst error is 1dB; for DF=40 the worst is 0.25dB. Since speakers are not precision devices, not accurate better than 1dB or 0.25dB, DF more than 10 or 40 is usually plenty. 100, 200 is not a lot better (0.1dB, 0.05dB) but generally "easy" with high-feedback transistor amps.
Anyway, 10 feet (3M) of speaker wire will give DF about 40 even with an infinite-DF amplifier. And speaker designers generally assume a few-tenths ohms of wire resistance (not that it makes any large difference to their calculations).
> the amplifier's "control" over the loudspeaker at resonance.
That's the common explanation, and a lot better than that "difference in travelspeed" garbage the Fry's boy was selling.
But the "control" is limited by the 6Ω resistance of the voice coil. And the designer has "controlled" the resistance, magnet, mass, stiffness, and size to give a "slightly uncontrolled" response. If the speaker were perfectly damped, Q=0, the response would droop 6dB/oct below 200Hz. We need to let it resonate with Q of 0.5 to 2.0 (0.8 to 1.1 in most hi-fi designs), and the designer balances resistance against all the other parameters to get in that range. The difference between resistance of 6.1Ω (DF=80) and 6.01Ω (DF=800) is about zero for damping, about 0.1dB for response. Room-effects make much more difference than this.
> I think it has to do with how many output transistors we use.
Not really. The output impedance of a transistor (or a parallel array of transistors) is proportional to current. More transistors is the same as one transistor at the same total current. In theory Zout may be 0.03Ω at 1 Amp (DF=240), but 0.3Ω at 0.1 Amp (DF=24) and 0.6Ω at a typical AB idle current of 0.05 Amps (DF=12). When you add the need for bias stability resistors, output Z is usually 0.1Ω to 0.5Ω for about any transistor amp with any reasonable number of output devices and any usable thermal stability. And that assumes low-Z drive to the emitter followers, which is often not the case. A naked emitter-follower speaker amp has DF in the area of 20.
> 3 stages (differential, VAS, current amp)
In general you need 4 stages to get just a "good" speaker amp. If the "current amp" is a darlington or similar, you have enough current gain for a "good" amp. It is a little marginal for a "great high-feedback" amp or for DF over 1,000.
The naked output Z is probably higher than 0.2Ω because of the hi-Z VAS and finite Beta in output devices. To get DF over 1,000, you will need 1,000:1 feedback factor. Closed-loop voltage gain of speaker amplifiers is usually about 20. You need an open-loop voltage gain of about 20*1,000= 20,000. This is very hard to get in one gain stage. And it is hard to get gain in the diff-stage without reducing gain in the VAS. And if you take gain in two stages, you will have a 2-pole open-loop response which is unstable at high feedback factor. If you can compensate it to look like a 1-pole response, and want DF=1,000 all the way to 20KHz, the gain-bandwidth has to be 20MHz at 20:1 gain, 400MHz if you keep it 1-pole all the way to unity gain. You can not get big output transistors that are flat above 20MHz; good rugged ones may get limp before 1MHz. So demanding hi-DF usually leads to unhappy compromises in stability and feedback margin.
I must point out that the path to high DF is what led to many nasty-sounding transistor amps. The high DF applied only below 1,000 Hz, because feedback was falling (and distortion rising) above 1KHz. So you get ultra-clean bass with a cloud of fuzzy treble. This does not "have to" happen, and I'm sure your "certain professional studio amp" is better-balanced than the old stuff. The DF=1,000 probably "just happened" after all else was optimized for good full-range feedback. But starting from scratch, you may be following a 25 year path from the big bad stuff of 1970 to modern amps.
Feed a constant signal, at less than maximum output. Load the output with no-load and with rated load. Measure the voltages you get.
For an easy measurement, try a very simple tube-triode amp. With no-load, crank the output up to 3 volts. Then apply an 8Ω load; the output will drop to about 2 volts. Be sure it is not distorting significantly in either case. Now a little math will tell you that the output Z is about 4 ohms, because 8/(4+8)= 2/3. If the Zout is 4 and the load is 8, the Damping Factor (I never saw DampENing Factor) is 8/4= 2.
With a modern "good" sand-state amplifier, you will read something like 3 volts no-load and 2.997 volts at 8Ω. Output Z is about ((3.000-2.997)/2.997)*8= 0.008Ω, DF=1,000. (At these absurd levels, the measurement may be limited by meter accuracy rather than the amp's actual performance.)
Some amps do not like no-load. Also it is probably not a real-world condition when driving a speaker. 50Ω and 8Ω (or same ratio for other design loads) is normally safe, though the exact math is harder.
Output Z is often measured around 1 volt, but usually varies with signal level. And output Z always varies with frequency.
Why do we care? If the speaker was always 8Ω, we might not care. But real speakers are 8Ω at 300Hz and very-high Ω at bass resonance. If the peak at resonance were infinity, then with DF=2 the voltage across the speaker terminals is 20*LOG(3/2)= 3.5dB higher at bass resonance than at 300Hz. If the bass resonance impedance were 50Ω, the rise is 2.85dB.
It is possible to make a speaker acoustic response "flat" from a hi-Z source. Most home radios of the 1930s-1950s were tuned this way. The undamped electrical rise at bass resonance was balanced against an open-back cabinet's and undersized transformer's bass droop. Some of these sound very good. But when you want a little better, it gets very difficult to design.
Most modern speakers are tuned to be flat with a constant voltage input despite speaker impedance variations. If the voltage is not perfectly flat: With a damping factor of 10, the worst error is 1dB; for DF=40 the worst is 0.25dB. Since speakers are not precision devices, not accurate better than 1dB or 0.25dB, DF more than 10 or 40 is usually plenty. 100, 200 is not a lot better (0.1dB, 0.05dB) but generally "easy" with high-feedback transistor amps.
Anyway, 10 feet (3M) of speaker wire will give DF about 40 even with an infinite-DF amplifier. And speaker designers generally assume a few-tenths ohms of wire resistance (not that it makes any large difference to their calculations).
> the amplifier's "control" over the loudspeaker at resonance.
That's the common explanation, and a lot better than that "difference in travelspeed" garbage the Fry's boy was selling.
But the "control" is limited by the 6Ω resistance of the voice coil. And the designer has "controlled" the resistance, magnet, mass, stiffness, and size to give a "slightly uncontrolled" response. If the speaker were perfectly damped, Q=0, the response would droop 6dB/oct below 200Hz. We need to let it resonate with Q of 0.5 to 2.0 (0.8 to 1.1 in most hi-fi designs), and the designer balances resistance against all the other parameters to get in that range. The difference between resistance of 6.1Ω (DF=80) and 6.01Ω (DF=800) is about zero for damping, about 0.1dB for response. Room-effects make much more difference than this.
> I think it has to do with how many output transistors we use.
Not really. The output impedance of a transistor (or a parallel array of transistors) is proportional to current. More transistors is the same as one transistor at the same total current. In theory Zout may be 0.03Ω at 1 Amp (DF=240), but 0.3Ω at 0.1 Amp (DF=24) and 0.6Ω at a typical AB idle current of 0.05 Amps (DF=12). When you add the need for bias stability resistors, output Z is usually 0.1Ω to 0.5Ω for about any transistor amp with any reasonable number of output devices and any usable thermal stability. And that assumes low-Z drive to the emitter followers, which is often not the case. A naked emitter-follower speaker amp has DF in the area of 20.
> 3 stages (differential, VAS, current amp)
In general you need 4 stages to get just a "good" speaker amp. If the "current amp" is a darlington or similar, you have enough current gain for a "good" amp. It is a little marginal for a "great high-feedback" amp or for DF over 1,000.
The naked output Z is probably higher than 0.2Ω because of the hi-Z VAS and finite Beta in output devices. To get DF over 1,000, you will need 1,000:1 feedback factor. Closed-loop voltage gain of speaker amplifiers is usually about 20. You need an open-loop voltage gain of about 20*1,000= 20,000. This is very hard to get in one gain stage. And it is hard to get gain in the diff-stage without reducing gain in the VAS. And if you take gain in two stages, you will have a 2-pole open-loop response which is unstable at high feedback factor. If you can compensate it to look like a 1-pole response, and want DF=1,000 all the way to 20KHz, the gain-bandwidth has to be 20MHz at 20:1 gain, 400MHz if you keep it 1-pole all the way to unity gain. You can not get big output transistors that are flat above 20MHz; good rugged ones may get limp before 1MHz. So demanding hi-DF usually leads to unhappy compromises in stability and feedback margin.
I must point out that the path to high DF is what led to many nasty-sounding transistor amps. The high DF applied only below 1,000 Hz, because feedback was falling (and distortion rising) above 1KHz. So you get ultra-clean bass with a cloud of fuzzy treble. This does not "have to" happen, and I'm sure your "certain professional studio amp" is better-balanced than the old stuff. The DF=1,000 probably "just happened" after all else was optimized for good full-range feedback. But starting from scratch, you may be following a 25 year path from the big bad stuff of 1970 to modern amps.
I have to say, I really love the Crown amplifiers that I use in PA and studio use. Their DF is very high, and they are also high current amplifiers. One thing I love about the MT/MA series amplifiers is their ability to control high power (1200W) subwoofers at loads down to 2 ohms. They control the speakers very well and you always get good, tight controlled bass using Crowns. BTW, the Crown MT/MA amps have a Damping Factor of 1000 from 10Hz to 400Hz, which is where it is needed most for subwoofer use. If that's not enough for you, the Crown Studio Reference series amps have a Damping Factor of 20,000 from 10Hz to 400Hz. Talk about speaker control! These amps sound great on higher frequency as well. They don't have any "fuzziness" in the mids and highs that was referred to in a previous post. One last thing, the correct term acording to Crown is, "Damping Factor." This is right from their data sheets.
I was also thinking about the speaker, in a roundabout mental analogy with CFB op-amps...
But on a very basic level: magnetodynamic interaction between the voice coil and magnet is determined by the induced flux in the coil, which in turn is proportional to the current flowing through it...
Intuitively it would seem that current drive is the logical way to go...
But on a very basic level: magnetodynamic interaction between the voice coil and magnet is determined by the induced flux in the coil, which in turn is proportional to the current flowing through it...
Intuitively it would seem that current drive is the logical way to go...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.