About dynamic headroom (controversial topic :-)

[ski2xblack]

Unless I am mistaken, Proton made some of the NAD amps back in the late '80's, specifically the 'Monitor Series'. NAD's Power Envelope / PowerDrive are class G/H types, hence their ability to provide ridiculous IHF dynamic power ratings they publish. I'm not sure of the duration of the reserve power, it's not specified, but only described "available not just for brief transients but for the full duration of complex chords and musical climaxes."

I guess the question on which approach is better regarding power supply topology is wheather you need sheer V-8 power, or if you can get away with a four-banger with a turbo, to drive your particular speakers at your maximum desired listening level. Class G/H makes sense for variable (music) signals.

 

[jeannot]

Finally, a standardized definition of "Dynamic Headroom"

Background: The standard by which Dynamic Headroom is measured has been set by IHF which was replaced by the EIA, and is now the CEA. This text was found on the Audio Engineering Society web site, which I have in high estime. The link is http://www.aes.org/e-lib/browse.cfm?elib=4898

The EIA RS-490 (former IHF A-202) amplifier test standard includes a "dynamic headroom" test employing a 20-mS tone-burst. In an informal survey of musical recordings, power bursts were found with durations from a few milliseconds up to several hundred milliseconds, with an apparent clustering in the 80-200-mS range. Since the practical value of an amplifier depends on its ability to reproduce musical dynamics, a more useful power rating would be obtained by amending the dynamic headroom test to employ a 200-millisecond (or similar) tone-burst.

So it appears that the standard is 20ms pulses. (May be even only 1 pulse) Knowing how manufacturers go after appearances first, I think we can assume that this is how they measure their DH.

In another article by Meyer Sound at http://www.meyersound.com/support/papers/amp_power.htm, it appears that even from a temporary high power audio signal, that 20ms just doesn't cut it:

Meyer Sound’s research has found that, in order to reproduce music without compressing the signal, the power amplifier should be capable of maintaining reproduction of a sine wave at full amplitude (i.e. where the sine wave’s peak amplitude reaches the maximum available voltage swing without clipping) into its intended load for at least 500 milliseconds. Meyer Sound refers to the average power during this 500 milliseconds as“true burst power.” Peak power output should last at least 100 milliseconds in order to be useful for music reproduction.

When we think about it, how much does a loud movie sound like an explosion lasts, may be 1-2 seconds. 20ms represents ONE cycle of a 50Hz sine wave, which would be probably be unrecognizable as a sound.

I found other research finding anywhere between 80ms-300ms, but never under 80ms. So it appears that the believers of DH are victims of deception with the 20ms measurement, may be as much as people who do not believe in DH. Hey, misery loves company

 

[highfigh]

Seems to me that "dynamic headroom" is just another term for transient response capability, which is a more well known terminology in electrical engineering circles (at least for SATCOM systems), especially for power amplifier design. I'm surprised that Dr. Mark has not chimed in about the Quad amps-their current dumping design would seem to be able to handle transients better than other more standard designs. Does NAD have a particular design feature that results in their dynamic headroom performance, or is it simply their way of characterizing the transient response capability of their amps?

Except 'dynamic headroom' uses rated output as the starting point, not some arbitrary level.

 

[Pyrrho]

... they don't specify their definition of dynamic power. ...

In which case, the claim is totally useless. Unless they are extremely clear about the other power ratings (which is typically not the case), the only one worth paying any attention to is the continuous RMS output, as there are some regulations regarding what that means.

 

[3db]

I think they are different, because they are tested differently. Dynamic Headroom is tested with a sine wave lasting 50-200ms, and the measure (in decibels) is the ratio of where these signals clip divided by where a continuous identical sine wave clips.

I would say the transient response is more related to slew rate, and is measured in V/uS using square waves. The slew rate is a measure of the parasitic capacitances within the amplifier circuit, among other things.

Some points I want to clear up on power supplies.

Its absurd to think that adding more windings to the secondary is a way of achieving dynamic headroom. Add more windings to the secondary increases voltage capacity only and actually diminishes current capability. Look at the transformer as VI (primary) = VI (secondary) x (parastic losses through the core and resistance of the windings). But that can be fixed by adding a BFC at the output of the secondary? Well no... It can't. Even if the charging time is fast enough, its still going to limited by the capability of the transformer to supply current to charge the device.

Dynamic headroom cannot be pinned to just the transformer alone but also to the BFC at the filtering stage. The bigger the caps, the longer it can sustain the peak demands before collapsing.

It also depends on slew rate of the amplifier as well. Its useless to have a robust power supply when the slew rate of the power transistors cannot track the change in signal fast enough.

Dynamic headroom is more complex than just power supply designs.


Here's an article by Stereophile on dynamic headroom.

http://www.stereophile.com/asweseeit/489/

 

[KEW]

Interesting discussion.

If I understand correctly, (in loose terms) the transformer determines the RMS power and the capacitance determines the dynamic power.

To coach it in terms of equipment, I have always wondered if there was a situation which would suggest two UPA-1 over the XPA-2:

2ea Emotiva UPA-1's (numbers below are doubled to reflect two monos):
Standard price $700
Transformer Size: 600VA
Secondary capacitance: 160,000uF

1ea Emotiva XPA-2:
Standard price $800
Transformer Size: 1,200VA
Secondary capacitance: 45,000uF

While the XPA-2 is in the "big transformer, decent capacitance" camp, the UPA-1 represents the most "lopsided" ratio of capacitance to transformer I've ever noticed in an amp!

Unfortunately Emotive doesn't provide a Dynamic power rating (that I saw).

 

[M Code]

Some points I want to clear up on power supplies.

Its absurd to think that adding more windings to the secondary is a way of achieving dynamic headroom. Add more windings to the secondary increases voltage capacity only and actually diminishes current capability. Look at the transformer as VI (primary) = VI (secondary) x (parastic losses through the core and resistance of the windings). But that can be fixed by adding a BFC at the output of the secondary? Well no... It can't. Even if the charging time is fast enough, its still going to limited by the capability of the transformer to supply current to charge the device.

Regarding the power transformer windings...
This is actually more complex than just a voltage output, essentially the power transformer consists of multiple windings of copper wiring. And frequently the transformer maker will wind the transformer with higher guage wire (> number) so it can equal the voltage output to a transformer wound of lower guage.. Unfortunately the transformer wound with the higher gauge wire (lower cost) will heat up qwiker and saturate with its voltage dropping..

The higher gauge wire is cheaper as the worldwide pricing for copper has increased significantly over the last 3 years.. This is common practice in the lower cost AVRs (SRP <$699). Since the power transformer is the highest cost single component within the AVR.

Dynamic headroom cannot be pinned to just the transformer alone but also to the BFC at the filtering stage. The bigger the caps, the longer it can sustain the peak demands before collapsing.

The major contribution of the bigger capacitors is for extended low freguency demands, especially if the loudspeakers are full bandwidth and low impedance/sensitivity sealed box design..


Just my $0.02...

 

[jeannot]

The major contribution of the bigger capacitors is for extended low freguency demands, especially if the loudspeakers are full bandwidth and low impedance/sensitivity sealed box design..

Let's all keep in mind that while we're talking about the filtering caps helping in brief high current situations, the Dynamic Headroom test is done with a 20ms pulse. It is likely that even the skimpiest filtering caps would give similar results as large caps. I haven't done these calculations for years, so I may be mistaken: At 60V, a 15,000uF capacitor in a resistance of 13ohms (the approximate average for a sine wave in 8 ohms) has a discharge time of 200ms for 67%. Of course a 30,000uF cap would take twice as long, and it is unlikely that a 20ms dynamic headroom test show a difference between these two caps.

In that regard, I believe that the quality of the transformer and size of the filtering caps are not visible in high reading of a 20ms Dynamic Headroom test. The high reading probably only indicates the potential output power if the power supply was perfect. However, the low reading is heavily dependent on the transformer, and to a small extent the filtering caps (if they are exhausted in between the rectifier cycles). The more the power supply collapses in continuous, the higher the Dynamic Headroom.

Although it is possible that the output stage of an amplifier behaves slightly differently in the high and low tests, I think the Dynamic Headroom is mostly (not 100%) a measure of how incapable a power supply is.

 

[M Code]

Let's all keep in mind that while we're talking about the filtering caps helping in brief high current situations, the Dynamic Headroom test is done with a 20ms pulse. It is likely that even the skimpiest filtering caps would give similar results as large caps. I haven't done these calculations for years, so I may be mistaken: At 60V, a 15,000uF capacitor in a resistance of 13ohms (the approximate average for a sine wave in 8 ohms) has a discharge time of 200ms for 67%. Of course a 30,000uF cap would take twice as long, and it is unlikely that a 20ms dynamic headroom test show a difference between these two caps.

In that regard, I believe that the quality of the transformer and size of the filtering caps are not visible in high reading of a 20ms Dynamic Headroom test. The high reading probably only indicates the potential output power if the power supply was perfect. However, the low reading is heavily dependent on the transformer, and to a small extent the filtering caps (if they are exhausted in between the rectifier cycles). The more the power supply collapses in continuous, the higher the Dynamic Headroom.

Although it is possible that the output stage of an amplifier behaves slightly differently in the high and low tests, I think the Dynamic Headroom is mostly (not 100%) a measure of how incapable a power supply is.

Keep in mind..
@ the time Dynamic Headroom Specification was created almost 25 years ago, few sources were digital then the major one being the CD.. And as the leading amplifier engineers began to design for high dynamic range digital sources they found out about the other related byproducts of slew rate, rise time when the amplifiers were pushed hard, and engineers like Matti Otala began to talk about TIM distortion.

A similar thing happened in loudspeakers when the transducers were pushed hard by a high dynamic range signal. Early transducer designs were not capable of handling this without the cones breaking up and/or distorting due to non-linear actions.

Also lower cost audio products typically use a rubber band power supply and save $ by using smaller caps, regulators & transformers as the amplifier's continuous output is limited by the collapse of the power supply which constrains the available power output.

A well-designed amplifier is crucial along with a matching power supply with adequate VA (voltage/current) capability, so that it can deliver reasonable continuous power output plus have enough headroom for the high dynamic range streams.

Just my $0.02...

 

[jeannot]

A well-designed amplifier is crucial along with a matching power supply with adequate VA (voltage/current) capability, so that it can deliver reasonable continuous power output plus have enough headroom for the high dynamic range streams.

I'm facing much disagreement on this post about this. My contention is that a 200W continuous amplifier capable 200W peaks will have a better, more controlled and solid sound than if it could give 300W peaks. And if these numbers are in 8 ohms, the 200W peaks amplifier would probably be more powerful in 4 ohms than the 300W peak.
Feels wery lonely on this side of the fence.

I'm not sure I agree with matching between power supply/amplifier.

With the exception of when a limited amplifier needs the "protection" of a weak power supply, if one builds the perfect power supply, absolutely every single amplifier will work and sound better (to varying levels, yes) with it.

So if "matching" there must be, it would be limited to the amplifier, to be "careful what current it wishes for".

You may be referring to the balance, given a limited cost, between the investments in the power supply VS the amp circuitry.

 

[PENG]

In which case, the claim is totally useless. Unless they are extremely clear about the other power ratings (which is typically not the case), the only one worth paying any attention to is the continuous RMS output, as there are some regulations regarding what that means.

If you want to be precise it is average power, not rms power.

 

[PENG]

Background: The standard by which Dynamic Headroom is measured has been set by IHF which was replaced by the EIA, and is now the CEA. This text was found on the Audio Engineering Society web site, which I have in high estime. The link is http://www.aes.org/e-lib/browse.cfm?elib=4898



So it appears that the standard is 20ms pulses. (May be even only 1 pulse) Knowing how manufacturers go after appearances first, I think we can assume that this is how they measure their DH.

In another article by Meyer Sound at http://www.meyersound.com/support/papers/amp_power.htm, it appears that even from a temporary high power audio signal, that 20ms just doesn't cut it:



When we think about it, how much does a loud movie sound like an explosion lasts, may be 1-2 seconds. 20ms represents ONE cycle of a 50Hz sine wave, which would be probably be unrecognizable as a sound.

I found other research finding anywhere between 80ms-300ms, but never under 80ms. So it appears that the believers of DH are victims of deception with the 20ms measurement, may be as much as people who do not believe in DH. Hey, misery loves company

Good findings, thanks for sharing.

 

[PENG]

Regarding the power transformer windings...
This is actually more complex than just a voltage output, essentially the power transformer consists of multiple windings of copper wiring. And frequently the transformer maker will wind the transformer with higher guage wire (> number) so it can equal the voltage output to a transformer wound of lower guage.. Unfortunately the transformer wound with the higher gauge wire (lower cost) will heat up qwiker and saturate with its voltage dropping..

The one that uses higher gauge number(thinner one) can equal the lower gauge one's voltage output but not current output so the one that uses high gauge number wire, everything else being equal WILL not have its power output rating equal to the one with lower gauge number=thicker wire one. That's basical electrical theory and I am sure you agree but I just want everyone to be clear.

 

[PENG]

Interesting discussion.

If I understand correctly, (in loose terms) the transformer determines the RMS power and the capacitance determines the dynamic power.

To coach it in terms of equipment, I have always wondered if there was a situation which would suggest two UPA-1 over the XPA-2:

2ea Emotiva UPA-1's (numbers below are doubled to reflect two monos):
Standard price $700
Transformer Size: 600VA
Secondary capacitance: 160,000uF

1ea Emotiva XPA-2:
Standard price $800
Transformer Size: 1,200VA
Secondary capacitance: 45,000uF

While the XPA-2 is in the "big transformer, decent capacitance" camp, the UPA-1 represents the most "lopsided" ratio of capacitance to transformer I've ever noticed in an amp!

Unfortunately Emotive doesn't provide a Dynamic power rating (that I saw).

Kew you raised an interesting point. Food for thought, the XPA-2's bigger transformer has one advantage, especially if it has only one big secondary winding, in that it will get you more juice for 1 channel output.

Overall monoblock is better when you get to the point of having 600VA per channel. It certainly should help produce better cross talk numbers.

 

[mtrycrafts]

lol, (wipes coffee spray off the monitor after seeing the boom boxes) I gather you mean your "other stuff".

Steve

Well, yes. I am kinder to my boomboxes. They are on their own 15A circuits Now that I think of it, one is on a 20A.

 

[jeannot]

If I understand correctly, (in loose terms) the transformer determines the RMS power and the capacitance determines the dynamic power.

Yes. Their effect does overlap a little bit though.

To coach it in terms of equipment, I have always wondered if there was a situation which would suggest two UPA-1 over the XPA-2:

2ea Emotiva UPA-1's (numbers below are doubled to reflect two monos):
Standard price $700
Transformer Size: 600VA
Secondary capacitance: 160,000uF

1ea Emotiva XPA-2:
Standard price $800
Transformer Size: 1,200VA
Secondary capacitance: 45,000uF

While the XPA-2 is in the "big transformer, decent capacitance" camp, the UPA-1 represents the most "lopsided" ratio of capacitance to transformer I've ever noticed in an amp!

Unfortunately Emotive doesn't provide a Dynamic power rating (that I saw).

The standard Dynamic Headroom 20ms pulse test may not show the difference in capacitance anyways.

Even though as PENG pointed out the advantages of crosstalk VS bigger supply availability may balance each other, I too would go monoblocks. Since the high power frequencies are concentrated where there is little stereo separation, the stereo amps pretty much split the power 50/50 anyways. But if one decides to drive the sub and the center for example, monoblocks may provide better sleep.

 

[3db]

The one that uses higher gauge number(thinner one) can equal the lower gauge one's voltage output but not current output so the one that uses high gauge number wire, everything else being equal WILL not have its power output rating equal to the one with lower gauge number=thicker wire one. That's basical electrical theory and I am sure you agree but I just want everyone to be clear.

Agreed as I eluded too in an earlier post.

 

[3db]

Regarding the power transformer windings...
This is actually more complex than just a voltage output, essentially the power transformer consists of multiple windings of copper wiring. And frequently the transformer maker will wind the transformer with higher guage wire (> number) so it can equal the voltage output to a transformer wound of lower guage.. Unfortunately the transformer wound with the higher gauge wire (lower cost) will heat up qwiker and saturate with its voltage dropping..

I've already stated that but you failed to mention the side affect of using thinner wire but more of it is reduced current capability of the secondary. Heating of the wire isn't core saturation. Heating of the wire is an I^2 xR loss. Core saturation occurs when the flux of the core cannot contribute to producing any more current.

The major contribution of the bigger capacitors is for extended low freguency demands, especially if the loudspeakers are full bandwidth and low impedance/sensitivity sealed box design..


Just my $0.02...

Thats one of the contributions. It also allows the instantaneous rush of current found in transients that a power transformer simply cannot supply in short order.

A good stout power supply needs large capacitors and a good transformer. Skimping on either one of these components compromises the design equally.

I also want to reiterate once more the fact of slew rate as I feel this spec is over looked. If the slew rate is low of the amplifier section, then not even the stoutest of power supplies will rescue its performance. The signal will clip.

 

[3db]

I want to change my last sentence about the amp clipping. It won't clip but it will skew or round the transient waveform severly to th point where it doesn't even sound the same.

 

[jeannot]

I want to change my last sentence about the amp clipping. It won't clip but it will skew or round the transient waveform severly to th point where it doesn't even sound the same.

That is what TIM is. It's the amplifier input stages running open loop, because the feedback signal is not back yet due to the amp low bandwidth.

Interestingly, when the war on THD started in the mid-70s (remember the ground-breaking 0.5% THD specs?) the TIM skyrocketed, because manufacturers went after high open loop gain and high feedback with the fewest transistors. THD was good, but amps sounded horrible because of TIM, and the domination of odd-harmonic distortion (which tubes had almost none of)

I noticed that the more transistors an amp has, the lower its open loop gain. It puzzled me at first, then realized that the added transistors act as current sources and pullers for the pushers, making each stage more linear. Hence, less need for feedback.