• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

How important is doubling down power to 2 ohm or 1ohm on Hi Fi Speakers on high dynamic range music?

ahofer

Master Contributor
Forum Donor
Joined
Jun 3, 2019
Messages
5,100
Likes
9,281
Location
New York City
Depends on the speakers - if the speakers have impedance down below 4 ohm, then 2 ohm doubling down from 4 ohm is important - but if they don't.... you need them to linearly "double down" to as low as your speakers go... anything more is wasted. (but nice to have for potential future speakers?)

Having said that - my current main's are run by Crown XLS2500 class D (drivecore amps) - 440W@8ohm, 775@4ohm, 1200@2ohm - no they do not double down.... but they have ample headroom at 2 ohm, for my speakers, which do go down to 2 ohm.
I don't think this is right. As we've seen further up thread, many "doubling" measurements are achieved by understating the 8 ohm performance.

What I think is important is that the amp have sufficient power for the load. If you have speakers that dip under 4 ohms, you want enough power to generate high sound pressure levels at that load (see speaker sensitivity). If the speaker has a large phase angle at the 4 ohm load, you need good heat dissipation. Thus a 400 watt (8 ohm) amp that merely produces 400 watts at 4 ohms is likely to be just fine. With classes outside A/AB, you will see different power responses to lower resistance. This should still be fine, as long as you get enough power for your speaker at the necessary resistance.
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,381
Likes
736
you need them to linearly "double down" to as low as your speakers go... anything more is wasted...Having said that - my current main's are run by Crown XLS2500..they do not double down....but they have ample headroom at 2 ohm

On the one hand, I can't agree with the "wasted" phrase. Speakers and especially amps are more complicated than what we measure. An amp that doubles power-tested only into resistors let us recall-likely has a stronger power supply which can handle capacitive or inductive phase angles. So a "4 ohm" speaker might need "2 ohm doubling" or even "1 ohm doubling" to not clip into the real speaker. On the other hand:
- You'd have to see actual clipping tests of the amp to see if it truly doubles. Almost all that are rated to double actually do not; they are simply under-rated at the higher impedances. I've only ever seen a few that really double at clipping out of hundreds of home amp reviews and automotive amps I tested myself: PS Audio's S300 and M700 and M1200, some Mark Levinson gigantoid thing in the distant past, and a Devialet Expert 140.
- As you say, the other point is if you have massive power, the doubling doesn't matter, like in the case of the Crown noted there. Someone in a past thread had pointed out what I would rephrase as "the clipping behavior doesn't matter if you have so much power you never clip." In other words, the most perfect 100W per channel amp will probably clip if you want to really crank average speakers, but a non-doubling 1000W amp probably won't.
 
Last edited:

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,381
Likes
736
I love the sound of the US 155mm artillery. It just doesn't get old. ...I like the sound. Like a 50 cal ripping off 500 rounds. Mortars, rockets oh heck, I like the sound of all of it!
Love the spell of napalm in the morning! And yes, a lot of folks should wear ear protection but don't. Either because they are really ignorant about hearing damage, which many people are, or the same kind of macho attitude which sees extreme skateboarding etc without helmets.
 

Head_Unit

Major Contributor
Forum Donor
Joined
Aug 27, 2018
Messages
1,381
Likes
736
Try to make a realistic load that has the same properties at 4 and 8 Ohms? A power cube can't and though it may tell a designer something, it does not give marketing anything they can put on a glossy back page ad or "fact" sheet ( the long narrow strip in a roll) given to reviewers so they have something to quote and make up a story about.
... We can pretty much identify what will sound like garbage. We can pretty much identify a group that is worth looking at. We have not yet, I do not believe, reached a understanding where they all sound the same.
Actually some companies (led by Rockford Fosgate) did use Power Cube graphs in their marketing and training to beat up on other companies. What do you mean by "same properties at 4 and 8 Ohms"?

Your second statement is a key. To me, what ASR achieves is to weed out some crap, stuff that has weird problems or high noise levels that can make a quiet CD noisier, uneven speaker response etc. In these hard-to-audition-in-person days that's a service, even if no we do not know how to "measure better sound." Unfortunately then some people attack on a basis of "Those people believe measurements are everything!!!!!" which is not how I take ASR at all.
 

ahofer

Master Contributor
Forum Donor
Joined
Jun 3, 2019
Messages
5,100
Likes
9,281
Location
New York City
On that video, keep in mind that they are measuring peak power. A 300W amplifier will give 600W peaks. Some 200W amplifiers might be able to do 300W dynamic power, and so also able to do 600W peaks. But still I don't understand that video since the Pan Sonic track is not dynamic.

Here is another CH-precision video that shows even higher power peaks. But the crest factor on that track is is >25dB so no wonder! I don't worry if that track would clip on my system because the crest factor is so out of the ordinary.
Incidentally, we had quite a debate up-thread about this. These meters are likely measuring apparent power, thus giving a very impressive reading at frequencies with wide phase angles, but not requiring the actual power delivery implied. There is a famous video of Harbeth 40s drawing 500 watts with some techno music that just happened to be blasting rhythmically at the precise frequency where phase angle was widest.

 

killdozzer

Major Contributor
Joined
Aug 2, 2020
Messages
1,615
Likes
1,636
Location
Zagreb
Dynamic range is a function of noise floor (SNR) and power output. Different manufacturers may specify amplifiers differently, even within their own product line. A conservatively specified amplifier may actually provide more power than one rated for higher power especially if it is specified for low-impedance loads and "doubling down". Thermal (heat) requirements, power supply capacity, output device safe operating area, and a myriad of other factors come into play. The opposite is also true; an amplifier not rated to "double down" all the way to 2 ohms may provide more power but simply not have the heat sink and power supply to sustain long-term high-current output for low loads. So it might not be rated to "double down" but in real life may sound far more powerful.

Like many things, there is not a simple single specification that tells everything about the amp.
But is power consumption predictable? On the speaker's end, if you had all the specs, could you precisely determine your real world headroom needs and set them apart from what is simply an excess? Could you differentiate among under powered speakers and idle power never being used, easily?

I'm asking because I have some plans on changing my amp. I'm trying to guess right what I need.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,952
Likes
16,820
Location
Monument, CO
But is power consumption predictable? On the speaker's end, if you had all the specs, could you precisely determine your real world headroom needs and set them apart from what is simply an excess? Could you differentiate among under powered speakers and idle power never being used, easily?

I'm asking because I have some plans on changing my amp. I'm trying to guess right what I need.
Yes, within reason, and assuming you have all the prerequisite parameters. But you can almost always get a reasonable estimate with more limited information. Here are the steps to gain a ballpark estimate:

First, look up the sensitivity of your speakers, usually expressed in dB/W/m. You also need to know their nominal impedance in ohms, the distance you sit from the speakers, and how close they are to the walls around them. It is helpful to know their minimum impedance to ensure your amplifier can handle that, but generally the assumption is that minima are fairly narrow in frequency (and large sonic peaks short in duration) so an amp rated for their nominal impedance will be able to handle the minima in impedance without overstress.

Next, estimate your average listening level, either measuring with an SPL meter or using an SPL comparison table (the calculator site has a table). You can use an online calculator like this: http://myhometheater.homestead.com/splcalculator.html to estimate how much power is required to reach your average listening level. You can also use it to estimate your peak power needs if you know how loud you ever expect it to be.

A rule of thumb for peak-to-average levels in music is 17 dB, a power factor of 50. That means you should have 50 times the average power to handle peaks without clipping. Much music may have less than that dynamic range, and movies are reputedly higher (perhaps 20 dB, a factor of 100x in power, though some have reported 30 dB, a factor of 1000x). I tend to discount movie peaks somewhat as higher distortion and even hard clipping on an explosion or gunshot is not likely to be noticed.

I listen at average levels closer to 70 dB, maybe 75 dB for some music, and peaks around 90~95 dB typically. I find 80 dB average very loud but YMMV. I chose amplifiers to reach 105 dB (per THX) just to be allow for louder than I ever expect to listen, but don't really expect to hit those levels. I also use subs so I know the midrange power is likely significantly (say 10 dB, a factor of 10 in power to 20 dB -- 100x in power) less (see equal-loudness curves).

Armed with that knowledge you can choose the amp you need.

IME most people overestimate their average listening power needs, but underestimate how much extra is needed for peaks. That is, they underestimate the peak-to-average power ratio. Many folk (again IME) think they listen at an average power of 10 to 50 W whereas it is really only a few watts (often close to 1 W). However, they think 3 dB to 6 dB of headroom is sufficient, but it is more like 15 to 20 dB. So they think they are using 50 W average and need 3 dB headroom (100 W), but in reality it is more like 1 W average and still 100 W max for (20 dB) peaks.

IME/IMO/FWIWFM/my 0.000001 cent (microcent)/YMMV/etc. - Don
 

Kijanki

Active Member
Joined
Dec 16, 2021
Messages
106
Likes
69
Perfect voltage source has zero output impedance. Perfect doubling of power at half impedance suggests extremely low output impedance and it is most likely achieved by use of deep negative feedback. Increasing negative feedback from 20dB tp 60dB lowers output impedance 100 times. Deep negative feedback can bring Transient Intermodulation Distortions (TIM) that adds unpleasant high order odd harmonics (bright sound). Class D has inherently low output impedance (one output Mosfet always shorting to low impedance - GND or VCC) while signal path is short (small delay), but for everything else I would stay away from amplifier that exactly doubles power at 4 ohm vs 8 ohm.
 

MakeMineVinyl

Major Contributor
Joined
Jun 5, 2020
Messages
3,558
Likes
5,878
Location
Santa Fe, NM
Class D has inherently low output impedance (one output Mosfet always shorting to low impedance - GND or VCC) while signal path is short (small delay), but for everything else I would stay away from amplifier that exactly doubles power at 4 ohm vs 8 ohm.
Its not nearly as tidy as you might imagine. A couple of ways where an amplifier can double its power into lower impedances is:

1) Simply under-rate the amplifier's power spec at higher impedances so that power at lower impedances power "doubles".
2) Use a higher voltage power supply in the driver stages vs the output stage. This amplifier uses that approach.

Neither of these involve massive feedback as you state, and the assumption is made that the power supply capacity and output transistor complement of the amp in question can actually support the increase in current into lower impedances.
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,952
Likes
16,820
Location
Monument, CO
:):):)

Not that easy. I'm interested in peaks at the impedance dips. And the speakers are supposedly even more hungry at the dip "edges", so to speak.
You'd have to know how your speaker's sensitivity was measured (assuming you did not do it yourself). There are probably lots of ways, but the two I have seen most often are to apply 1 W (or, 2.83 Vrms) at 1 kHz and measure the output using a calibrated mic and signal chain from 1 m away. The other way is similar but uses pink noise (broadband weighted noise) instead of a 1 kHz tone. Recently the spot-checks I have done on speakers, when I could find the info at all, used a 1 kHz tone.

If the sensitivity was measured at 1 kHz, then you can look at the speaker's frequency response and impedance graphs to calculate the power required at other frequencies. If you dig deeper you'll want to get impedance real and imaginary (resistive and reactive) components using the magnitude and phase from the impedance plot to calculate real and apparent power. Depends on how deep down the rabbit hole you want to go. I suspect if you were really deep into the engineering side of things you wouldn't need to ask, so for an approximation I'd assume the number is for 1 kHz, look at the impedance magnitude there, then just use the impedance magnitude and frequency response at other frequencies to estimate the maximum power required relative to the 1 kHz value.

Of course, then you might run into the difference between anechoic and in-room response, since e.g. a 3 dB difference is a doubling (or halving) of power. Then factor in the frequency response of recordings and how we hear (loudness curves again) to estimate the relative power needs over frequency based upon the target SPL over frequency. You can easily turn this into a life's work.

In the real world I am not sure it's worth the effort unless you have very troublesome speakers and/or amplifier(s). Most any amplifier can (should) handle a halving in impedance and/or doubling in power for a brief peak. Peak power is almost almost required in the deep bass where we are less sensitive to the sound and thus tend to crank it up. I have subs for that.

HTH - Don
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,952
Likes
16,820
Location
Monument, CO
Perfect voltage source has zero output impedance. Perfect doubling of power at half impedance suggests extremely low output impedance and it is most likely achieved by use of deep negative feedback. Increasing negative feedback from 20dB tp 60dB lowers output impedance 100 times. Deep negative feedback can bring Transient Intermodulation Distortions (TIM) that adds unpleasant high order odd harmonics (bright sound). Class D has inherently low output impedance (one output Mosfet always shorting to low impedance - GND or VCC) while signal path is short (small delay), but for everything else I would stay away from amplifier that exactly doubles power at 4 ohm vs 8 ohm.
Maybe, but that (very high feedback to create low output impedance to enable doubling-down in power at low impedance) has not been my experience (albeit audio amps are not my day job). Low output impedance and high feedback are not necessary for the amp to "double down" into lower-impedance loads. For example, a class A amplifier can do it with high enough standing bias current with no change in feedback or output impedance. More likely IME is that the amp is "underrated" at 8 ohms to allow for doubling at lower impedance, mainly for marketing purposes, or the voltage rails are high but thermally the amp won't handle the higher power steady-state (essentially the same thing).

Class D uses feedback to provide low distortion, and modern designs close the loop around the big LC output filter required to get the high-frequency switching noise to acceptable levels, so they also have lots of feedback -- often more than class AB in my limited experience, because they often have higher loop gain-bandwidth so can use greater feedback than more conventional A/AB designs. Without feedback, the output impedance of class D is limited by the series resistance of the output switching devices and the filter inductor so would be comparable to many other amplifiers.

FWIWFM - Don

Edit: @MakeMineVinyl posted as I was typing; what he said.
 
Last edited:

Kijanki

Active Member
Joined
Dec 16, 2021
Messages
106
Likes
69
Its not nearly as tidy as you might imagine. A couple of ways where an amplifier can double its power into lower impedances is:

1) Simply under-rate the amplifier's power spec at higher impedances so that power at lower impedances power "doubles".
2) Use a higher voltage power supply in the driver stages vs the output stage.

Neither of these involve massive feedback as you state, and the assumption is made that the power supply capacity and output transistor complement of the amp in question can actually support the increase in current into lower impedances.
When amplifier has DF=10 it implies that 10% of output power is lost in output impedance at 8 ohms and 20% at 4 ohm - difference of 10%. This difference will be apparent not only at max power but also at any output power (no PS limitation). Choosing amplifier with DF=100 would reduce this difference to 1% - practically doubling, but at DF=100 amplifier has deep negative feedback for sure. My AHB2 has DF=350, suggesting also deep NFB, but its feedback is non-recursive (no TIM). Also, very low output impedance might not be always desired and could cause brightness with some speakers that have low impedance at high frequencies, like electrostatic speakers. I'm not sure if very low output impedance is even necessary to obtain good sound. There are amps that sound great, got all sorts of awards, and have DF in order of 1 or 2.
 
Last edited:

EdW

Senior Member
Joined
Jul 5, 2020
Messages
355
Likes
452
Location
Cambridge, UK
But is power consumption predictable? On the speaker's end, if you had all the specs, could you precisely determine your real world headroom needs and set them apart from what is simply an excess? Could you differentiate among under powered speakers and idle power never being used, easily?

I'm asking because I have some plans on changing my amp. I'm trying to guess right what I need.
i too am mulling over the right spec for amplifier current delivery. A little time ago I thought that taking the lowest impedance that your chosen speaker exhibits at low frequencies is likely the region where your amplifier would have deliver most voltage swing and therefore demand the most current. Not really too concerned about impedance troughs at high frequencies since with real music no appreciable power is involved. Obviously the amplifier must be stable with a difficult high frequency load. Perhaps this is a good enough set of requirements. But a couple of other thoughts:

1. With a class AB power amps an awkward phase angle can further stress the amplifier (thermally) - see Keith Howard’s article in Stereophile:


but I’m considering a class D Purifi amplifier where this shouldn’t be a problem.

2. Douglas Self in his book “Audio PowerAmplifier Design“ shows that it is possible with the complex impedance of a real speaker to postulate a rather artificial asymmetric drive waveform which causes huge current spikes. He dismisses this as not a real problem for a good amplifier. On the basis that his knowledge is a lot better than mine I’ll relax a little :)

3AEA592B-6195-4EC1-AF6A-C8B052A37A9F.png
 

Kijanki

Active Member
Joined
Dec 16, 2021
Messages
106
Likes
69
As for class D, yes output impedance is limited by Zobel network and transistors in series, but it is usually very low likely in order of 0.5 ohm total.
Atma-Sphere specifies both amount of feedback and output impedance in new class D amp. Feedback is 35dB and output impedance is 0.01 ohm (DF=800). Since output impedance is lowered in linear fashion by the feedback it would be 56x (35dB) higher without any feedback, resulting in 0.56 ohm output impedance (DF=14). I would be hard to do with class A or AB amp without any feedback. We need to remember that these amplifiers, even without global NFB, have many local NFB. Even emiter (or cathode) resistor is a form of NFB.

35dB is not very deep and likely 20dB would be plenty since class D amps also have other forms of feedback (timing feedback), at least Icepower does.
 

dlaloum

Major Contributor
Joined
Oct 4, 2021
Messages
3,252
Likes
2,519
First the obvious admission - I am not an Electrical Engineering Geek.....(yes I did do some in Uni days but that was a long time ago)

Typical sensitivity measurements are done at 1W or 2.83Vrms... the former is dependent on impedance, which varies with frequency... the latter is not... but I doubt that most speaker sensitivity measurements are done wideband (using spots, sweeps, or pink noise)

I haven't yet been down the rabbit hole of working out what SPL I actually need at each part of the frequency spectrum... and I have a feeling that knowing this will potentially change my perspective of what amp power/impedance handling my system/speakers require.

With regards to "doubling down" - I have noted that several well regarded high current designs, (which don't double down) have the following factors:

Power @ 8ohm
Power @ 4 ohm = (Power @ 8 ohm) x 1.7
Power @ 2 ohm = (Power @ 4 ohm) x 1.5

I wonder whether in real life, the above is what most "double down amps" actually deliver, and they simply quote the 8ohm figure more conservatively to boost the 4 ohm and 2 ohm figures?

Crown XLS2500/2502 have 440W / 775W / 1200W - if it were to be quoted in "double down" marketing mode that would be 300W / 600W / 1200W
 

killdozzer

Major Contributor
Joined
Aug 2, 2020
Messages
1,615
Likes
1,636
Location
Zagreb
@EdW very interesting. Thank you.

Watt being what is expensive, I'm wondering can I go weaker with my new purchase, but it would seem I shouldn't. I should stick in the 90-110w range. Mr. Oclee said it's not even the lowest impedance dip that is most power hungry.

Later edit: here it is "the most difficult frequencies are either side of the impedance dip where the magnitude is still fairly low and the phase angle deviates from zero. Such impedances make the amplifier output stage work much harder and can cause protection systems in the amp to trigger"

So there's more to worry than the dip.
 
Last edited:

dlaloum

Major Contributor
Joined
Oct 4, 2021
Messages
3,252
Likes
2,519
Can anyone suggest a link to a primer on the phase angle impact on amplifiers? I "get" impedance, but am trying to wrap my head around how phase angle (which I though was changes in timing?) would make things difficult for an amp...
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,272
Likes
17,283
Location
Riverview FL
Can anyone suggest a link to a primer on the phase angle impact on amplifiers? I "get" impedance, but am trying to wrap my head around how phase angle (which I though was changes in timing?) would make things difficult for an amp...

 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,952
Likes
16,820
Location
Monument, CO
The hand-waving explanation is that phase determines how voltage and current "peaks" are related. If the phase angle is 0 degrees, voltage and current are exactly in phase, and the power factor is 1. Think of a sine wave with both voltage and current in phase so the curves overlie. Power peaks where the sine waves peak, voltage and current, then falls off since they both "fall" together. The amplifier's power demand is also 1 (relatively).

As the phase angle increases, voltage and current peak at different times, sort of "spreading" the power (the product of voltage and current) over more of the cycle, and apparent power increases. At 45 degrees, current and voltage peaks are a quarter-cycle apart, and you get a series of peaks one after another (alternating current and voltage peaks). The power factor drops to sqrt(2)/2 = 0.7071, meaning real and reactive power are different due to the phase difference, but the amplifier actually delivers twice the average power (power demanded is 2). Instead of two peaks over each cycle (one positive, one negative) now you get four. As the angle goes above 45 degrees the difference in current and voltage peaks lessens again and the amplifier's power drops back towards 1.

HTH - Don

Edit: This is not technically rigorous, take "hand-waving" literally, but helps provide a picture that might help understand why phase matters.
 
Top Bottom