How much power do you need?

[goldenpiggy]

This is how I answered this question for myself (please correct me if I'm wrong):

The dynamic range in music varies greatly, but a rule of thumb I've seen is ~10LU with 1LU ~ 1.3db soooo - lets just say 12db of padding to be really safe.

I enjoy listening with average DB around 80-85... add to that the head room and it equals 92 - 97db.

Next we examine efficiency of the speakers - mine are 85db at 1W at 1m.

Next we figure out what it would take to get to 97db with my speakers (keeping in mind that we need double the power for every 3db) 88db = 2W, 91db = 4W, 94 = 8W, 97db = 16W ...... at 1 meter, but I don't listen at 1 meter...

So next we need to add the listening distance (wich I believe quadruples the power for double the distance) so at 2M = 64W and at 4M = 256W...

And all of that assumes into 8ohm speakers... (some amps produce 2x power into 4ohm speakers, and some AV receivers actually lose power into 4 ohm speakers).... yadayadayada

Annnnnyyyways - you can see by the quick calculation - to be confident of not having clipping - it's not unreasonable to have a 200W amp or bigger.

I think you nailed it. Most of us crank the volume either for more bass or because the recording has such extreme dynamic range (e.g. classical music.) It obviously takes more power to boogie the LF drivers. Our hearing sensitivity is not linear, and is not nearly as sensitive in the lower octaves, so crank it up even more. As one operates at or below a driver's resonant frequency, it sucks a lot more power -- you're just heating up the voice coil at that point. As much as I don't like to brick wall things, when I do live sound, I'll always use 48dB/oct slope on the sub's high pass for exactly this reason. Go just a few Hz below Fs and it can easily draw 2-3x the power, and before you know it, you're clipping or bottoming out the drivers. But most people usually won't see this in a home environment.

The loading on the speaker enclosure also comes into play. If you're at 1/4 space (in a corner), it'll be louder in the lower octaves than half space (just a back wall in a very wide room), which is like doubling the power in some freq.

Finally, not every speaker behaves like a point source or abides by the -6dB SPL loss with doubling of distance. Line columns (not to be mistaken with line array) work on cylindrical wavefront and (theoretically over certain freq range) only loose half as much over distance. You don't see many line columns in the home though.

 

[Koeitje]

Thank you NTK, that document is indeed useful.
View attachment 120827
View attachment 120828

It's amazing to me just how LITTLE power is required for every day listening.

Using https://www.crownaudio.com/en/tools/calculators I'm showing that my setup requires only 10w to stay under the 85db threshold for long term hearing damage.

View attachment 120830
On the other hand, it's also crazy how quickly power levels increase. 3kw at 110db aka "rock concernt / full orchestra (must be drums)

View attachment 120831

That all be said, this doesn't really answer my question. 10w at 85db... but 85db of bass is a different power level than 85db of treble right? So if I wanted to be accurate, I would need to know how much power I needed drive my woofers to the db I wanted.
View attachment 120832

Anyways, at this pointed I'm leaning toward the extra power of the Purifi.

Those numbers are for fairly efficient loudspeakers, try doing the same down to 85 or even 83dB sensitivity.

 

[RayDunzl]

I have read the on average music has transient peaks that are 6 to 25 dB above the average level.

If you have a mic and REW, you can observe your own Leq and Peak levels...

REW SPL meter, set to integrate unweighted Leq and display unweighted peak.

18.9dB difference, indicating my 20dB fudge factor is useful for what I tend to listen to critically.

 

[jshelbyj]

I got a Purifi amp today and A/B tested vs the smsl.

Volume matched each amp at 80db @ 2m.

Honestly the 20 watt smsl is more than enough for apartment listening. I think one would need a big room or outdoors to need more power.

 

[AdamG]

If you have a mic and REW, you can observe your own Leq and Peak levels...

REW SPL meter, set to integrate unweighted Leq and display unweighted peak.

18.9dB difference, indicating my 20dB fudge factor is useful for what I tend to listen to critically.

View attachment 122304

Been there done that. I rarely ever crank it up anymore. My biggest power consumers are my Subs. They are feed with 2400W amps. The days of ear bleeding levels are behind me. My number one focus now is preserving the “hearing” I have left.

My point was the needed reserves of power or head room can be quite substantial and mostly under estimated. With the new class d amps I have 450/500 Watts at 4 ohms on tap. They exceed my head room needs.

 

[pma]

When listening to Tchaikovsky "Pathétique", Decca 483 0656, about 47W equivalent sine power in fff. I have recorded the voltage wave from amplifier output during listening. So at least 2 x 100W amplifier is a must, to me.

 

[dlaloum]

I run a set of Gallo Reference 3.2's.... the tweeter is a capacitive load, and drops down to around 1.6 ohm - the rest of the frequency range the load is relatively benign.

I have run these speakers using 140W/Channel (8ohm RMS) Quad 606 amplifier - and it sounded good... I have run it from 140w AVR's and it didn't sound as good ... my assumption is that the power supply in the AVR is more constrained, so when the load gets difficult the Amp can't quite stretch up to it...

I talked to the speaker designer (Anthony Gallo) - he recommended substantially more power, and loads of current - ie ability to handle the low impedance without stress. He demoed the speakers using the 500W/channel Spectron class D amps.

The Spectron was well beyond my means at the time, but there were discussions about the Crown XLS series of Drivecore amps.... the XLS2500 puts out an easy 440W / channel into 8ohms.... and is available quite cheaply on the used market.

So I went from a well regarded (albeit "classic" now) audiophile amp, to a pro class D amp with almost 3x the power.

It worked - the XLS2500 sounds substantially better than the Quad 606 or the AVR.

Does it make any sense? not really.... I tend to listen to movies at reference -15db ... so not overly loud, I run music at the same level

The Speaker sensitivity is 88db/w@1m.... it shouldn't require that much power... and I certainly am NOT using that much power!

The Quad is a great amp, and sounded absolutely fabulous driving Quad ESL988 electrostatics - but on the Gallo's the Crown sounded better.

A number of people have mentioned that their Valve amps of sub 30w work well with the Gallo's... - perhaps they have a better ability to provide high current into the demanding tweeter load?

Either way - the number of watt's an amp provides, does not necessarily provide a relevant indication of how well it will drive a specific speaker.

Some good indicators of an amps ability with "difficult" speaker loads might be, whether the amp can double down into 4 ohm and then again into 2 ohm..... so a 100W amp @8ohm should be 200W @4ohm and 400W@2ohm... and one would hope 800W@1ohm.

Most amps won't do that.... most amps simply won't drive 1ohm loads (!) - and perhaps 1ohm is asking too much - but given my Gallo CDT tweeter does indeed drop down below 2 ohm....

So yeah - in theory a 100W amp should be more than ample for my needs... in fact most likely a 50W amp would do the job! It just has to be the right kind of 100W.... alternatively a 440W pro amp, capable of driving 1200W into 2ohm loads seems to do the job just fine.

 

[Koeitje]

Thank you NTK, that document is indeed useful.
View attachment 120827
View attachment 120828

It's amazing to me just how LITTLE power is required for every day listening.

Using https://www.crownaudio.com/en/tools/calculators I'm showing that my setup requires only 10w to stay under the 85db threshold for long term hearing damage.

View attachment 120830
On the other hand, it's also crazy how quickly power levels increase. 3kw at 110db aka "rock concernt / full orchestra (must be drums)

View attachment 120831

That all be said, this doesn't really answer my question. 10w at 85db... but 85db of bass is a different power level than 85db of treble right? So if I wanted to be accurate, I would need to know how much power I needed drive my woofers to the db I wanted.
View attachment 120832

Anyways, at this pointed I'm leaning toward the extra power of the Purifi.

Remember that the desired level at the listening position should be the peak output you want. If you want to listen at 85dB average and the music has 10dB peaks then you need 95dB at your listening position. If you use 20dB for peaks you are pretty safe, not much music has that kind of dynamics.

For me amplifier power seems a paradox at first sight, because you probably need both less and more power than you initially think. Average SPL is easily achieved, but for the peaks you suddenly need a lot of power.

 

[Willem]

I used to drive my Quad 2805s in a large room with a 2x45 watt Quad 303. This was clearly not enough so I bought a refurbished 2x140 watt Quad 606-2, and that was far better. It was only in extremis that I could sense that it was still stressed a bit. I now have a subwoofer and a high pass filter for the signal into the power amplifier, and that has I think solved the issue. To be honest, subwoofers and a high pass filter are probably the easiest and cheapest way to have a lot of headroom.

 

[peng]

In theory, for a lot of speakers, the amp's "watts" output into 8/4 ohm can be a reliable specs for predicting if one can get good results. For some speakers such as the Gallo Reference, I suspect it is necessary to see the full specs including the impedance/phase angle vs frequency graphs and also the amp's output and current specs as well, not just watts. Obviously if the amp is rated 500 WPC 8 ohms then you mostly can ignore its individual voltage and current limits because even in the worse scenario such an amp should be able to do 250 W 4 ohms or 125 W 2 ohms. The Gallo Ref speakers would have to have some really weird electrical characteristic to present a challenge to a 500 W 8 ohm amp.

 

[Schollaudio]

Here is my McIntosh MC462 driving KEF R11s. This is loud rock. Does look like we need lots of power for content like this. ...and these are relatively efficient speakers.

The power meters are showing upwards of 10DB peaks. However, they are much to slow to show the true voltage peaks. Using an oscilloscope is much better to see voltage peaks. MAC amps are hard to determine the point of clipping because they use step up autoformers but that amp is probably getting close to clipping.

Comparing an O-scope to power meters and clip indicators I find 10DB headroom is needed. On one of my amps the meters will show 20 watts but the clip indicator is flickering and the scope is peaking at 60V p-p or peaks or 200watts.

The OPers question is hard to answer but using 10 times more power than the users likes to listen at isn't ridiculous.

 

[peng]

The power meters are showing upwards of 10DB peaks. However, they are much to slow to show the true voltage peaks. Using an oscilloscope is much better to see voltage peaks. MAC amps are hard to determine the point of clipping because they use step up autoformers but that amp is probably getting close to clipping.

Comparing an O-scope to power meters and clip indicators I find 10DB headroom is needed. On one of my amps the meters will show 20 watts but the clip indicator is flickering and the scope is peaking at 60V p-p or peaks or 200watts.

The OPers question is hard to answer but using 10 times more power than the users likes to listen at isn't ridiculous.

I checked into that before, according to the manual:

• Illuminated Power Meters
The Illuminated Power Output Watt Meters on the
MC462 are peak responding, and indicate the true
power output of the amplifier. The Peak Watt Hold
Mode allows the meters to temporarily stay at the
highest power output and then slowly decay.

I think those meters (assume they are truly analog type) do show the true voltage peaks.

 

[gags11]

The power meters are showing upwards of 10DB peaks. However, they are much to slow to show the true voltage peaks. Using an oscilloscope is much better to see voltage peaks. MAC amps are hard to determine the point of clipping because they use step up autoformers but that amp is probably getting close to clipping.

Comparing an O-scope to power meters and clip indicators I find 10DB headroom is needed. On one of my amps the meters will show 20 watts but the clip indicator is flickering and the scope is peaking at 60V p-p or peaks or 200watts.

The OPers question is hard to answer but using 10 times more power than the users likes to listen at isn't ridiculous.

I have run a frequency sweep with this amp powering Dirac equalized KEF R11s, going to about average 104db at 1 meter. The power meter on the Mac went all the way to max passing through around 100hz region.

Manual states the meters are calibrated for output somewhere at 2khz region. Not sure what to make of this.

Edit: This observation is why I think in some cases even pros may get compression at a certain frequency, thinking it was caused by the woofers or speakers per se. but in actuality, their amp may be clipping or running out of power given the load, eq, etc.

Unless you got what Amir has, a monster Classe, Mc, or similar amp that can go close to 1k power, I won’t believe any woofer compression data (please convince me otherwise if you got the data).

 

[gags11]

I checked into that before, according to the manual:



I think those meters (assume they are truly analog type) do show the true voltage peaks.

Compare those to the fake D’Agostino meters!!
Sorry, could not resist….

 

[Schollaudio]

I checked into that before, according to the manual:



I think those meters (assume they are truly analog type) do show the true voltage peaks.

OK, these meters are clearly more realistic then typical electro mechanical types. Good job MAC. Sometimes ya get what you pay for.