Two Basic Amplifier Questions

[radix]

Thanks for the reply, appreciate it. While I can't argue with your logic -- seems very good -- I can't help but wonder why Amir includes dynamic range specs in his testing. If you're correct, it has pretty much nothing to do with real world listening. Any thoughts?

modern amps are exceedingly clean. You'll see 100, 105, 110, 120 dB of signal-to-noise, which amir converts to equivalent bits. yep, anything past 16 bits is not very audible. But, it is indicative of excellent engineering. And people with 105dB sensitivity speakers might care.

 

[Deleted member 48726]

People have underestimated how much power they think they need forever.

I would estimate there are perhaps one or two members on ASR who might just have enough power on tap to cover the full audible dynamic range of 16 bit audio in their systems.

True. And every discussion on this topic results in people with their 5 w amplifiers posting SPL measurements lacking any resemblance to full range sound and no recognition of peoples different taste of music.

If all amps had clipping indicators this topic wouldn't be controversial.

 

[restorer-john]

I can't help but wonder why Amir includes dynamic range specs in his testing. If you're correct, it has pretty much nothing to do with real world listening. Any thoughts?

He's perfectly right doing dynamic range (electrical) tests.

How that translates to audible dynamic range, when it comes to loudspeakers in a room is complete other story.

I just did the experiment. 2mV was the minimum level I could just perceive a 1kHz tone with my ear right near the speaker (<1ft). That was with complete silence in my room in a very quiet location. So that translates to ~0.5uW.

60dB above that would be 2V
70dB above that would be 6.3V
80dB above that would be 20V- close to the Topping amp's limit. It's done at this point as it has no dynamic reserves.
90dB above that would be 63V and at 500wpc. I'm still going with plenty in reserve and more importantly a decent dynamic capability.
96dB above that would be 1990 Watts required. Nearly 2kW! I can't do that. Nor would I want to, but you get the picture.

 

[Deleted member 48726]

He's perfectly right doing dynamic range (electrical) tests.

How that translates to audible dynamic range, when it comes to loudspeakers in a room is complete other story.

I just did the experiment. 2mV was the minimum level I could just perceive a 1kHz tone with my ear right near the speaker (<1ft). That was with complete silence in my room in a very quiet location. So that translates to ~0.5uW.

60dB above that would be 2V
70dB above that would be 6.3V
80dB above that would be 20V- close to the Topping amp's limit. It's done at this point as it has no dynamic reserves.
90dB above that would be 63V and at 500wpc. I'm still going with plenty in reserve and more importantly a decent dynamic capability.
96dB above that would be 1990 Watts required. Nearly 2kW! I can't do that. Nor would I want to, but you get the picture.

You've made this simple calculation a couple of times before (this may even be copy-pasted?) And it should really be in an FAQ for reference and reality checks..

 

[restorer-john]

You've made this simple calculation a couple of times before (this may even be copy-pasted?) And it should really be in an FAQ for reference and reality checks..

I thought I had done something like it, but figured it was easier to just type it up again.

People will keep buying little weiner amplifiers as long as they are praised to the hilt by reviewers, regardless of whether they are really fit for the purpose of driving typical inefficient modern loudspeakers, in a room. Letalone even coming within 20dB of the dynamic range of 16 bit on a bad day.

 

[pma]

He's perfectly right doing dynamic range (electrical) tests.

How that translates to audible dynamic range, when it comes to loudspeakers in a room is complete other story.

I just did the experiment. 2mV was the minimum level I could just perceive a 1kHz tone with my ear right near the speaker (<1ft). That was with complete silence in my room in a very quiet location. So that translates to ~0.5uW.

60dB above that would be 2V
70dB above that would be 6.3V
80dB above that would be 20V- close to the Topping amp's limit. It's done at this point as it has no dynamic reserves.
90dB above that would be 63V and at 500wpc. I'm still going with plenty in reserve and more importantly a decent dynamic capability.
96dB above that would be 1990 Watts required. Nearly 2kW! I can't do that. Nor would I want to, but you get the picture.

Hi John, I am confused by the result you get (2mV threshold). IMHO, it is very, very high. Sometimes I do the same test, so I have repeated it this morning. The procedure:

- I set Arta generator to 1kHz/0dB,
- I set amplifier output level, now it was 300mV for 1kHz/0dB,
- I decrease Arta generator dB level as long as something is audible.

Now I have ended at -80dB, which makes 300mV/10000 = 0.03mV. Almost 100x less than you reported. Please take into account you must not use DVM to measure such low voltage level, as it shows all the amp noise background as well.

In my test, 0.03mV is barely audible, 0.1mV is well audible. Your chart would look completely different, then.

 

[restorer-john]

Hi John, I am confused by the result you get (2mV threshold). IMHO, it is very, very high. Sometimes I do the same test, so I have repeated it this morning. The procedure:

- I set Arta generator to 1kHz/0dB,
- I set amplifier output level, now it was 300mV for 1kHz/0dB,
- I decrease Arta generator dB level as long as something is audible.

Now I have ended at -80dB, which makes 300mV/10000 = 0.03mV. Almost 100x less than you reported. Please take into account you must not use DVM to measure such low voltage level, as it shows all the amp noise background as well.

In my test, 0.03mV is barely audible, 0.1mV is well audible. Your chart would look completely different, then.

I'm using the millivoltmeter- 200kHz bandwidth in my distortion analyzer. Class AB amp, so no HF garbage to throw it off. It'll read down to 30uV or so. There's around 100uV of residual in the amplifier (65uV A-Wtd) Obviously at 2mV there is also a contribution of noise from the preamp, amp and source. But that doesn't really matter, I couldn't hear any noise at ~1ft and the absence of the tone (switched to shorted input) dropped the reading to the baseline 100uV.

The baseline is the tone- can you hear it or not and at what level can you hear it on your loudspeakers in your room.

I'll revisit the test again later when it cools down (31 degrees C here) and see if I get different numbers. I may have made a mistake- who knows.

 

[JiiPee]

The lowest sound level detected varies among humans. Person A can detect more than 12 dB quieter sounds than person B, while both of them are still considered to have normal hearing ability.

For many of us 20dB background noise is already considered very quiet, and 80 dB above that will be very loud. Also, in most cases using two speakers instead of one gives You approx. 3dB extra and so does a typical room gain.

 

[pma]

I'm using the millivoltmeter- 200kHz bandwidth in my distortion analyzer. Class AB amp, so no HF garbage to throw it off. It'll read down to 30uV or so. There's around 100uV of residual in the amplifier (65uV A-Wtd) Obviously at 2mV there is also a contribution of noise from the preamp, amp and source. But that doesn't really matter, I couldn't hear any noise at ~1ft and the absence of the tone (switched to shorted input) dropped the reading to the baseline 100uV.

The baseline is the tone- can you hear it or not and at what level can you hear it on your loudspeakers in your room.

I'll revisit the test again later when it cools down (31 degrees C here) and see if I get different numbers. I may have made a mistake- who knows.

Just to add, the speaker midrange (active box) used has 86dB/2.83V/m quite usual sensitivity. Arta generator being double-checked for linearity, I think no one would doubt about it. Output level re-measured by calibrated divider/soundcard. Basic test level at 0dB/300mV re-checked by DVM.
In the past, I several times experienced DVM errors when measuring about 1mVac at amp output.

We spoke about audibility "near speaker". This is vague, so I add the level of audibility at 1m distance from speaker. In my case, the 1kHz tone is audible at -72dBV level / 1m, i.e. 0.25mV.

 

[pma]

I think this kind of chart might be of better practical use:

 

[restorer-john]

This is vague, so I add the level of audibility at 1m distance from speaker. In my case, the 1kHz tone is audible at -72dBV level / 1m, i.e. 0.25mV.

Ok. 251uV at 1kHz into your loudspeaker at 1M. Let me see what I can come up with when everyone has gone to bed, the fans are off and the Koalas are keeping quiet.

It's so dry here, one came out of the bush tonight to find water. There was water running down the gutter and he climbed in to take a drink. Poor little guy...

 

[IAtaman]

I think this kind of chart might be of better practical use:

I was just trying to do that math for 8 ohm speaker. Thanks for that.

Looking up from the table, I see I am gonna need 20mV to reach to ca 43bB, just above the noise floor in a quite room, and to have the full 90db headroom of 16 bits, I'd need an amp that can deliver 4kW power, which is more or less what restorer-john said no?

 

[pma]

Looking up from the table, I see I am gonna need 20mV to reach to ca 43bB, just above the noise floor in a quite room,

This thought about noise floor in a quiet room might be confusing, because listening to a single tone you can hear deep below the integrated room noise level. A microphone cannot, human ear can, with its selective brains capability. I measured 38dBA ambient noise in my room with a SPL level meter.

 

[Sokel]

I won't do the math,but I set the pre at full (that's not unity gain,it amplifies) ,put a 1Khz sine on Multitone ,started from -100db play gain and going up I started to hear it at -93db at one meter,measured the speaker terminals and they say 1.4mV.
Noise floor here is 37dbA.

 

[Sokel]

Taking @pma 's post about background noise into account I also measured without signal at the exact same state (everything at 100% ) at speaker terminals and they say 0.8mVAC.
So,signal needed is probably lower but it's hard to measure accurately,way above my pay grade.

 

[peng]

People have underestimated how much little power they think they need forever.

I would estimate there are perhaps one or two members on ASR who might just have enough power on tap to cover the full audible dynamic range of 16 bit audio in their systems.

FIFY Seriously though, both could be true, depending on usage conditions/applications.

 

[Sokel]

FIFY Seriously though, both could be true, depending on usage conditions/applications.

Taking serious applications by people who probably know what they're doing into account,we can look at stuff like 8381A with it's 6kW per channel.
I don't think they carelessly calculated the needs and we must take into account the number of drivers and the fact that the whole thing is active so less power losses than a passive one.

That's about compromises,nothing more.

But smaller stuff are into kW area as well.

 

[peng]

Thanks for the reply, appreciate it. While I can't argue with your logic -- seems very good -- I can't help but wonder why Amir includes dynamic range specs in his testing. If you're correct, it has pretty much nothing to do with real world listening. Any thoughts?

Great point, as you alluded to, for real world listening, i.e., in practical terms, most of us will be limited by, but not limited to, the following facts:

- The maximum peak voltage/current your speaker can take when listening to the most dynamic music you listen to.
- Your speaker's sensitivity, in dB/2.83 V/m
- Your seating distance
- Impedance/phase angle characteristics
- The contents we listen to, for example: heavily amplified live rock concerts recordings, vs huge symphony orchestras (classical such Beethoven's Symphony No.9)

Just because many members on ASR might have the likes of the McIntosh 2 kW per channel amp, or bridged Crown audio XLS2502, doesn't mean your 125 WPC AVR (just as example) are clipping in your own applications. For example, I know in one of my 2 channel system, my 500 W 4 ohm rated amp would never be pushed to more than 50 W under the most demanding condition. My own rule of thumb is, if the calculated (easy to do) maximum power I need is say 20 W 4 ohms, I would go with an amp rated 200 W, preferable much higher, based on my speaker's recommended power handling spec of, say, 200 W. That's just me, to me, you can't have too much power whether you need it or not.

 

[fpitas]

you can't have too much power whether you need it or not.

I agree. I always considerably oversize my amps.

 

[JeremyFife]

He's perfectly right doing dynamic range (electrical) tests.

How that translates to audible dynamic range, when it comes to loudspeakers in a room is complete other story.

I just did the experiment. 2mV was the minimum level I could just perceive a 1kHz tone with my ear right near the speaker (<1ft). That was with complete silence in my room in a very quiet location. So that translates to ~0.5uW.

60dB above that would be 2V
70dB above that would be 6.3V
80dB above that would be 20V- close to the Topping amp's limit. It's done at this point as it has no dynamic reserves.
90dB above that would be 63V and at 500wpc. I'm still going with plenty in reserve and more importantly a decent dynamic capability.
96dB above that would be 1990 Watts required. Nearly 2kW! I can't do that. Nor would I want to, but you get the picture.

There's some interesting threads on ASR at the moment about 'real world' issues; amp behaviour under difficult loads, power over a full frequency range etc. This seems to fit onto that.

So, for a real world question: what kind of Speakers are needed to handle this sort of power? A mid-range example like Kef R3 has a recommended max power of 180W. The Kef Information Sheet does not qualify this so I don't know if that's peak (surely not) or only for a frequency and impedance range. Whatever, it's lower than 500W and nowhere near 2kW

Does this comment on 'real power' needed for a full dynamic range in music mean that most speakers can't handle the power required anyway?

Do speaker specifications describe what is needed, or is this another area where specs are not enough?