Assuming the above was meant to spell "irrelevant", i dare disagree. Higher voltage appears to be part of the secret to a sound that lacks strain. Unlike the OP i have never played with amps capable of more than 400W/8ohms and perhaps more importantly 800W/4ohm but there is no mistaking their sound for even a powerful sounding 50W class A. At any power level. Even after midnight 🙂
If higher voltage eliminates strain, then you needed the higher voltage in the first place. IOW, the lower voltage amp would have been clipping.
If massively paralleled output transistors eliminates strain, then you needed the current capability. IOW, a lesser amp would have been current limiting in some way. Maybe not a hard limit like a protection circuit kicking in - perhaps just running out of beta, loading the VAS, and dropping the loop gain enough to cause artifacts you CAN hear.
If massively paralleled output transistors eliminates strain, then you needed the current capability. IOW, a lesser amp would have been current limiting in some way. Maybe not a hard limit like a protection circuit kicking in - perhaps just running out of beta, loading the VAS, and dropping the loop gain enough to cause artifacts you CAN hear.
A bit of conjecture in some posts by folks who I gather have never designed or built amplifiers. Mother nature does not always follow arm-chair logic.
You only need the current requested by the input and demanded by the load.
You will only need the voltage to achieve that current in a linear fashion.
The higher the current through a BJT, the more non-linear. Well understood. Multiple outputs also have well understood advantages in thermal stability though one must deal with current hogging. ( I like my FET outputs)
Rail collapse, due to insufficient energy storage can easily cause both VAS and output limitations. ( as I think wg-ski is suggesting)
So many here are focusing only on power, but the entire execution matters. Even swapping IPS local feedback vs global feedback can swap even for odd order distortion as predominate. Dominant pole vs Miller compensation effects the distortion distribution. The latter having a large effect on the "openness and dynamics" of an amplifier. What the open loop gain is, and what the poles and zeros are to maintain stability matters. BW of the VAS matters. Many believe the inherent linearity of a FET input is critical. The list goes on and on. A really big issue is the degradation caused by monitoring and safety circuits. I recently put a small comercial amp design into Spice and found the current monitoring and shutdown to have a 100 x effect on THD. Simulated with all monitoring in place, matched the manufactures specs.
BTY, I have played with 2000W amplifiers. They were PA class D and I would never live with one in my living room. Horrific distortion. The OP did not mention what amplifiers were in play. As 2000W amplifiers with the same level of SNAID as a 20W would cost tens of thousands of dollars, I can only assume it was budget PA amps. Now some like the sound of distortion as they think it is "fuller and more dramatic" To each their own. One size fits, well, only those with that size.
You only need the current requested by the input and demanded by the load.
You will only need the voltage to achieve that current in a linear fashion.
The higher the current through a BJT, the more non-linear. Well understood. Multiple outputs also have well understood advantages in thermal stability though one must deal with current hogging. ( I like my FET outputs)
Rail collapse, due to insufficient energy storage can easily cause both VAS and output limitations. ( as I think wg-ski is suggesting)
So many here are focusing only on power, but the entire execution matters. Even swapping IPS local feedback vs global feedback can swap even for odd order distortion as predominate. Dominant pole vs Miller compensation effects the distortion distribution. The latter having a large effect on the "openness and dynamics" of an amplifier. What the open loop gain is, and what the poles and zeros are to maintain stability matters. BW of the VAS matters. Many believe the inherent linearity of a FET input is critical. The list goes on and on. A really big issue is the degradation caused by monitoring and safety circuits. I recently put a small comercial amp design into Spice and found the current monitoring and shutdown to have a 100 x effect on THD. Simulated with all monitoring in place, matched the manufactures specs.
BTY, I have played with 2000W amplifiers. They were PA class D and I would never live with one in my living room. Horrific distortion. The OP did not mention what amplifiers were in play. As 2000W amplifiers with the same level of SNAID as a 20W would cost tens of thousands of dollars, I can only assume it was budget PA amps. Now some like the sound of distortion as they think it is "fuller and more dramatic" To each their own. One size fits, well, only those with that size.
I love this discussion. Thanks to everyone for participating. I am not an electrical engineer (mechanical engineer), but have no doubt that there are many factors that play a role in dynamics, power supply, capacitors, heat, etc. Hence the need to measure the dynamic output of the loudspeaker. Except I want to change the word loudspeaker to system. In my mind the system is everything including the room, loudspeaker, preamp, amp, source, wires, etc.
I have a basic question. If you are playing a tone in REW and you set your volume control as needed to get 70 db output from your system. If you raise the level in REW by 20 db, how many db should the sound increase? Don't move the microphone, volume setting, just the value of the input. System A may take .25 watts and system B may take 2 watts. But should the output increase the same amount for system A and B? Does anybody know?
Dan
I have a basic question. If you are playing a tone in REW and you set your volume control as needed to get 70 db output from your system. If you raise the level in REW by 20 db, how many db should the sound increase? Don't move the microphone, volume setting, just the value of the input. System A may take .25 watts and system B may take 2 watts. But should the output increase the same amount for system A and B? Does anybody know?
Dan
Some friends had a small house that had a huge pa stack that belonged a local metal band, would practice there once or twice a week. We just used the system to watch Pink Panther cartoons. It was an abandoned put-put golf course, so no neighbors to get angry.
If the loudspeakers were perfect, yes. But they are not and they are not linear. As you know, you are dealing with a damped mass. I stumbled in second semester dynamics, so you will know what I mean. There is much discussion about steady state measurements, but that goes to hell in a resonant system. With some effort, we can explain and measure that. What we do not have a good enough grasp on is what clues triggers your brain to be music. Basically, we do not know the target transfer function to build to. Perfect is likely not it. Or maybe, it is if we were trained to recognize it, but we have only been exposed to highly flawed reproduction and that is what our poor brains expect. It is complicated.
I am sure quite biased, but I have studied, designed, and built both amplifiers and speaker systems off and on for over 40 years, so I have some basis for my bias. I have modified one of two identical amps to both model in Spice and listen to and made some general conclusions. My ace in the hole is my wife's very sensitive hearing. She could care less how hard I worked, what I think, or what it costs. She only cares about the music and has provided very reliable feedback over the years. Frequently disappointing to my efforts, sometimes confirming them.
There are things I can measure ( or others can) I can't hear
There are things I can hear I can simulate so I assume with enough money can be measured
There are things I think I maybe can hear
There are things I hear I think I may know why
And there are things I can hear I haven't the clue why.
I only concern myself with the last two as that is where learning is.
Decades ago, before the trend of using paralleled output transistors, a paper was written describing about that very thing, and it's benefits as well as its (potential) issues.
What I can remember from it is that utilizing multiple/paralleled devices manages to create some new distortion products due to each transistor not being exactly the same as its partners.
Basically, a single NPN/PNP or NPN/NPN pair will exhibit less distortion, because there's a "single" turn-on/off node involved, and a simpler job for the feedback system to deal with.
Now, when you've got say, 2,3, or more pairs of outputs in parallel, there's a small, minute, bit of difference in each pair added in- manufacturing differences, tolerences, and that is what I remember from that writing.
Even while being "matched", close, this remains an issue, however those types of amps are better used for PA service where such issues are of no critical concern.
Don't forget - your hearing ability changes as you age, but of course you likely already know that.
I've had people bring me their stuff in the shop and claim it doesn't sound "like it used to".
When all that was an issue was their own hearing.
Sometimes I'd mod/boost the tweeters to compensate for "their loss"
There are major differences between BJTs and MOSFETs with respect to linearity and current. Do not lump them together.
With regards to paralleling outputs, maybe "distributing" distortion products was an idea back in the day, but if you know how to read a transistor spec sheet ( curves) you will have a better understanding it is more likely about current and temperature.
I have been trying to think of an objective bench test that might put a value on the ability to maintain the rails without opening the amp and measuring directly. Some sort of time dependent tone burst and look at the last wave compared to first, both in amplitude and distortion created at that instant. The later may not be significant. Then some sort of complementary "gain compression" test, though global feedback should maintain gain with level and temperature. It should. Maybe it does, maybe not. Never seen it measured. I can think of a few other implementation parameters with regards to the bias tracking and how far it lags on-die temperatures. Then correlate these measures with the subjective preferences of human hearing. Having a measured value does not automatically give us a "preferred transfer function" for want of a better term. Our brain remains the wild card.
FWIW, back in the olden days, purely subjective, my favorite two amps I could not afford were the Aragon and the Nak PA5 Stasis. A close follower was the Adcoms. I never listened carefully to Threshold or Martin Levinson as their price was not just unaffordable, but almost abstract. Well, I do remember being blown away by an early 100W Krell. Things they had in common were multiple outputs and robust supplies. This is a hindsight analysis as in those days, I had no education on amplifier design and had never seen the schematics. Amps of that time, sadly some today, had a clear sonic signature. I had a Sanyo Plus 55 MOSFET as that was the best I could afford. If the heat pipe did not gurgle so loud, I might still have it. I don't think I matched it's sound, or lack there of, until I built my own after over a dozen in between. It would have been cheaper to buy the Nak but it cost more than my car at the time.
You may be mixing few issues here.
First thing first, amplifiers should be linear, that is, voltage in, voltage out multiplied by gain.
Assuming voltage, not current amplifiers, where gain varies with speaker resistance.
So wheather you use 60 watt amp or 5000 watt amp, if you do not clip the 60 watt amp, they will behave equally. Voltage in, voltage out. There should be no significant difference.
Now, when you talk about system A and system B as two different speakers, you will get different dynamics based on the construction of the speakers.
Speaker, that is normal dynamic speaker, is a mechanical device, where the travel of speaker cone is limited by the spider and surround, so in a sense, non-linear. In certain small region of cone movement its linear, but beyond certain signal the cone does not follow input signal, till it completely hits max excursion. So the signal in does not equal excursion or signal out.
This is different for each speaker system, so expect different dynamics for each, depending on spl. In very low level, most speakers will be linear. Exceed certain level and some system will start compressing the signal. Some will suffer this sooner then other systems. So choose your speakers based on spl you need.
THE problem is that it is possible for real-world use conditions to cause an amplifier NOT the behave as a voltage source. It’s just easier to do than simple models would suggest, and that real voltage requirements can be shockingly high.
Adason
Thank you for your response. I agree with everything you said. Is there a way to measure system response to show that one system is more dynamic than another system?
Thank you for your response. I agree with everything you said. Is there a way to measure system response to show that one system is more dynamic than another system?
Slew rate, current, and not clipping make the sound dynamic. These are virtues found in many larger amps but not exclusively.
I assume 'dynamic' means a lack of intermodulation distortion. So, high voltage rails could help reduce beta modulation (IIRC, or transconductance for FETs. In many cases, I've already played around with the effects before reading about the name.)
For personal use, I only need about +/-5V output headroom to cover all daily music and av needs, and that's with low sensitivity speakers. When the kids are older, they might want 4x that to put loud movies on blast (except that I'm going to higher sensitivity speakers anyway).
But from that 10V pk-pk swing, the rails only really need 30-ish V for MOSFETs, which will distort a lot more than with 40-50V. But then it's a game of bulking up the whole amplifier "because you promised more power!".
I'd rather devote the upper 50% of |output voltage| (or let's say 0.1% actual output and 99.9% hot-plugging phone jacks) to soft clipping.
For personal use, I only need about +/-5V output headroom to cover all daily music and av needs, and that's with low sensitivity speakers. When the kids are older, they might want 4x that to put loud movies on blast (except that I'm going to higher sensitivity speakers anyway).
But from that 10V pk-pk swing, the rails only really need 30-ish V for MOSFETs, which will distort a lot more than with 40-50V. But then it's a game of bulking up the whole amplifier "because you promised more power!".
I'd rather devote the upper 50% of |output voltage| (or let's say 0.1% actual output and 99.9% hot-plugging phone jacks) to soft clipping.
Last edited:
Pardon me for asking, why would lack of intermodulation distortion be dynamics?
Any literature refference?
Any literature refference?
Last edited:
My understanding of IM thd is when two signals combine to create new signals. Above and below the frequencies of orinal two.
Not sure how lack of this is dynamics.
To me dynamics is simply a difference between low and high signal level. Without any compression. Small speakers with small membrane surface area simply can not reploduce both low signal and full fortissimo of the orchestra, because of the limits in peaks.
That would be my understanding of the dynamics. Not sure how IM thd comes in.
But i may be wrong.
With complete lack of any data from thread starter, this thread is meaningless.
Not sure how lack of this is dynamics.
To me dynamics is simply a difference between low and high signal level. Without any compression. Small speakers with small membrane surface area simply can not reploduce both low signal and full fortissimo of the orchestra, because of the limits in peaks.
That would be my understanding of the dynamics. Not sure how IM thd comes in.
But i may be wrong.
With complete lack of any data from thread starter, this thread is meaningless.
Well that's just it. The differences in signal level are heard on a logarithmic scale, so a drooping supply rail makes no literal sense to me. But it does make sense in terms of distortion, especially if a complex music signal seems to selectively raise the noise floor during the peaks.
I also have my pet theories about speakers interacting with amplifiers, and depending on what the designer intended, there can be great variability in the end result.
I also have my pet theories about speakers interacting with amplifiers, and depending on what the designer intended, there can be great variability in the end result.
Slew rate does not effect "dynamics". It does effect the distortion.
Clipping is of course the wrong amplifier for the load and level required.
Insufficient current reserve will cause the rails to droop, which causes "dynamic compression" as you approach clipping.
Correct negative feedback should account for Beta droop in a BJT output, but my ears tell me MOSFETS linearity are still my preference.
A lot of "dynamic" sound is nothing but distortion. Like it, don't like it is everyone's choice. You may find a very clean amplifier sounds less "dynamic" than what you are accustomed to. I have heard this criticism about the newest very clean class D as well as the Benchmark. Basically, lack of distortion.
Take a good look at the speakers before you blame the amp.