I mentioned 2Hz to 200KHz as the -3dB points for my general designs, and as a recommendation for a good 20-20KHz bandwidth. That bandwidth is not for some super-human hearing, or for any woo-woo subjectivist reasoning, just a nice, wide pass-band for preamps and power amps.
I have several CDs with low-end information (information that is physical more than musical, but fun) down to 5Hz. Specifically, my 5TH Element soundtrack CD has tremendous and clean bass recordings and is overall, a great sounding disc that will definitely test the low-end capability of one’s system.
So limiting -3dB to 35Hz is just ridiculous as much would be missed. This isn’t the ‘70s with huge sloppy woofers in ported cabinets that need to be protected from themselves or from wow and crap spewing from the 8-track.
Then again, if you’re designing amps and preamps for vinyl (Isn’t vinyl supposed to be pure analog and perfect?), then you are already coddling to horrendous equalization and bandwidth limitations, so there is much more to consider and one may want (have) to EQ and/or limit the bandwidth.
For high-end response, again, I use 200KHz as it gives me a nice, flat -0.1dB at 20KHz and is an easily obtainable figure with properly designed electronics. Plus, considering the wall filters in CD and SACD players, there is no HF information above 50KHz to worry about.
I choose not to employ tone controls or extraneous filtering in my signal path and leave bandwidth limitations to the source unit. But, that’s just what I like.
I have several CDs with low-end information (information that is physical more than musical, but fun) down to 5Hz. Specifically, my 5TH Element soundtrack CD has tremendous and clean bass recordings and is overall, a great sounding disc that will definitely test the low-end capability of one’s system.
So limiting -3dB to 35Hz is just ridiculous as much would be missed. This isn’t the ‘70s with huge sloppy woofers in ported cabinets that need to be protected from themselves or from wow and crap spewing from the 8-track.
Then again, if you’re designing amps and preamps for vinyl (Isn’t vinyl supposed to be pure analog and perfect?), then you are already coddling to horrendous equalization and bandwidth limitations, so there is much more to consider and one may want (have) to EQ and/or limit the bandwidth.
For high-end response, again, I use 200KHz as it gives me a nice, flat -0.1dB at 20KHz and is an easily obtainable figure with properly designed electronics. Plus, considering the wall filters in CD and SACD players, there is no HF information above 50KHz to worry about.
I choose not to employ tone controls or extraneous filtering in my signal path and leave bandwidth limitations to the source unit. But, that’s just what I like.
There's no "cut" at 2 Hz, that's the -3dB frequency.darkfenriz said:Actually in most hi-fi applictions cutting bass around 35Hz can be superior, because most woofers are happier when not tortured with freqs they cannot cope with (coil heating etc..). Absolutely no need to cut at 2Hz. And please do not say anything about phase of bass...
And the power amplifier is certainly not the place to add a high-pass filter (if you want to add one at all), the correct link in the audio chain for that is the preamp/equalizer.
Re: high pass filter
I don't think you want a high-pass filter in the feedback loop? What would be the reason to have one?
The input cap, AFAIK, is there basically to block DC. Make it as large as practical. True, it's a high-pass filter together with whatever resistance is in series with it. So you want its corner frequency to be about 2Hz or lower, as indicated by DCPreamp.mikelm said:well, I agree, but many designs include an input capacitor and a feedback capacitor which give plenty of opportunuity for a high pass filter and, unfortunately, it also gives plenty of scope for messing up the audio signal ....
I don't think you want a high-pass filter in the feedback loop? What would be the reason to have one?
Hi,
re post33,
I agree with the Baxandall idea that too little feedback is worse than sufficient. I recall the graph he showed (did it come from someone else?) with the rising distortion as the frequency/harmonics went up (it was quite a while ago).
The rise in distortion that we see in conventional power amps starting to come out of the noise floor at around 1kHz and becoming quite significant at 20kHz is due to the reduced feedback at the higher frequencies.
If some intermodulation mechanism were to bring super high frequency artefacts down into the audio band then increased bandwidth and increased feedback would only help make for a better audio amp.
But I fear the target has been set too high.
The clever designers do what they can with the technology that exists and minor tweaking moves the target forward only very slowly it seems. Me, I just copy the bits I think show merit.
re post33,
I agree with the Baxandall idea that too little feedback is worse than sufficient. I recall the graph he showed (did it come from someone else?) with the rising distortion as the frequency/harmonics went up (it was quite a while ago).
The rise in distortion that we see in conventional power amps starting to come out of the noise floor at around 1kHz and becoming quite significant at 20kHz is due to the reduced feedback at the higher frequencies.
If some intermodulation mechanism were to bring super high frequency artefacts down into the audio band then increased bandwidth and increased feedback would only help make for a better audio amp.
But I fear the target has been set too high.
The clever designers do what they can with the technology that exists and minor tweaking moves the target forward only very slowly it seems. Me, I just copy the bits I think show merit.
AndrewT said:
With today's semiconductors
there are is nothing 'extreme' about upper -3dB at 100.000 Hz.
Even for Very high power amplifiers.
As wwood and others have mentioned:
to minimize phase distortion within audio band
it is good trying not to have too low limit.
This is of course if your circuit allows this, without other bad effects.
There is no reason I can see to 'aim for 2 - 50 kHz'.
When we while careful use of capacitances can get 2x, 4x that bandwidth.
Of course, if you think it is difficult to
do the work of final trimming adjustments,
..... the easy solution is to restrict bandwidth to 50 kHz.
I know most wouldn't do this.
I wouldn't.
Because I want as good amplifier as possible.
And some of my criteria for a good amplifier is:
- minimum phase distortion
- very good transient response
- reasonable low noise
- low THD harmonic distortion
Because designers have different criteria
for what is a good amplifier.
The bandwidth and other quality in amplifier can be different.
Therefore, it is in the end up to your own preferences:
------------------------------------------------------------------------------
- lower upper -3dB for minimal THD distortion
- higher -3dB for better transient response
lineup
Lineup Ultra Amplifiers
http://lineup.awardspace.com/
Hi all
Many amplifiers running with 6 MHz transistors can provide a bandwidth of 200 kHz. This seems to me to be the minimum requirement as it still allows possibilities for low distortion.
200 kHz is ten times the maximum audio which means in practice that 20 kHz signals (don't think I can hear 20 kHz) will be pretty accurately reproduced.
But the newer 30 MHz transistors may allow this to be increased. As the power ratings go up the speed of the output devices needs to rise too to maintain slew rate.
Many years ago a letter to Wireless World reported that "after 150 kHz all amplifiers tend to sound the same". This was only a letter with no data to back this observation up. This was based on subjective effects I presume.
Trouble is that input and output filters all act together and reduce the overall bandwidth. As most people like to have inductors in the output lead this reduces the overall bandwidth - so the amplifier has to stay high to maintain the performance.
Cheers
John
Many amplifiers running with 6 MHz transistors can provide a bandwidth of 200 kHz. This seems to me to be the minimum requirement as it still allows possibilities for low distortion.
200 kHz is ten times the maximum audio which means in practice that 20 kHz signals (don't think I can hear 20 kHz) will be pretty accurately reproduced.
But the newer 30 MHz transistors may allow this to be increased. As the power ratings go up the speed of the output devices needs to rise too to maintain slew rate.
Many years ago a letter to Wireless World reported that "after 150 kHz all amplifiers tend to sound the same". This was only a letter with no data to back this observation up. This was based on subjective effects I presume.
Trouble is that input and output filters all act together and reduce the overall bandwidth. As most people like to have inductors in the output lead this reduces the overall bandwidth - so the amplifier has to stay high to maintain the performance.
Cheers
John
Gigapod:
Blocking DC is indeed a high pass filter.
The Low pass filter is the -3db bandwidth of the amp and typically 100 khz.
High pass: Pass frequencies above a certain -3db point. e.g 0.5 Hz.
Low pass: Pass frequencies below a given -3db point. e.g pass frequencies below 100 kHz
Bandpass: A combination of low and high pass filters.
I think your terminology needs some re-adjusting.
Blocking DC is indeed a high pass filter.
The Low pass filter is the -3db bandwidth of the amp and typically 100 khz.
High pass: Pass frequencies above a certain -3db point. e.g 0.5 Hz.
Low pass: Pass frequencies below a given -3db point. e.g pass frequencies below 100 kHz
Bandpass: A combination of low and high pass filters.
I think your terminology needs some re-adjusting.
john_ellis said:
John,
good post.
imo, an important factor of the rapidly increasing popularity of mosfets in the mid 1980s : very linear power amplifier designs with a CL bandwidth of 300KHz and very much higher. Unfortunately, going hand in hand with drastic NFB numbers.
Result=> a whole army of overanalytical products with a highly unnatural harmonic distr. spectrum
I just read through the thread
Opinions range on upper bandwidth figure from 50khz to 1meghz
but I did not notice that even one person that said that that they had tried a number of different high bandwidths and, based on these experiments, stated which bandwidth they prefer based on how they sounded.
and just to remind, I am talking about the amplifier internal bandwidth not including the input filter or output filter.
did anyone actually try this ?
darkfenriz - you have actually designed and built a very fast amp - what where the subjective benefit that you noticed with extra speed ?
do you think that any faster may improve sound even more ?
mike
Opinions range on upper bandwidth figure from 50khz to 1meghz
but I did not notice that even one person that said that that they had tried a number of different high bandwidths and, based on these experiments, stated which bandwidth they prefer based on how they sounded.
and just to remind, I am talking about the amplifier internal bandwidth not including the input filter or output filter.
did anyone actually try this ?
darkfenriz - you have actually designed and built a very fast amp - what where the subjective benefit that you noticed with extra speed ?
do you think that any faster may improve sound even more ?
mike
Bandwidth helps. What kills the sound is the creation of odd harmonics with solid state devices vs even for tube devices. So, distortion has to be lower for solid state amps to compete with a much higher distortion from tube amps Those odd harmonics are hard on the ears.
Next problem: What are you driving
1. A flame speaker
http://www.swtpc.com/mholley/PopularElectronics/May1968/Flame_Amplification.htm
2. Electrostatics
3. Horns
4. Domes
5. a 15" woofer
6 a 10" woofer
Each of these devices need different drivers (Amplifiers). Some are a high capacitance load, some highly inductive, some a voltage device, some a current device. One size doesn't fit all and neither does the type of music.
a 15" woofer has no business in hammer dulcimer music. Domes and electrostacs should preform well.
A horn and candle flame has no business in rock music.
Classical and Horns go nicely together. Horns and "valve" amps go nicely together.
Class D amps drive subwoffers well. A tube amp. It might grunt.
Electrostatics and domes need a fast amp. Use the rules, the voltage across a cap can't change instantaneouly and the current through an inductor can't change instantaneously so drive a capacitor (Electrostatics) with a voltage source (FET) and a voice coil with a current source (BJT). On order to drive a horn the voltage has to be high (transformer couples tube amps are great at this), but you might need a 300 W (BJT) amp to get the same sound.
Match the AMP to what's driven and don't compare tube amps to transistor amps but rather to what they excel in.
You might end up with a killer sound system if you could combine signal processing (FFT) to select a speaker appropriate for the signal and pair with it's best driver.
Think of it:
Guy hits a drum and it comes out of a 15" woofer with a class D amp.
Violin comes out of a horn with a tube amp.
A cymble comes out of a pair of electrostatics driven with a high speed amp with a BJT output.
A piano needs a nice midrange and a dome tweetr to hear the hammer attack.
Get the idea.
There's the guy that comes into the stereo store and asks one question: "How many watts is that?"
Another guy: "What's the distortion"?
Are they going to select a system based on the type of music they like to listen to and at what SPL level? Doubt it.
Next problem: What are you driving
1. A flame speaker
http://www.swtpc.com/mholley/PopularElectronics/May1968/Flame_Amplification.htm
2. Electrostatics
3. Horns
4. Domes
5. a 15" woofer
6 a 10" woofer
Each of these devices need different drivers (Amplifiers). Some are a high capacitance load, some highly inductive, some a voltage device, some a current device. One size doesn't fit all and neither does the type of music.
a 15" woofer has no business in hammer dulcimer music. Domes and electrostacs should preform well.
A horn and candle flame has no business in rock music.
Classical and Horns go nicely together. Horns and "valve" amps go nicely together.
Class D amps drive subwoffers well. A tube amp. It might grunt.
Electrostatics and domes need a fast amp. Use the rules, the voltage across a cap can't change instantaneouly and the current through an inductor can't change instantaneously so drive a capacitor (Electrostatics) with a voltage source (FET) and a voice coil with a current source (BJT). On order to drive a horn the voltage has to be high (transformer couples tube amps are great at this), but you might need a 300 W (BJT) amp to get the same sound.
Match the AMP to what's driven and don't compare tube amps to transistor amps but rather to what they excel in.
You might end up with a killer sound system if you could combine signal processing (FFT) to select a speaker appropriate for the signal and pair with it's best driver.
Think of it:
Guy hits a drum and it comes out of a 15" woofer with a class D amp.
Violin comes out of a horn with a tube amp.
A cymble comes out of a pair of electrostatics driven with a high speed amp with a BJT output.
A piano needs a nice midrange and a dome tweetr to hear the hammer attack.
Get the idea.
There's the guy that comes into the stereo store and asks one question: "How many watts is that?"
Another guy: "What's the distortion"?
Are they going to select a system based on the type of music they like to listen to and at what SPL level? Doubt it.
OOPs my reference to try to achieve -1db @ 50kHz is based on the whole amplifier including ALL components and is based on listening tests.mikelm said:
If the input filter is extracted from that and knowing that I generally use about 0.7 to 1uS then that is about 0.5 down @ 50kHz resulting in the amp (no filter) just about having to match the same 0.5db down @ 50kHz. This, assuming of course, that a small amount of amp peaking is not cancelling some of the droop coming from the passive input filter.
It appears that to meet my -1db @ 50kHz target, I would need about -3db @ between 160kHz and 200kHz from the amp without the input filter (assuming a single pole roll off at least initially).
I think "fast amplifiers" is the wrong engineering solution for a problem that could be solved otherwise.
You can't get a faster signal or one with less phase-distortion than the SOURCE delivers independant of using an amp with 50 kHz or 1 MHz bandwidth. And the usual source is heavily limited in this respect nowadays. The higher bandwith amp does just deteriorate the signal less than the slower one does in terms of phase.
But there would be cheaper solutions to this problems than using a power amp that is suitable as RF transmitter.
Regards
Charles
You can't get a faster signal or one with less phase-distortion than the SOURCE delivers independant of using an amp with 50 kHz or 1 MHz bandwidth. And the usual source is heavily limited in this respect nowadays. The higher bandwith amp does just deteriorate the signal less than the slower one does in terms of phase.
But there would be cheaper solutions to this problems than using a power amp that is suitable as RF transmitter.
Regards
Charles
mikelm said:
Well, it is hard to say.
At first I run the prototype heavily over-compesated, so having probably around 300kHz BW and <20V/us SR.
After changing compensation to 'optimal' cap value and probably improving BW to some 1.2Mhz and SR to around 90V/us and having much more feedback at 10-40kHz range, the difference was hardly noticable, maybe some trebles slightly improved along with resolution of percussion timbre (cymbals, rides, crashes), but this wasn't revolutionary improvement, maybe placebo effect as this wasn't AB comparison.
What's interesting the stability was actually better with optimum compensation compared to over-lagging.
regards
Adam
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.