Hi,
I do not agree.
First, the "punch" and "drive" in Rock is almost invariably heavy compression (trust me, I know, I used to sit at the mixing desk for an East German Hard Rock band, our FX rack had more compressors than the audiences had Secret Police informers I have seen even worse in the West).
Moreover, I am unclear what a "so-called choke input supply" is. It either IS a choke input supply and thus draws current via the rectifier throughout the whole cycle or it is a capacitor input supply, even if only a small value capacitor (a few uF) is used upfront, or it is a capacitor input supply with a LC filter after. It cannot be "so-called".
In fact, compared to capacitor input supplies true choke input supplies deliver by far superior regulation and thus less supply sag than conventional Capacitor input supplies. The problem is that chokes needed for this are nearly as massive the mains transformers.
So at least technically speaking in the correct application the axiom of your countrymen is wrong. In fact, I have encountered many axioms among audiophiles worldwide which are often directly contradicting to each other (so by definition one group must be wrong) and equally often contradict reality.
Well, most speakers voice coils will near instantly vapourise with a real 60A. So given that I am not trying to make exploding wire magnetisators but Amplifiers for listening to music I am not worried about delivering 60A. If we assume a Speaker with 1.6 Ohm DCR for the Voice Coil (two 4 Ohm drivers in parallel) we need a minimum of around +/-100V Rails for Solid State Amplifiers and deliver pulse power of 6KW.
For speakers with a more normal Impedance (say 3.2 Ohm for two paralleled 8 Ohm Drivers) the requirements for the rails in terms of Voltage and the peak voltages are simply ridiculous.
I cannot imagine any such situation under domestic or even studio conditions, but as I remarked elsewhere, I don't have an imagination.
I find that saying such things is meaningless unless we include the Speakers. Once we do I partially agree, insofar that the more compromised the amplifiers and Speakers are the more true I fond this "If they are good at one music they are not with another".
I have found at least one system (there is a second one but I do not want to be falsely accused of bearing the rules and promoting products I have involvement with so I shall not mention that) that does perform very even handed and is probably the best I had in my long terms for Jazz, Classical, Rock, Pop and even (classic) Rap and Solo Pipe Organ. It is also absolutely superb with movies... Dynamics were explosive and realistic, except for movies, where they are hyper realistic.
In this case we are talking about 15" Tannoy Monitor Red's in Corner enclosures (some might call them corner horns, but strictly speaking they are not - measured at 97dB/2.83V/1m and posessing a 16 Ohm rated impedance - thus a real 100dB/1W/1m). Lower cutoff was somewhere a bit below 30Hz.
They where driven by 300B SE Amplifiers which I optimised for clipping / overload behaviour in addition to linearity inside the intended power range. This amplifier could be driven with peaks that are at least 6dB above the rated 7 Watt without sounding subjectively distorted or even compressed. The output stage incorporated a kind of "expansive variable quiescent current" regulation scheme, that reduced effective bias voltage and increased anode current more the more the amplifier was over driven (grid current to the 300B)...
The Amplifier of course used choke input supplies (not so-called ones) and further LC filtering, no electrolytic capacitors and relatively low values of power supply capacitance and additional noise cancellation schemes which incidentally also opposed any power supply modulation by the Amplifier (they cancelled > 26dB any disturbance on the power supply lines, no matter if mains noise or signal modulation).
Incidentally, most of these design techniques can be found among others in 1930's WE Theatrical Amplifiers and Telefunken Public Address amplifiers, they are nothing new, nothing I invented or claim as my original work. Most of these techniques cannot be ported directly to solid state gear, however it is possibly to learn from them and get similar results using different circuitry.
I eventually planned to make an 18 Watt version of this Amplifier using one of these Super 300B's, but stuff came up and it never happened. In the fullness of time I developed other design techniques that allow me more power with a similar overload handling using standard 300B's and I also have worked on other Amplifiers and a variety of speakers, but that, as they say is another story.
Ciao T
I do not agree.
First, the "punch" and "drive" in Rock is almost invariably heavy compression (trust me, I know, I used to sit at the mixing desk for an East German Hard Rock band, our FX rack had more compressors than the audiences had Secret Police informers I have seen even worse in the West).
Moreover, I am unclear what a "so-called choke input supply" is. It either IS a choke input supply and thus draws current via the rectifier throughout the whole cycle or it is a capacitor input supply, even if only a small value capacitor (a few uF) is used upfront, or it is a capacitor input supply with a LC filter after. It cannot be "so-called".
In fact, compared to capacitor input supplies true choke input supplies deliver by far superior regulation and thus less supply sag than conventional Capacitor input supplies. The problem is that chokes needed for this are nearly as massive the mains transformers.
So at least technically speaking in the correct application the axiom of your countrymen is wrong. In fact, I have encountered many axioms among audiophiles worldwide which are often directly contradicting to each other (so by definition one group must be wrong) and equally often contradict reality.
Well, most speakers voice coils will near instantly vapourise with a real 60A. So given that I am not trying to make exploding wire magnetisators but Amplifiers for listening to music I am not worried about delivering 60A. If we assume a Speaker with 1.6 Ohm DCR for the Voice Coil (two 4 Ohm drivers in parallel) we need a minimum of around +/-100V Rails for Solid State Amplifiers and deliver pulse power of 6KW.
For speakers with a more normal Impedance (say 3.2 Ohm for two paralleled 8 Ohm Drivers) the requirements for the rails in terms of Voltage and the peak voltages are simply ridiculous.
I cannot imagine any such situation under domestic or even studio conditions, but as I remarked elsewhere, I don't have an imagination.
I find that saying such things is meaningless unless we include the Speakers. Once we do I partially agree, insofar that the more compromised the amplifiers and Speakers are the more true I fond this "If they are good at one music they are not with another".
I have found at least one system (there is a second one but I do not want to be falsely accused of bearing the rules and promoting products I have involvement with so I shall not mention that) that does perform very even handed and is probably the best I had in my long terms for Jazz, Classical, Rock, Pop and even (classic) Rap and Solo Pipe Organ. It is also absolutely superb with movies... Dynamics were explosive and realistic, except for movies, where they are hyper realistic.
In this case we are talking about 15" Tannoy Monitor Red's in Corner enclosures (some might call them corner horns, but strictly speaking they are not - measured at 97dB/2.83V/1m and posessing a 16 Ohm rated impedance - thus a real 100dB/1W/1m). Lower cutoff was somewhere a bit below 30Hz.
They where driven by 300B SE Amplifiers which I optimised for clipping / overload behaviour in addition to linearity inside the intended power range. This amplifier could be driven with peaks that are at least 6dB above the rated 7 Watt without sounding subjectively distorted or even compressed. The output stage incorporated a kind of "expansive variable quiescent current" regulation scheme, that reduced effective bias voltage and increased anode current more the more the amplifier was over driven (grid current to the 300B)...
The Amplifier of course used choke input supplies (not so-called ones) and further LC filtering, no electrolytic capacitors and relatively low values of power supply capacitance and additional noise cancellation schemes which incidentally also opposed any power supply modulation by the Amplifier (they cancelled > 26dB any disturbance on the power supply lines, no matter if mains noise or signal modulation).
Incidentally, most of these design techniques can be found among others in 1930's WE Theatrical Amplifiers and Telefunken Public Address amplifiers, they are nothing new, nothing I invented or claim as my original work. Most of these techniques cannot be ported directly to solid state gear, however it is possibly to learn from them and get similar results using different circuitry.
I eventually planned to make an 18 Watt version of this Amplifier using one of these Super 300B's, but stuff came up and it never happened. In the fullness of time I developed other design techniques that allow me more power with a similar overload handling using standard 300B's and I also have worked on other Amplifiers and a variety of speakers, but that, as they say is another story.
Ciao T
Sorry, but it is IMO completely untrue, both from empirical and technical point of wiev..
Amplifier are intended to amplify, independent to genre of music.,Try to think about about role of coil before filtering capacitors (coils "behind" filtering capacitors are technical nonsense..) . It will only lower peak output voltage by some hundreds mV, but it will lower ripple, too. Or You are using PSU without filtering capacitors??
I have no clue why fuses have impact on the sound. All I know, empirically, is that they do.
Well, you are right about there is no use talking about fuses.
The reason I looked at the fuse is that in all my other gear there are Hi-Fi tuning fuses installed. So I wanted to see the value of the fuse in that CDP. Seeing that it's a propriety fuse, I decided to let it be. Out of curiosity, after the burning in, I'll try to replace it, though I doubt if any other fuse will make any improvement. I learned to trust your design and parts choosing decisions.
You often play into short circuit??
Technically there is no problem to monitor supply rails with A/D converter and memory (even behind the fuse, if someone mentions fuse non-linearity as a source of problems). Same for output voltage of the amplifier playing into real speaker. I do it regularly.
It is nice to care, but some audiophile reasoning and explanations are driving me mad
No problem, and my apologies for the tardy response.
If you look at the frequency response graphs I publish in Stereophile - for example, see figs.1 & 2 at Music Reference RM-200 Mk.II power amplifier Measurements | Stereophile.com - I do publish the response into resistive loads ranging from 2 ohms to 8 or 16 ohms, as well as into the simulated loudspeaker.
I don't follow, Surely it's just a matter of Ohm's Law?
John Atkinson
Editor, Stereophile
That's a big cap. Often over an inch in diameter or one side in the box shape. Wouldn't it be fairly low impedance at 60Hz?
Do you put it across hot and neutral, or to ground? Have you found any differences there?
Guys, I propose to differ between AB amps and A amps, and to think about whether LC filter (choke input) could really give benefits for AB amp.
I do not use AB, I use A, and even more, a load connected in parallel to active device. My case is just the best application for LC power supply filter. But I find reasonable to go with consequent RC-RC-RC-RC, simpler and could be even better.
L and C are chosen from definite requirements, in particular, for definite current consumption from power supply. If the last is strongly varying like in AB amp case, how LC can be used effectively?
Consider bass note lasting 1s, with average current to load 10A. That means 10 C (coulomb) charge is transferred to load during 1s. Typical lytics reservoir lets say 0,1F, would require 100V voltage drop in order to provide such charge. This simply indicates, that in this examplary case output current must be provided by increased current consumption from PS, not from lytics.
If one use L taken for 2A consumption, what will be effect of this L for 10A demand?
I do not use AB, I use A, and even more, a load connected in parallel to active device. My case is just the best application for LC power supply filter. But I find reasonable to go with consequent RC-RC-RC-RC, simpler and could be even better.
L and C are chosen from definite requirements, in particular, for definite current consumption from power supply. If the last is strongly varying like in AB amp case, how LC can be used effectively?
Consider bass note lasting 1s, with average current to load 10A. That means 10 C (coulomb) charge is transferred to load during 1s. Typical lytics reservoir lets say 0,1F, would require 100V voltage drop in order to provide such charge. This simply indicates, that in this examplary case output current must be provided by increased current consumption from PS, not from lytics.
If one use L taken for 2A consumption, what will be effect of this L for 10A demand?
John, I think Wavebourn is talking about the acoustic SPL out of the loudspeaker does not necessarily correlate to the dips and peaks of the voltage on the terminals, which is determined roughly by Ohms law (with complex variables). I never examined it myself so I don't know how much non-minimum phase behavior this voltage/current can exhibit but I would surmise it is small, while the SPL can have measurable non-minimum phase behavior.
Last edited:
If you are saying that mains current drawn by the PSU of a Class B amplifier varies at the syllabic rate of the music then you are correct, but so what? That is common knowledge. You just have to ensure that any LC resonances in the PSU are absent (no L), well damped (C not too big) or well away in frequency (C not too big). The modern trend to use huge caps is sometimes based on naive 'bigger is better engineering' rather than real calculations.VladimirK said:
Dear Vladimir,
Well, choke filtered supplies where de rigeur for high power Tube PA Amplifiers operating in Class AB precisely because of better regulation. There is however an issue.
Cored chokes (as opposed to air core ones) will saturate at a certain value of current and for choke input supplies (not CLC) to work correctly a certain minimum current must flow, or the supply will no longer work correctly. In tube Amplifiers even most notional Class B tube circuits have a fair bit of quiescent current, so this is easy to do, however in transistor circuits only very warm running AB Amplifier may apply for true choke input supplies.
As far as I know only musical fidelity commercially made an amplifier with a choke input supply (choke input as distinct from choke filtered) and that was a fairly low powered near Class A job that you could fry eggs upon (I repaired one).
You need to have quiescent current to stay above the critical current of the choke and you need to have a large enough core to not saturate at high current draw.
There are some kinds of chokes build called "swinging chokes" (they where popular in the 60's) which have a non-linear inductance that rises strongly at low currents, so maintaining choke input operation for a wider range of current. Again, I have not seen such applied to SS AB Amplifiers, but in principle there is no issue in doing so.
I am sorry Vladimir, I cannot do that. (to be said in a clam, slightly mechnical voice)
Almost all musical notes consist of attack, sustain and different speeds and levels of decay. On old analogue synth's you could actually dial these up on convenient rotaries, playing with the resulting sounds provides oodles and hours of fun, all of which was lost when keyboards became Samplers and digital synthesis (preprogrammed samplers). They did however sound a lot more like real instruments, which is what most people wanted. Still, there is something hypnotic to firing up a MiniMoog and try some harmonies while twiddeling the nobs...
A bass note that has a 1 second sustain and a really substantial level is not something I have ever encountered. I suspect a bass player attempting to get acoustic feedback Jimmy Hendrix Style at the low E (not that I have ever heard it done) might manage, with great effort.
But in reality an actual note, even if the decay last for over a second will be at high power only for a small fraction of that time before it decays.
That would be DC, not music.
Yet, as it's charge is replenished 100 times or 120 times per seconds it only needs to sustain the 10A for 10mS (Europe) or 8.3mS (US). So in fact the voltage drop is only 1V in that period, another way is to say 1V ripple...
The stored charge (stored for 10mS or less) must be taken from the supply to be stored. This may need a very large current spike if the conduction angle is low (current is drawn only at the mains waveform peak -typical capacitor input supplies) but it may be a sine wave current without large peaks if a choke input supply is used, as the choke stores the charge, not the capacitor, or more precisely, the capacitor is constantly recharged by the energy in the choke as it is discharged by the load.
If one uses a choke limited to 2A where one for 10A is needed I suspect one should hand back ones degree and retake EE101.
Ciao T
Well, choke filtered supplies where de rigeur for high power Tube PA Amplifiers operating in Class AB precisely because of better regulation. There is however an issue.
Cored chokes (as opposed to air core ones) will saturate at a certain value of current and for choke input supplies (not CLC) to work correctly a certain minimum current must flow, or the supply will no longer work correctly. In tube Amplifiers even most notional Class B tube circuits have a fair bit of quiescent current, so this is easy to do, however in transistor circuits only very warm running AB Amplifier may apply for true choke input supplies.
As far as I know only musical fidelity commercially made an amplifier with a choke input supply (choke input as distinct from choke filtered) and that was a fairly low powered near Class A job that you could fry eggs upon (I repaired one).
You need to have quiescent current to stay above the critical current of the choke and you need to have a large enough core to not saturate at high current draw.
There are some kinds of chokes build called "swinging chokes" (they where popular in the 60's) which have a non-linear inductance that rises strongly at low currents, so maintaining choke input operation for a wider range of current. Again, I have not seen such applied to SS AB Amplifiers, but in principle there is no issue in doing so.
I am sorry Vladimir, I cannot do that. (to be said in a clam, slightly mechnical voice)
Almost all musical notes consist of attack, sustain and different speeds and levels of decay. On old analogue synth's you could actually dial these up on convenient rotaries, playing with the resulting sounds provides oodles and hours of fun, all of which was lost when keyboards became Samplers and digital synthesis (preprogrammed samplers). They did however sound a lot more like real instruments, which is what most people wanted. Still, there is something hypnotic to firing up a MiniMoog and try some harmonies while twiddeling the nobs...
A bass note that has a 1 second sustain and a really substantial level is not something I have ever encountered. I suspect a bass player attempting to get acoustic feedback Jimmy Hendrix Style at the low E (not that I have ever heard it done) might manage, with great effort.
But in reality an actual note, even if the decay last for over a second will be at high power only for a small fraction of that time before it decays.
That would be DC, not music.
Yet, as it's charge is replenished 100 times or 120 times per seconds it only needs to sustain the 10A for 10mS (Europe) or 8.3mS (US). So in fact the voltage drop is only 1V in that period, another way is to say 1V ripple...
The stored charge (stored for 10mS or less) must be taken from the supply to be stored. This may need a very large current spike if the conduction angle is low (current is drawn only at the mains waveform peak -typical capacitor input supplies) but it may be a sine wave current without large peaks if a choke input supply is used, as the choke stores the charge, not the capacitor, or more precisely, the capacitor is constantly recharged by the energy in the choke as it is discharged by the load.
If one uses a choke limited to 2A where one for 10A is needed I suspect one should hand back ones degree and retake EE101.
Ciao T
Last edited:
Correct, and every charging pulse must deliver 0,1 Coulomb, what can be hardly achievable with typically 2H choke standing just after diode bridge. This is clear, that voltage drops at the capacitor will be larger with choke than without it.
For tube amps, this is smaller among various devils, since nobody expect them to be especially "punchy".
I think you need to read up on the theory of choke input supplies. The 2H choke can supply as much current as you want (limited by DC resistance of course). It can still provide current going forward even when the choke itself is reversed biassed, so it holds the rectifier on for longer. Choke input supplies do not have the narrow charging pulses of a cap input supply.VladimirK said:
Vladimir,
Actually, IF you had a 2H that would not saturate at 10A it would do that splendidly. Pavel showed one supply, allow me to up the primary voltage, add 2H inductance and use Duncanamps PSUD for simulation.
Attached is one PSU like Pavel's Sim, showing the charging pulses through the capacitor and another with the 2H Choke added, which shows the current through the choke. The ripple is mostly a sinewave and the charge current is mostly DC. If you looked at the current through the Diode you would see that each diode conducts a constant current through it's halve of the mains cycle...
It is also worthy of note that the ripple of the capacitor input version is around 2V Peak-Peak, while the choke input supply has a ripple of a few mV peak-peak.
The reason is that the energy storage in the supply has effectively moved to the Choke, from the capacitors, who now have not much of a job. In fact, if you have a constant current draw circuit to supply, you can completely dispense with capacitors and use only inductors and shunt regulators and get a perfectly working supply!
Yes, it will be, because the output from a Choke input supply is around 0.9 * VRMS while a capacitor input supply is 1.41 * VRMS, approximately.
However, if resistive losses in the choke are contained the choke input supply will hold it's 0.9 * VRMS output by far stiffer (less variation in Volt for a given current variation) than the capacitor input supply holds it's output voltage.
This kind of knowledge may be lost on the new generations who knoweth not the might valve amplifier, but it remains completely true.
Ciao T
Actually, IF you had a 2H that would not saturate at 10A it would do that splendidly. Pavel showed one supply, allow me to up the primary voltage, add 2H inductance and use Duncanamps PSUD for simulation.
Attached is one PSU like Pavel's Sim, showing the charging pulses through the capacitor and another with the 2H Choke added, which shows the current through the choke. The ripple is mostly a sinewave and the charge current is mostly DC. If you looked at the current through the Diode you would see that each diode conducts a constant current through it's halve of the mains cycle...
It is also worthy of note that the ripple of the capacitor input version is around 2V Peak-Peak, while the choke input supply has a ripple of a few mV peak-peak.
The reason is that the energy storage in the supply has effectively moved to the Choke, from the capacitors, who now have not much of a job. In fact, if you have a constant current draw circuit to supply, you can completely dispense with capacitors and use only inductors and shunt regulators and get a perfectly working supply!
Yes, it will be, because the output from a Choke input supply is around 0.9 * VRMS while a capacitor input supply is 1.41 * VRMS, approximately.
However, if resistive losses in the choke are contained the choke input supply will hold it's 0.9 * VRMS output by far stiffer (less variation in Volt for a given current variation) than the capacitor input supply holds it's output voltage.
This kind of knowledge may be lost on the new generations who knoweth not the might valve amplifier, but it remains completely true.
Ciao T
Attachments
- Status
- Not open for further replies.