Agree, properly designed 2H choke can provide even 100A, but power supply will be very unhappy if 500A will be needed during 1s. Chokes do not like dynamically varying consumption above their rated current.
One could simply simulate two different cases we are speaking about, with choke (chosen for 1-2A) and without choke, and compare voltage drop at C during 1s high current (10A) consumption.
Vladimir,
But the charge in the choke is also replenished every 10mS and come on, 60A was unrealistic, 500A! Get off it already and keep it real.
Of course not. But why would anyone use a choke in an application where it's rated saturation would be dramatically exceeded?
That is like by far exceeding the SOAR of a Semiconductor in an Amp you design and to them complain the transistors where no good because they blew up!
Why on earth do you insist on using a 2A rated choke where a 10A one is needed? Who would do something like that?
Ciao T
But the charge in the choke is also replenished every 10mS and come on, 60A was unrealistic, 500A! Get off it already and keep it real.
Of course not. But why would anyone use a choke in an application where it's rated saturation would be dramatically exceeded?
That is like by far exceeding the SOAR of a Semiconductor in an Amp you design and to them complain the transistors where no good because they blew up!
Why on earth do you insist on using a 2A rated choke where a 10A one is needed? Who would do something like that?
Ciao T
Pavel, L must be high enough in order current through choke is not interrupted (in contra to simple C input, when current flows as short pulses only). This is one of requirements for correct operation of LC filter. Try something near 1H.
Rough formular is L > coeff * Rload max / f , i.e. L is chosen for minimum consumed current, idle current for AB amp. With lower L, that will not be correct choke-input filter.
Last edited:
Originally Posted by Wavebourn View Post
Also, dips and peaks on impedance curve of loudspeakers are not directly translated into dips and peaks on sound pressure curve.
In professional loudspeaker systems there actually are two schools of design. One uses very rugged and lower efficiency loudspeakers with greater amplifier power used. That appeals to those who buy their rental PA by the watt! Having a 100,000 watt PA sounds much more impressive than a smaller 10,000 watt unit.
However for the arenas I work in efficiency is a good thing. One well known issues is that as frequency halves cone motion must increase to maintain the same sound pressure level. Or the inverse is that as frequency increases your driver should have a rising response with frequency. Now if such a driver were presented to the domestic use crowd they would be upset that the driver did not have a flat frequency response. So those units are designed such that the cone mass and other issues are balanced to provide a flat frequency response.
With the drivers designed for maximum efficiency (around 30% of the energy in is converted to sound energy) port resonances really show up. I am sure you are aware of how at the low frequency resonances the impedance shoots up and back down from acoustic loading and then rises again due to impedance increases.
However if you take a look at the older and new driver data sheets you can see both drivers have increasing output up to about 125 hz. The newer driver also has more of an on axis frequency rise at high frequencies due to less cone break up (My Assumption.)
Now the issue is that if you were listening to the newer driver with out any equalization you would find an amplifier with rising output impedance most likely would sound better. Of course in a large system this would translate into less amplifier power required.
But back to the original premise that dips and peaks in the impedance curve do not show up in the output frequency response, I think that is very clear from these examples designed to be driven from a voltage source amplifier.
http://www.jblpro.com/pub/components/2206.pdf
http://www.jblpro.com/pub/components/2227.pdf
ES
Also, dips and peaks on impedance curve of loudspeakers are not directly translated into dips and peaks on sound pressure curve.
In professional loudspeaker systems there actually are two schools of design. One uses very rugged and lower efficiency loudspeakers with greater amplifier power used. That appeals to those who buy their rental PA by the watt! Having a 100,000 watt PA sounds much more impressive than a smaller 10,000 watt unit.
However for the arenas I work in efficiency is a good thing. One well known issues is that as frequency halves cone motion must increase to maintain the same sound pressure level. Or the inverse is that as frequency increases your driver should have a rising response with frequency. Now if such a driver were presented to the domestic use crowd they would be upset that the driver did not have a flat frequency response. So those units are designed such that the cone mass and other issues are balanced to provide a flat frequency response.
With the drivers designed for maximum efficiency (around 30% of the energy in is converted to sound energy) port resonances really show up. I am sure you are aware of how at the low frequency resonances the impedance shoots up and back down from acoustic loading and then rises again due to impedance increases.
However if you take a look at the older and new driver data sheets you can see both drivers have increasing output up to about 125 hz. The newer driver also has more of an on axis frequency rise at high frequencies due to less cone break up (My Assumption.)
Now the issue is that if you were listening to the newer driver with out any equalization you would find an amplifier with rising output impedance most likely would sound better. Of course in a large system this would translate into less amplifier power required.
But back to the original premise that dips and peaks in the impedance curve do not show up in the output frequency response, I think that is very clear from these examples designed to be driven from a voltage source amplifier.
http://www.jblpro.com/pub/components/2206.pdf
http://www.jblpro.com/pub/components/2227.pdf
ES
Hi,
The critical current in mA is approximatly output Voltage (35V) divided by Inductance (1H). This is the kind of shortcut formula you know well if you ever worked with tube gear.
We can re-arrange this to solve for inductance at a given current and voltage.
So for a minimum current of (say) 1000mA and 35V we get 35mH. For current of 10A steady around 3.5mH are needed.
Ciao T
The critical current in mA is approximatly output Voltage (35V) divided by Inductance (1H). This is the kind of shortcut formula you know well if you ever worked with tube gear.
We can re-arrange this to solve for inductance at a given current and voltage.
So for a minimum current of (say) 1000mA and 35V we get 35mH. For current of 10A steady around 3.5mH are needed.
Ciao T
For 50V voltage and 1H one gets 50mA, that is close for idle current of AB amp.
If for such amp we take 0,1H, then at low signals current through the choke will be pulse like, almost like without choke.
This makes clear again, that for class A SE amps using of chokes is OK, but for AB amp it is doubtful.
Dear Vladimir,
I am afraid you read too many audiophile magazines and visit too many audiophile forums.
Pavel; it is a sub-culture called Audioportal. They have gurus, myths, they vote democratically for better rules of thumbs, like amount of capacitance needed per watt, and so on. Since Vladimir is here on this international forum he has a good chance to question that myths and come to own conclusion. However, we are as well full of other myths, though...
Vladimir,
First, AB Amp biased optimally needs 26mV across the emitter restsor of the BJT, this is hard to garantee, but let's run with this and use 3 pairs of BJT's as outputs on 50V rails (about adequate I'd say) and 0.22 Ohm emitter resistors. Our Iq will be 3 * 26mV / 0.22 Ohm or 350mA.
This means we need 0.14H. Though, if we go to the expense of having choke input supplies (not an insignificant expense) we can probably afford a few more pairs and the heatsink's to go, so lets double up to 6 Pairs of outputs and 700mA Iq and we only need around 70mH with, if we limit ourselves to continous output into 4 Ohm only 9A current handling. Of course this is still a lot of inductance.
As I remarked before, only very warm running Class AB.
Well, the only real problem here is that the rail voltage and ripple will rise a lot when the choke drops out. If this is a "bad thing" or not is down to many factors. If all the parts have the necesary voltage rating, why not? It would depend on the rest of the circuit how much a problem all this could cause.
But as remarked, you can make a choke that the gap in the choke's core partially bridged with iron lams (or other methods may be used to achieve the same results), so at low currents this partial iron path is actually not saturated and the choke's inductance rises, I have seen these with a 1:10 ratio between "low current" and "just before saturation" current.
So a 10mH choke at 10A may very well be 100mH at 100mA...
No, sorry, it does not make it doubtful at all. Just do not expect your chokes cores to be much small than those of the mains transformer... That's all.
The Chokes used in the PSU's of some of the gear I designed have double C-cores big enough for 50VA mains transformers, yet they are only rated for 20mA (but at 40H)... The mains transformer feeding them has a 120VA core!
Ciao T
First, AB Amp biased optimally needs 26mV across the emitter restsor of the BJT, this is hard to garantee, but let's run with this and use 3 pairs of BJT's as outputs on 50V rails (about adequate I'd say) and 0.22 Ohm emitter resistors. Our Iq will be 3 * 26mV / 0.22 Ohm or 350mA.
This means we need 0.14H. Though, if we go to the expense of having choke input supplies (not an insignificant expense) we can probably afford a few more pairs and the heatsink's to go, so lets double up to 6 Pairs of outputs and 700mA Iq and we only need around 70mH with, if we limit ourselves to continous output into 4 Ohm only 9A current handling. Of course this is still a lot of inductance.
As I remarked before, only very warm running Class AB.
Well, the only real problem here is that the rail voltage and ripple will rise a lot when the choke drops out. If this is a "bad thing" or not is down to many factors. If all the parts have the necesary voltage rating, why not? It would depend on the rest of the circuit how much a problem all this could cause.
But as remarked, you can make a choke that the gap in the choke's core partially bridged with iron lams (or other methods may be used to achieve the same results), so at low currents this partial iron path is actually not saturated and the choke's inductance rises, I have seen these with a 1:10 ratio between "low current" and "just before saturation" current.
So a 10mH choke at 10A may very well be 100mH at 100mA...
No, sorry, it does not make it doubtful at all. Just do not expect your chokes cores to be much small than those of the mains transformer... That's all.
The Chokes used in the PSU's of some of the gear I designed have double C-cores big enough for 50VA mains transformers, yet they are only rated for 20mA (but at 40H)... The mains transformer feeding them has a 120VA core!
Ciao T
Last edited:
Both of you guys are 'sophomores' in choke input design. You are both right, and both wrong, and both 'half baked', and both opinionated without good cause.
I have at my office a dual 50mH 10A (continuous) choke. It weights at least 35 pounds, and I used it for a door stop, for years. It was custom made for the Gale amplifier in 1976 that was never put into production. It was 'John's folly' to specify such a thing, but it would have worked well, just like you two have implied. However, it is DANGEROUS to make choke input power supplies, unless you know that you have a Class A load, ALWAYS, somehow, in some way. IF you significantly reduce the load, the voltage will rise to peak from average and perhaps explode the power supply caps, unless they are overrated in voltage to handle this condition. You might even get shock ringing on top of peak voltage. It is very difficult to make a Class AB amplifier, with less than 1A continuous loading with a choke input. However, many tube amps have used 1H chokes, including (I think) the Dyna Stereo 70 tube amp, but as a pi section filter. NOW, you don't get everything that pure choke input will give you in current smoothness, but you get plenty of ripple reduction. It is a good compromise.
Speculations about peak current, it is double your calculation Thorsten, and sustained currents: It gets refreshed every 1/100th of a second, so 10mS is a good working number, just like PMA stated.
Dynamic range, you should have gone to a live GD concert and you would find an incredible dynamic range in some of the songs. One example, I remember, completely confusing a Burwen 3dB into one dB compressor/expander that we were trying out. It was laughable! Enough said for now.
I have at my office a dual 50mH 10A (continuous) choke. It weights at least 35 pounds, and I used it for a door stop, for years. It was custom made for the Gale amplifier in 1976 that was never put into production. It was 'John's folly' to specify such a thing, but it would have worked well, just like you two have implied. However, it is DANGEROUS to make choke input power supplies, unless you know that you have a Class A load, ALWAYS, somehow, in some way. IF you significantly reduce the load, the voltage will rise to peak from average and perhaps explode the power supply caps, unless they are overrated in voltage to handle this condition. You might even get shock ringing on top of peak voltage. It is very difficult to make a Class AB amplifier, with less than 1A continuous loading with a choke input. However, many tube amps have used 1H chokes, including (I think) the Dyna Stereo 70 tube amp, but as a pi section filter. NOW, you don't get everything that pure choke input will give you in current smoothness, but you get plenty of ripple reduction. It is a good compromise.
Speculations about peak current, it is double your calculation Thorsten, and sustained currents: It gets refreshed every 1/100th of a second, so 10mS is a good working number, just like PMA stated.
Dynamic range, you should have gone to a live GD concert and you would find an incredible dynamic range in some of the songs. One example, I remember, completely confusing a Burwen 3dB into one dB compressor/expander that we were trying out. It was laughable! Enough said for now.
Hi,
Ahh. A bit like the subculture I have ties with (Sound Practices). There we have a very simple measure of the quality of Amplifiers.
It is called the Dunker Faktor, expressed as the ratio of Weight in Kg divided by Power in Watt. Higher is better.
I believe the record currently stands around a DF of 50. But I heard for the next ETF some guys are shooting at 200...
Ciao T
Ahh. A bit like the subculture I have ties with (Sound Practices). There we have a very simple measure of the quality of Amplifiers.
It is called the Dunker Faktor, expressed as the ratio of Weight in Kg divided by Power in Watt. Higher is better.
I believe the record currently stands around a DF of 50. But I heard for the next ETF some guys are shooting at 200...
Ciao T
One more thing would be interesting to learn from you, John. Could you predict some changes in sound capabilities of your JC-1 monoblocks, if you would put 100mH chokes just after diode bridges? What kind of changes could happen?
John,
I have a pair somewhere that where made for an abortive 833 SE Amp for the filament, similar size/weight... In 1976 I probably was near winding my first pair of output transformers, by hand, with a bicycle counter and crank... Chokes too.
I disagree. There is no magic here. Choke input supplies are trivial to design in most cases, unless cost is an object. Incidentally, if I have the choice between a choke or a tube rectifier, I take the choke...
The max is around a 1.5 time rise. One would hope that anyone designing powersupplies did not sleep through EE101.
If you read the old books, you can add a snubber for this...
Well, I would not say that it is very difficult, but I would not want to lift a 150Watt/8R and 2 Ohm compatible Stereo Class AB amp with a Choke input supply (not even using swinging chokes at that) and I am certainly a Big Boi...
I agree, I posted a few pages back on what I have assembled for my next pet (not commercial) project, which would scale nicely to tubes...
Sorry, I don't follow. I stated the refresh every 10mS (8.3mS for the US at 120Hz) and I gave no peak current...
Sure. As we say at home, the exceptions underline the rule.
Try going to concert right now measure the dynamics... I fear they are trying hard for DC, to go with AC... But it sure is LOUD!
Ciao T
I have a pair somewhere that where made for an abortive 833 SE Amp for the filament, similar size/weight... In 1976 I probably was near winding my first pair of output transformers, by hand, with a bicycle counter and crank... Chokes too.
I disagree. There is no magic here. Choke input supplies are trivial to design in most cases, unless cost is an object. Incidentally, if I have the choice between a choke or a tube rectifier, I take the choke...
The max is around a 1.5 time rise. One would hope that anyone designing powersupplies did not sleep through EE101.
If you read the old books, you can add a snubber for this...
Well, I would not say that it is very difficult, but I would not want to lift a 150Watt/8R and 2 Ohm compatible Stereo Class AB amp with a Choke input supply (not even using swinging chokes at that) and I am certainly a Big Boi...
I agree, I posted a few pages back on what I have assembled for my next pet (not commercial) project, which would scale nicely to tubes...
Sorry, I don't follow. I stated the refresh every 10mS (8.3mS for the US at 120Hz) and I gave no peak current...
Sure. As we say at home, the exceptions underline the rule.
Try going to concert right now measure the dynamics... I fear they are trying hard for DC, to go with AC... But it sure is LOUD!
Ciao T
- Status
- Not open for further replies.