Hi, EVA,
Do you mean current flowing through R5 to the load (speaker)? I think there is no current will flow from A to B to speaker.
The current that will drive the loudspeaker will be from collectors of T6 and T7, none from R5.
The function of R5 is only as a "voltage sensor" for T6 and T7 to determine whether they should give or sucks current. There is current here in R5 (between A and B), but that's it, it will never reach the speaker.
If you look the "Alexander Current Feedback Amplifier", maybe it will be clearer that R5 is forming a voltage sensor only.
Do you mean current flowing through R5 to the load (speaker)? I think there is no current will flow from A to B to speaker.
The current that will drive the loudspeaker will be from collectors of T6 and T7, none from R5.
The function of R5 is only as a "voltage sensor" for T6 and T7 to determine whether they should give or sucks current. There is current here in R5 (between A and B), but that's it, it will never reach the speaker.
If you look the "Alexander Current Feedback Amplifier", maybe it will be clearer that R5 is forming a voltage sensor only.
Just to clarify - here is the simulated circuit. R17 (like lumanauw's R5) and R18 are normally not used, and Q3 emitter connected to Q5 collector, Q4 emitter to Q6 collector. The gain is close to +1. After connecting the R17 only, the gain is decreased. After connecting the R17 with R18, the both resistors create feedback divider and the overal gain is more than +1.
For the values shown, the gain is about 15dB.
For the values shown, the gain is about 15dB.
Attachments
Hi, PMA,
In your simulation, the gain will be far less than +1, because the ratio of the value R7/R17 (or R8/R17) is only 1/3. The local EC-CFP of Q5-Q6 in your simulation only senses 1/3 of the error happened between A-B (in R17 from your sch).
If you put R17=10ohm, then the error voltage magnification between A-B will be amplified by factor of 10 (100ohm/10ohm). This way, the whole gain will be closer to +1 (but will never get to +1).
Hi, EVA
It is true, that this EC will never get the perfect +1 gain. It will always less than +1, no matter how small it is.
But this is not caused because R5 (Or R17 in PMA's) has to contribute driving the load/output stage (even a very small fraction), but because the little fraction of current has to drive the EC mechanism itself. This is because the whole EC mechanism doesn't have external current source / external transistor, so the energy to drive it has to be drawn from the system itself. In Hawksford it has external current source and external transistor, so the gain can be made <1 or set =1 or made>1 (negative impedance). This EC cannot do that, it cannot even reach the perfect +1gain. (but I think this is a merit, it is harder to oscilate
)
Infact the current is happening "Negative way". In positive current (node B is positive towards speaker/ground), the current in R5 is flowing from B to A, not from A to B.
If you see the "Alexander Current Feedback Amplifier", you will see this R5 as the resistor which connects the output of opamp to the output node to the amplifier. Does it means the opamp is driving (or partly driving) the speaker? It is not. The output node has voltage more positive/negative than the opamp (driven by +/-15V only, the amp is driven by +/-50V for example).
In case of positive swing, the output node is more positive than the opamp's output, so in positive swing the current actually "get in" to the output node, not "get out". The configuration looks like the opamps output is contributing "a little fraction of output" for the speaker, but infact it is "receiving voltage difference".
The difference of "Alexander Current Feedback Amplifier" and R5 that I use, is that in "Alexander" it is forming a voltage divider (having extra R to ground), in R5 it don't have voltage divider, just define the gain of the error voltage in R5 (error between A-B)by a factor of (100ohm/10ohm) to be seen by T6/7.
Thanks for simulating. R18 shouldn't be used if your Q5-Q6 wants to operate as EC or voltage follower. If we use R18, the whole thing will become output-stage with voltage gain, like QSC uses, with opamp driver (+/-15V) can make output stage of +/-70V
In your simulation, the gain will be far less than +1, because the ratio of the value R7/R17 (or R8/R17) is only 1/3. The local EC-CFP of Q5-Q6 in your simulation only senses 1/3 of the error happened between A-B (in R17 from your sch).
If you put R17=10ohm, then the error voltage magnification between A-B will be amplified by factor of 10 (100ohm/10ohm). This way, the whole gain will be closer to +1 (but will never get to +1).
Hi, EVA
Your assumption of 1/70th current is true IF we take the feedback from point B, then R5 will contributing driving the load. But here, the feedback is taken from point A, it is very different mechanism happening.
It is true, that this EC will never get the perfect +1 gain. It will always less than +1, no matter how small it is.
But this is not caused because R5 (Or R17 in PMA's) has to contribute driving the load/output stage (even a very small fraction), but because the little fraction of current has to drive the EC mechanism itself. This is because the whole EC mechanism doesn't have external current source / external transistor, so the energy to drive it has to be drawn from the system itself. In Hawksford it has external current source and external transistor, so the gain can be made <1 or set =1 or made>1 (negative impedance). This EC cannot do that, it cannot even reach the perfect +1gain. (but I think this is a merit, it is harder to oscilate
)Infact the current is happening "Negative way". In positive current (node B is positive towards speaker/ground), the current in R5 is flowing from B to A, not from A to B.
If you see the "Alexander Current Feedback Amplifier", you will see this R5 as the resistor which connects the output of opamp to the output node to the amplifier. Does it means the opamp is driving (or partly driving) the speaker? It is not. The output node has voltage more positive/negative than the opamp (driven by +/-15V only, the amp is driven by +/-50V for example).
In case of positive swing, the output node is more positive than the opamp's output, so in positive swing the current actually "get in" to the output node, not "get out". The configuration looks like the opamps output is contributing "a little fraction of output" for the speaker, but infact it is "receiving voltage difference".
The difference of "Alexander Current Feedback Amplifier" and R5 that I use, is that in "Alexander" it is forming a voltage divider (having extra R to ground), in R5 it don't have voltage divider, just define the gain of the error voltage in R5 (error between A-B)by a factor of (100ohm/10ohm) to be seen by T6/7.
In the view of final gain, it is
. But isn't that Triple Darlington EF cannot achieve perfect +1 gain also?How about looking with another eyeglass.
R5 enables us to
1. Float the feedback node from speaker, so that small errors caused by "back EMF" can be corrected locally by T6-T7, doesn't entering input differential. Maybe you dont agree with this, how about 2nd below?
2. R5 enables us to make the whole amp drive in "current mode" not voltage follower mode (like ordinary amp). There are several examples of "Voltage feedback/Current drive" schematics that sounds good, NP's Aleph is one of them. The other is MauroPenasa' "MyReferenceRev3", sounds good too. F1/F2 is using current drive also, but not using feedback.
Hi, EVA,
This is the Alexander amp http://www.scopeboy.com/original.gif.
The brain of this amp is opamp A1 (SSM opamp). It has local feedback divider towards ground (16.8+33.2ohm?not so clear), and it's output node is connected to amp's output node via 750ohm/2watt/noninductive resistor.
Because the opamp's output is connected to amp's output via 750ohm resistor, does this mean the opamp is partially driving the speaker? Not at all. 750ohm here is giving information back from the output node to the opamp. The reference is in SSM's output node. The voltage of (1+[16.8/33.2]) should fit (1+[750/(16.8+33.2)], otherwise the opamp will take action via it's supply node.
In positive swing, the current is actually moving towards the opamp's output, not "from" the opamp's output to the amp's output node, in a sense that the opamp is partially driving the output node.
Base currents of output devices (differential) will always flow through R5, no matter where you take global feedback. There is no other path they could follow.
And T3 will have to source base currents for T4 and T5, that assuming a gain of 100 would be 1/100th of the current flowing through R5, that is 1/7000th of the output current (assuming beta=70 for output devices).
This actually means that 1/70th of the speaker current is sourced by T4 or T5, and 1/7000th of speaker current comes from T3 (by the way, that's how impedances are estimated, by tracing back base currents).
Sorry if it sounds weird, but this is how bipolar transistors work.
PD: your link for Alexander amp appears to be broken, but I think it has nothing to do with your circuit.
And T3 will have to source base currents for T4 and T5, that assuming a gain of 100 would be 1/100th of the current flowing through R5, that is 1/7000th of the output current (assuming beta=70 for output devices).
This actually means that 1/70th of the speaker current is sourced by T4 or T5, and 1/7000th of speaker current comes from T3 (by the way, that's how impedances are estimated, by tracing back base currents).
Sorry if it sounds weird, but this is how bipolar transistors work.
PD: your link for Alexander amp appears to be broken, but I think it has nothing to do with your circuit.
Lumanauw,
I know how my circuit works. The one linked in http://www.diyaudio.com/forums/showthread.php?postid=700078#post700078 has been in operation for 3 years and for the other showed in http://www.diyaudio.com/forums/showthread.php?postid=700155#post700155 I have done a simulation with and without R18, with R17 as a parameter stepped from 0 to 600 Ohms by step of 30 Ohms.
I know how my circuit works. The one linked in http://www.diyaudio.com/forums/showthread.php?postid=700078#post700078 has been in operation for 3 years and for the other showed in http://www.diyaudio.com/forums/showthread.php?postid=700155#post700155 I have done a simulation with and without R18, with R17 as a parameter stepped from 0 to 600 Ohms by step of 30 Ohms.
lumanauw said:
hi David,
Since the feedback from the output node is taken back to voltage divider and then fed to the output of the opamp..
Now lets examine whats comprising the output of an opamp...The answer is simply the Emitter followers at the output in the opamps or should we say the output of an opamp is formed of emmiters of both transistor in it...then it clearly perspectigates thats the current node summing is well within the domain....
regards,
K a n w a r
Hi, EVA,
Sorry for the broken link. Here I attach Alexander current feedback amp.
Directly, no, I just want to show you what happened in R5. R5 is in the same position as 750ohm in Alexander, but without the voltage divider. Please look at around A1 (SSM2131?). It is connected from output of the opamp to the output of the amp via 750ohm resistor. Do you think the opamp "contribute" driving the loudspeaker, because there is 750ohm resistor connected to the speaker?
Make the case positive swing. The output node (that is connected to the speaker) maybe has +30V (relative to ground).
At the same time, the output node of the opamp maybe only 1V. How come there is a current flowing from 1V to 30V?
So 750ohm here is not a path for the opamp to "contribute" driving the load.
It works in reversed current flow. The opamp actually "receives" current instead of "contributing" to output current, in case of positive swing.
The opamp's own output will be VA=1+(16.8/33.2).
The output node's voltage divider will be VB=1+(750/[16.8+33.2]).
There are 2 voltages, one is made by the opamp's output, and one is made by output node's voltage divider.
Since these 2 voltages is happening in the same point (the opamp's output node), the opamp will do anything to make the 2 voltages the same value (VA=VB). If they are not the same value, the opamp will react via it's voltage terminals.
In case of R5, because there is no voltage divider (750ohm towards 16.8+33.2), T6 and T7 in post#1 will always try to make VA=VB in R5, without magnification (because there is no voltage divider).
Can you see that 750ohm here is just a "voltage sensor"? It is not a path for driving the load.
Hi, Kanwar,
I'm not quite understanding you. You are talking about Alexander amp (which has opamp)? Another explenation, maybe
Sorry for the broken link. Here I attach Alexander current feedback amp.
Directly, no, I just want to show you what happened in R5. R5 is in the same position as 750ohm in Alexander, but without the voltage divider. Please look at around A1 (SSM2131?). It is connected from output of the opamp to the output of the amp via 750ohm resistor. Do you think the opamp "contribute" driving the loudspeaker, because there is 750ohm resistor connected to the speaker?
Make the case positive swing. The output node (that is connected to the speaker) maybe has +30V (relative to ground).
At the same time, the output node of the opamp maybe only 1V. How come there is a current flowing from 1V to 30V?
So 750ohm here is not a path for the opamp to "contribute" driving the load.
It works in reversed current flow. The opamp actually "receives" current instead of "contributing" to output current, in case of positive swing.
The opamp's own output will be VA=1+(16.8/33.2).
The output node's voltage divider will be VB=1+(750/[16.8+33.2]).
There are 2 voltages, one is made by the opamp's output, and one is made by output node's voltage divider.
Since these 2 voltages is happening in the same point (the opamp's output node), the opamp will do anything to make the 2 voltages the same value (VA=VB). If they are not the same value, the opamp will react via it's voltage terminals.
In case of R5, because there is no voltage divider (750ohm towards 16.8+33.2), T6 and T7 in post#1 will always try to make VA=VB in R5, without magnification (because there is no voltage divider).
Can you see that 750ohm here is just a "voltage sensor"? It is not a path for driving the load.
Hi, Kanwar,
I'm not quite understanding you. You are talking about Alexander amp (which has opamp)? Another explenation, maybe
Attachments
Oh, that schematic has erotic content
The the op-amp tries to track the input voltage but the voltage divider connected to its output tries to track the output voltage.
Any imbalance forces a current to flow through the output of the op-amp, that is further forced to flow through its supply rails, throught the current mirrors and through the bases of the predrivers (Q8 and Q9?).
This means that the error current flowing through the output of the op-amp is multiplied by 22500 and forced to flow through the gates of the output devices (assuming a gain of 150 for predrivers and drivers).
However, that error current stops flowing when the gates are charged to the right value. Thus, error current flowing through the output of the op-amp is negligible at DC and audio frequencies, so it isn't able to produce any distortion or output impedance degradation.
In comparison, your error current is roughly 1/70th of speaker current, it's huge thus producing lots of distortion and degrading output impedance.
Note that Alexander circuit has an outrageous open loop current gain (almost infinite at DC) while current gain of your output devices is pretty low. That makes a strong difference.
The the op-amp tries to track the input voltage but the voltage divider connected to its output tries to track the output voltage.
Any imbalance forces a current to flow through the output of the op-amp, that is further forced to flow through its supply rails, throught the current mirrors and through the bases of the predrivers (Q8 and Q9?).
This means that the error current flowing through the output of the op-amp is multiplied by 22500 and forced to flow through the gates of the output devices (assuming a gain of 150 for predrivers and drivers).
However, that error current stops flowing when the gates are charged to the right value. Thus, error current flowing through the output of the op-amp is negligible at DC and audio frequencies, so it isn't able to produce any distortion or output impedance degradation.
In comparison, your error current is roughly 1/70th of speaker current, it's huge thus producing lots of distortion and degrading output impedance.
Note that Alexander circuit has an outrageous open loop current gain (almost infinite at DC) while current gain of your output devices is pretty low. That makes a strong difference.
amplifierguru:
The input stage of any amplifier is nothing but a discrete op-amp, and there are op-amps that will do that job the same way or better.
I admit that the concept of using current consumption from an op-amp as a drive signal for an output stage may appear to be conceptually very complex to untrained minds, but it's a great idea anyway. It's very clever since current has no other path to flow.
The input stage of any amplifier is nothing but a discrete op-amp, and there are op-amps that will do that job the same way or better.
I admit that the concept of using current consumption from an op-amp as a drive signal for an output stage may appear to be conceptually very complex to untrained minds, but it's a great idea anyway. It's very clever since current has no other path to flow.
amplifierguru said:
Hi Greg,
perspectigates = senstive interactivity
Yes i do know Continental Device India limited
This company makes both 556/546 in A', B' C' versions and there products are well available in India at New Delhi....
kindly visit www.CDIL.com
regards,
K a n w a r
Hi, EVA,
I feel exhausted like having to run 10km, only to explain what happened in R5 / current drive mode.
And finally, you give me this :
I think you're right about not enough current gain (current drive) compared to alexander amp.
How about parrareling output stages, maybe use 5 or 7 pcs of C5200 or C2922 kind of transistor (per-rail) as the current generator, will it be better?
So, what is your conclusion, sch in no.1 is not good enough?
If I use sch in post#1 just as it is, not adding more transistors, what will I get and what will I loose in audio power amp reproduction, compared to standard 3 stages amp for example.
Which part is erotic?
I feel exhausted like having to run 10km, only to explain what happened in R5 / current drive mode.
And finally, you give me this :
Please EVA, mercy on me.........No, just kidding
I think you're right about not enough current gain (current drive) compared to alexander amp.
How about parrareling output stages, maybe use 5 or 7 pcs of C5200 or C2922 kind of transistor (per-rail) as the current generator, will it be better?
So, what is your conclusion, sch in no.1 is not good enough?
If I use sch in post#1 just as it is, not adding more transistors, what will I get and what will I loose in audio power amp reproduction, compared to standard 3 stages amp for example.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.