I have learn audio power amp with 3 stages, that is Differential, VAS, Current stage. This is the most common configuration, common people see it as minimal, as each stage has its purpose.
Clever thinking can lead to 2 stages, like Aleph or Folded cascode.
From what I learn from Mr.Pass, fewer gain stage will lead to more accurate reproduction, since in every stage (transistor) will introduce phase shift and time delay. In extreme example, many stages power amp wont have the differential produce the right error signal, since what happens in the final stage is delayed too much than what is received in differential. He call this "Hall of Mirror Effect" which makes sense.
Mr. Pass has also made single stage power amp, like ZEN and its variations. But these amps can operate only in class A/single ended which is very hot and efficiency is poor (but the sound is good).
I just think about a single stage power amp. Single stage, because the input and output is just from 1 stage transistor. But this configuration can be operated in class B, AB, A, therefor have more efficiency.
The power supply is 2 unit. The first is +/-15V, with its ground is the overall ground. The second unit is the big voltage (any voltage), and floating.
Look at the schematic. Since the big supply is floating, it can wonders all around. But this PS is stiff. The distance between +,CT,- is always steady.
We only have to modulate T3 and T4 about +/-0.6V to make it active. Therefor, the +/-15V supply can be reduced (Maybe +/-5V is enough, but I put +/-15V so it can drive accesories opamps)
The signal comes to a buffer (to higher input impedance) and this buffer has RG, to set the 0V DC operating point. The gain is -RF/Rin which is inverting amp gain.
The signal is injected to point P, between emitors of bias/Vbe multiplier, which can be adjusted with VR.
The T1 and T2 is small transistor, its purpose is to set the bias for T3 and T4. Since this T1 and T2 is put between 2 CCS (put between collectors), it is floating, can wonder around. The CCS is about 2-3mA, just to bias T1 and T2. C is to help stabilize this bias voltage.
If the input signal is positive, it will make T3 on, and T3 will pull + to ground. Therefor the CT will be more negative than ground (the speaker is put between CT and ground).
In other way, if signal is negative, it will make T4 on, and T4 will pull - to ground. The speaker will experience negative, relative to ground.
The configuration of 2 CCS + T1 + T2 is a way to use bipolars or MOSFETS as T3 and T4, where they are to set bias point for T3 and T4.
If we can get big enough FETs (with negative Vgs), the CCS and T1+T2 wont be necessary, the signal input can be directed to gate of fets itself (if this kind of power FETs exist).
Because it is only 1 stage transistor (T3+T4), is this Single Stage power amp? It merits also in bias, can be operated from class B to class A. It also can swing fully, to voltage rail set by PS.
What is the open loop gain of this configuration?
Clever thinking can lead to 2 stages, like Aleph or Folded cascode.
From what I learn from Mr.Pass, fewer gain stage will lead to more accurate reproduction, since in every stage (transistor) will introduce phase shift and time delay. In extreme example, many stages power amp wont have the differential produce the right error signal, since what happens in the final stage is delayed too much than what is received in differential. He call this "Hall of Mirror Effect" which makes sense.
Mr. Pass has also made single stage power amp, like ZEN and its variations. But these amps can operate only in class A/single ended which is very hot and efficiency is poor (but the sound is good).
I just think about a single stage power amp. Single stage, because the input and output is just from 1 stage transistor. But this configuration can be operated in class B, AB, A, therefor have more efficiency.
The power supply is 2 unit. The first is +/-15V, with its ground is the overall ground. The second unit is the big voltage (any voltage), and floating.
Look at the schematic. Since the big supply is floating, it can wonders all around. But this PS is stiff. The distance between +,CT,- is always steady.
We only have to modulate T3 and T4 about +/-0.6V to make it active. Therefor, the +/-15V supply can be reduced (Maybe +/-5V is enough, but I put +/-15V so it can drive accesories opamps)
The signal comes to a buffer (to higher input impedance) and this buffer has RG, to set the 0V DC operating point. The gain is -RF/Rin which is inverting amp gain.
The signal is injected to point P, between emitors of bias/Vbe multiplier, which can be adjusted with VR.
The T1 and T2 is small transistor, its purpose is to set the bias for T3 and T4. Since this T1 and T2 is put between 2 CCS (put between collectors), it is floating, can wonder around. The CCS is about 2-3mA, just to bias T1 and T2. C is to help stabilize this bias voltage.
If the input signal is positive, it will make T3 on, and T3 will pull + to ground. Therefor the CT will be more negative than ground (the speaker is put between CT and ground).
In other way, if signal is negative, it will make T4 on, and T4 will pull - to ground. The speaker will experience negative, relative to ground.
The configuration of 2 CCS + T1 + T2 is a way to use bipolars or MOSFETS as T3 and T4, where they are to set bias point for T3 and T4.
If we can get big enough FETs (with negative Vgs), the CCS and T1+T2 wont be necessary, the signal input can be directed to gate of fets itself (if this kind of power FETs exist).
Because it is only 1 stage transistor (T3+T4), is this Single Stage power amp? It merits also in bias, can be operated from class B to class A. It also can swing fully, to voltage rail set by PS.
What is the open loop gain of this configuration?
Attachments
Been there, done that.
http://www.diyvideo.com/forums/attachment.php?postid=182299
This is the Transnova patent.
QSC, Crown, have done it for years with bipolars.
http://www.diyvideo.com/forums/attachment.php?postid=182299
This is the Transnova patent.
QSC, Crown, have done it for years with bipolars.
Hi, DJK,
You are right. The transformer CT output has been used by QSC for a long long time.
What I want to emphasis here is the single stage power amp, which can swing to any voltage, with any bias. This will reduce the time delay and phase shift.
In the patent you pointed, there is an amplifier/opamp in the path of differential-feedback. Wouldn't this mean the patent is having more than 1 stage (at least inside the opamp)
I have question about my own drawings. What is the open loop gain of my cct?
You are right. The transformer CT output has been used by QSC for a long long time.
What I want to emphasis here is the single stage power amp, which can swing to any voltage, with any bias. This will reduce the time delay and phase shift.
In the patent you pointed, there is an amplifier/opamp in the path of differential-feedback. Wouldn't this mean the patent is having more than 1 stage (at least inside the opamp)
I have question about my own drawings. What is the open loop gain of my cct?
You are better off to use a CFP like QSC does.
In addition to the cases of the outputs being at ground potential and improved thermal performance the emitter resistors shift to the power supply side.
The open loop gain is greatly reduced by the emitter resistor when the emitters are grounded as per your drawing.
If you used a pair of MJE15030/31 as outputs for a low powered amplifier with no emitter resistors the open loop gain would be about 160 for a 16 ohm load, 80 for an 8 ohm load. A 0R1 emitter resistor in the ground side would require twice as much input voltage for the same output current.
In addition to the cases of the outputs being at ground potential and improved thermal performance the emitter resistors shift to the power supply side.
The open loop gain is greatly reduced by the emitter resistor when the emitters are grounded as per your drawing.
If you used a pair of MJE15030/31 as outputs for a low powered amplifier with no emitter resistors the open loop gain would be about 160 for a 16 ohm load, 80 for an 8 ohm load. A 0R1 emitter resistor in the ground side would require twice as much input voltage for the same output current.
Hi, Djk,
Is this 160 to 16 ohm and 80 to 8ohm considered big or small? Is it possible to make low openloop gain / low feedback power amp with this configuration?
Is this 160 to 16 ohm and 80 to 8ohm considered big or small? Is it possible to make low openloop gain / low feedback power amp with this configuration?
Opamps with two voltage gain stages run about 50,000X voltage gain open loop, so 80 is quite low.
Your buffer would need about 100mA to drive the outputs to about 15W/8R.
Your buffer would need about 100mA to drive the outputs to about 15W/8R.
Hi, DJK,
I'm not quite understand what is mean by Low openloop gain and low feedback.
A power amp is said to have low feedback if it only have 25dB? feedback? If we want the gain 10X that means the open loop is about 100X? So the feedback is about 10X?
I'm not quite understand what is mean by Low openloop gain and low feedback.
A power amp is said to have low feedback if it only have 25dB? feedback? If we want the gain 10X that means the open loop is about 100X? So the feedback is about 10X?
I've done this circuit as a single stage, and considered using
it as part of the Zen Variations, which I suppose I still might.
The performance was very similar to the ZV5, but didn't require
as clean a power supply. On the other hand, I found that I
really only got acceptable performance in Class A mode - the
distortion was simply too high at lower bias (of course that
was with Mosfets - with BJT's the input impedance is very,
very low) 😎
it as part of the Zen Variations, which I suppose I still might.
The performance was very similar to the ZV5, but didn't require
as clean a power supply. On the other hand, I found that I
really only got acceptable performance in Class A mode - the
distortion was simply too high at lower bias (of course that
was with Mosfets - with BJT's the input impedance is very,
very low) 😎
Hi, Mr. Pass,
why is it that I always get answer from you in this forum, instead from direct mail to nelson@passlabs.com?
Really looking forward what this topology can do in the hand of the master. Zen ver 8 maybe? When will this be released?
If you take your sketch and change the BJT to FETs this is essentially what you end up with:
http://www.diyvideo.com/forums/showthread.php?threadid=15919&highlight=
http://www.diyvideo.com/forums/showthread.php?threadid=15919&highlight=
- Status
- Not open for further replies.