M. Curl :
I design this DC Servo on the base of your description:
“I have 2 separate servos in each line amp, because it has differential outputs. One servo tries to keep the average of the two outputs near ground, and the other compares each output and adjusts them to be close to each other. The servos must be very low in frequency cut-off in order not to effect any asymmetrical low frequency transient that might occur on program material.”
However running simulations tests vs. traditional DC Servos, the differences don’t seem really significant within the audio bandwidth…, but beyond…
Differential Servo
Traditional Servo
Graphs represent the rejection at the output of a symmetrical complementary balanced line amplifier of AC source at the inputs of the DC servos.
It’s obvious that a mismatch in the servo values considerably lowers rejection.
I design this DC Servo on the base of your description:
“I have 2 separate servos in each line amp, because it has differential outputs. One servo tries to keep the average of the two outputs near ground, and the other compares each output and adjusts them to be close to each other. The servos must be very low in frequency cut-off in order not to effect any asymmetrical low frequency transient that might occur on program material.”
However running simulations tests vs. traditional DC Servos, the differences don’t seem really significant within the audio bandwidth…, but beyond…
Differential Servo
Traditional Servo
Graphs represent the rejection at the output of a symmetrical complementary balanced line amplifier of AC source at the inputs of the DC servos.
It’s obvious that a mismatch in the servo values considerably lowers rejection.
This completely misses the point. Current output amps are standard for DC motor control, but most loudspeakers are designed to be driven by voltage, not current. Emitter resistors are not a big problem, if they are small valued enough.
The biggest problem is output gain. If you have only 2 stages in your amp, then you may not have enough gain to make the amp practical, because the output stage can have negative gain with a low Z load.
If you use negative feedback, then you can develop IIM distortion, where the output from the loudspeaker feeds back to the input and causes overload.
The biggest problem is output gain. If you have only 2 stages in your amp, then you may not have enough gain to make the amp practical, because the output stage can have negative gain with a low Z load.
If you use negative feedback, then you can develop IIM distortion, where the output from the loudspeaker feeds back to the input and causes overload.
Sorry John, 
I must say that my last post was rather confused, very short and badly explained! I agree !
I just tried to understand the benefits of your DC Servo compound over a classical op amp integrator for each side of a balanced line preamp. In my mind if both sides are near ground, inevitably the two balanced output are close one to the other. This is for the DC point of view… But… If you adopted a more sophisticated configuration it’s for sure that it’s best suited… And I guessed that for a balanced mode preamp output, your design could be very interesting for an AC point of view. So, I went further to see the influence at the line preamp outputs of your DC servo compound vs. a classic one. The goal was to compare for both servos designs the rejection of an AC signal source at the DC servos inputs, measured at the line amp outputs, for single end and for differential outputs.
Yes! I didn’t explain and show it very well on my last post!
Here the graphs of these 2 simulations:
In red and blue the rejection ratio at the output of the line amplifier referred to ground (single end), in green the rejection ratio at the output in balanced mode, this for each type of DC servo (Differential or Dual Classic Integrators).
Dual integrator.
Differential Integrator.
I hope it's clearer.
I must say that my last post was rather confused, very short and badly explained! I agree ! I just tried to understand the benefits of your DC Servo compound over a classical op amp integrator for each side of a balanced line preamp. In my mind if both sides are near ground, inevitably the two balanced output are close one to the other. This is for the DC point of view… But… If you adopted a more sophisticated configuration it’s for sure that it’s best suited… And I guessed that for a balanced mode preamp output, your design could be very interesting for an AC point of view. So, I went further to see the influence at the line preamp outputs of your DC servo compound vs. a classic one. The goal was to compare for both servos designs the rejection of an AC signal source at the DC servos inputs, measured at the line amp outputs, for single end and for differential outputs.
Yes! I didn’t explain and show it very well on my last post!
Here the graphs of these 2 simulations:
In red and blue the rejection ratio at the output of the line amplifier referred to ground (single end), in green the rejection ratio at the output in balanced mode, this for each type of DC servo (Differential or Dual Classic Integrators).
Dual integrator.
Differential Integrator.
I hope it's clearer.
...Dad, I think that you are making more of this than I ever had thought to do. First, when you have a differential output, you need 2 separate servos in order to get the outputs to not have any DC offset.
This is because: One servo sums the two DC outputs and references them to ground. The other servo differences the two outputs and nulls the difference between their DC outputs.
Sorry that this is an incomplete explanation, but it is off the cuff.
Let me try it another way:
IF only one servo is in operation, the 2 outputs will either be averaged around ground (one is slightly positive, the other is slightly negative) or both outputs will be either above ground, or below ground, but they will not have any DC difference between each other.
With both servos in operation, both outputs will be at 0V offset, referenced to ground.
This is because: One servo sums the two DC outputs and references them to ground. The other servo differences the two outputs and nulls the difference between their DC outputs.
Sorry that this is an incomplete explanation, but it is off the cuff.
Let me try it another way:
IF only one servo is in operation, the 2 outputs will either be averaged around ground (one is slightly positive, the other is slightly negative) or both outputs will be either above ground, or below ground, but they will not have any DC difference between each other.
With both servos in operation, both outputs will be at 0V offset, referenced to ground.
Hi John,
Yes I have understand the idea, have you seen my schematic on post 369? I think that it does the job?
I was mistaken, I thought that your DC servo compound was also designed to avoid the drawbacks of a DC servo on sound quality?
So your compound is just intended for DC offset considerations (balanced and single ended)?
Thank you for your patience!
Yes I have understand the idea, have you seen my schematic on post 369? I think that it does the job?
I was mistaken, I thought that your DC servo compound was also designed to avoid the drawbacks of a DC servo on sound quality?
So your compound is just intended for DC offset considerations (balanced and single ended)?
Thank you for your patience!
Not clear again !!! Sorry! Sorry! Sorry!
Outputs (+/-) are the connections at the balance line preamp outputs to sense the offsets.
Servo (+/-) are the DC servo outputs connected at the suitable place on the preamp circuit to correct the offset.
The lower schematic is the classical configuration for a DC servo, I use 2 like these for the sim's.
Clearer?
Outputs (+/-) are the connections at the balance line preamp outputs to sense the offsets.
Servo (+/-) are the DC servo outputs connected at the suitable place on the preamp circuit to correct the offset.
The lower schematic is the classical configuration for a DC servo, I use 2 like these for the sim's.
Clearer?
Yes, John, I notice this a few years ago in doing tests with a small power amp where the DC servo was tied by a resistor divider to the fet sources of the input complementary fets and the sound was much better when I increase the resistor value at the output of the op amp (R2 was pretty small 500R) I choose to quite limit the servo op-amp AC/DC gain ! and setting the amplifier gain by these 2 resistors, (very bad Idea)!!!
I thought at that time that the problem came from mixing 2 references, the ground and the DC servo output, at this node.
Of course due to the DC servo Op-Amp AC/DC gain voluntary limited and the increasing of R2, the offset correction was not very useful !
About jfets, I read somewhere, probably in the forum, and coming from you I think, that the fet’s you use were selected in noise, if this is the
case how do you achieve this measure?
I thought at that time that the problem came from mixing 2 references, the ground and the DC servo output, at this node.
An externally hosted image should be here but it was not working when we last tested it.
Of course due to the DC servo Op-Amp AC/DC gain voluntary limited and the increasing of R2, the offset correction was not very useful !
About jfets, I read somewhere, probably in the forum, and coming from you I think, that the fet’s you use were selected in noise, if this is the
case how do you achieve this measure?
Yes, R2 should be a much higher value.
The fets are difficult to measure for noise, unless you are a professional and make your own test rig to with adjacent equipment, or you have a Quantech Noise Analyser. However, Toshiba makes pretty accurate noise specs, so just get the data sheets for the 2sk170, and the 2sj74 on the internet.
The fets are difficult to measure for noise, unless you are a professional and make your own test rig to with adjacent equipment, or you have a Quantech Noise Analyser. However, Toshiba makes pretty accurate noise specs, so just get the data sheets for the 2sk170, and the 2sj74 on the internet.
"Toshiba makes pretty accurate noise specs" Do you mean that it's not useful to noise select them, as well as for the 2sk240/2sj103 ?
I am not a professional in audio design, but I am sound manager in a broadcast studio so I have some test material like Audio precision portable 1, a Tektronix AM700 and an old 250 Mhz scope, so If I could make a test rig for this purpose even for others fet brands, some advises are welcome
I am not a professional in audio design, but I am sound manager in a broadcast studio so I have some test material like Audio precision portable 1, a Tektronix AM700 and an old 250 Mhz scope, so If I could make a test rig for this purpose even for others fet brands, some advises are welcome
- Status
- Not open for further replies.