John Curl's Blowtorch preamplifier

[Justcallmedad]

John if we tie the servo op amp output TO A BIG R but not to the middle point between the fets, but twice to the fets sources, the op amp disturbance could be less "annoying"?

[Justcallmedad]

Originally posted by Upupa Epops
He cut eis, Richard...listen closely ...

eis? what is this?

[Justcallmedad]

Ice !!! Upopa ! I can't listen now, to late here, but I think you are right remembering the song Have you a nice Q sound?

[Justcallmedad]

Originally posted by john curl
None of these circuits is optimum. Circuit one does not have enough attenuation to remove the servo imperfections. The other servos should do even more change to the sound.

Yes I am agree now with you John, it was a few years ago...
In the fig 3 there is only I servo and one virtual ground.

What I would like to test more in depht is the virtual ground, have you ever tested something like?

[Terry Demol]

Originally posted by john curl
None of these circuits is optimum. Circuit one does not have enough attenuation to remove the servo imperfections. The other servos should do even more change to the sound.

Hi John,

I'm really enjoying this thread having done a lot on servos myself.

I have a question related to implementing a 'gain reduction' or
some kind of attenuator at the OP of the servo, as you have
suggested, to reduce any possible servo induced artifacts.

Let's say we have an open loop circuit with a load (OP) R of 1k
with a servo OP fed into a virtual gnd or fairly low impedance
current IP point.

Essentially the servo is inputting a correction current that will
manifest at the OP R as a voltage. The ratio of the OP R on the
servo opamp to the circuit load R (1K) will determine the servo
swing for a given circuit OP swing.

So is there any advantage in implementing a gain reduction
network, say, referenced to ground at the servo output over just
increasing the resistor size on the OP of the servo to result in the
same correction ratio.

At the end of the day is there not just a ration of servo swing
compared to circuit OP swing and the optimum is to make this
ratio as high as possible to attenuate servo artifacts.

Cheers,

Terry

[john curl]

Terry, I can't judge specifically which circuits will work from your description, but I will say this: It is important to ACTUALLY attenuate the servo output so that the servo must work many times more in voltage for the correction it is trying to implement. For example, a 20mV input offset might take 2V of swing on the servo, and the servo will sit at 2V out most of the time. This way, the servo is not adding itself to the input signal in any significant way. This is because the servo IC's and caps are usually inferior to the actual circuit in their transfer function, so you don't desire that the IC servo's sound is added to the audio passing through.
I completely understand that some of you, like Elso, have had bad experiences with servos. I would not use servos either, if they were not necessary in my designs. Please remember that I make components for others, who are sometimes very sensitive to a click or pop to the point of neurosis. I just can't afford to allow my absolute polarity switch to click or pop just because I felt that 10 mV of offset was OK.

[Justcallmedad]

Originally posted by john curl
It is important to ACTUALLY attenuate the servo output so that the servo must work many times more in voltage for the correction it is trying to implement...

Very clear indeed.

Forget one moment the dc servo. and consider simply the virtual ground for the input stage (gate and source reference) is it better because we avoid ground currents as the op amp is very close to the audio input stage and the current at the virtual ground op amp is negligible.

[Terry Demol]

Originally posted by Justcallmedad


Very clear indeed.

Forget one moment the dc servo. and consider simply the virtual ground for the input stage (gate and source reference) is it better because we avoid ground currents as the op amp is very close to the audio input stage and the current at the virtual ground op amp is negligible.

Hi Guys,

At first it is clear... but is it? You need to think this through.

I will clarify a few points;

- Assume the servo feeds *current* into a part of the cirdcuit.
It does not matter where, at this point, only the fact that the servo
is injecting a correction current into the circuit.

- Assume load R (at OP) of the circuit is 1k.

- Assume the servo has a 100k resistor at it's OP.

- The servo will swing 2V to effect a 20mV offset at the circuit OP

- Let us introduce a 10:1 attenuation at the servo OP. Say 10k
series, 1k to fround.

- To get the same 2V servo swing for 20mV audio OP correction we
then will need a 10k feed R *after* the 10:1 servo attenuator.

Have we actually achieved any thing with the servo OP attenuator of 10:1 if the servo OP swing to audio circuit correction swing is
the same (in this example 100:1)?

I have also implemented attenuator circuits at sservo OP's to
supposedly minimise their effect. But after careful circuit analysis
I have had to ask myself -exactly what was acheived?-

It depends on how you look at it.

Cheers,

Terry

[Justcallmedad]

1) My question only concerned the virtual ground properties, and my question is more to know if technically speaking it’s a must but I don’t think, or it’s just to avoid the ground currents and ground loops in a real PCB implantation.

2) The purpose of the DC servo is to compensate the output offset, so whatever you do (a resistor or an attenuator) after the DC Servo you always will get the same current or voltage at the offset compensation point (we suppose that op amp swing is not limited )
The purpose of the resistor is to "move away", isolate, the op amp and the capacitor from the audio circuit.

In short it is what I understood of John’s lesson.

[Justcallmedad]

Good morning John… Teaching time… …

Do you really use an LM317/337 as current source on the Blowtorch PSU as suggested by Darry, (if I remember well?) If this is true why this choice over a single BJT?