Looking forward for the (more later) info to come.
I had enourmous trouble "voicing" my Riaa preamp because it depends so strongly upon caps behaviour... I Experimented with many and now I am satisfyed with my present "cooking" but it is far from perfect. I could not use expensive caps so I had to try severall bypass combinations until I got pleasant results.
Without a major research facility at your disposal you could never get into the realm of the extreme.
Even two #40 wires just touching with 1uA in them involves millions of atoms and ~10^12 electrons per second. Gold wire stretched until there is a single atom left will demonstrate the quantized Hall effect as it breaks, this would be a good place to start.
Last edited:
Use of DC servos
As John said, the use of DC servos is a preferred technique for managing DC offset and avoiding the use of DC blocking capacitors that may degrade sound quality. In fact, I have an entire chapter on the use of DC servos in my Designing Audio Power Amplifiers book. Although treated there in the context of power amplifiers, most of the material applies equally well to the use of DC servos in phono preamps. There are some design subtleties that are important in applying DC servos that need to be understood.
Servo is a feedback loop and all of its components are in the signal path. This is especially true of the op amp(s) and the integrating capacitor(s). The op amp should be of audio grade, as it can inject noise and distortion into the signal path. Similarly, the integrating capacitor should be of high quality, polypropylene at minimum. Fortunately, the integrating capacitors used in DC servos do not usually have to be very large, so cost consideration of using high-quality capacitors are mitigated a bit.
Never use a DC servo that employs a non-inverting integrator. Unfortunately, this approach is popular with some designers. This circuit configures the integrator as a differential amplifier, using two integrating capacitors. In most circuits the use of the non-inverting integrator saves an op amp at the expense of an additional integrating capacitor. This is a mis-guided and poor trade-off in terms of both board area and cost. Worse yet, the common mode rejection of the non-inverting integrator depends on precise matching of the two capacitors, which is usually not available. Poor common mode rejection in the differential integrator can alter the servo frequency characteristics and allow unintended amounts of output back to the input stage.
Cheers,
Bob
As John said, the use of DC servos is a preferred technique for managing DC offset and avoiding the use of DC blocking capacitors that may degrade sound quality. In fact, I have an entire chapter on the use of DC servos in my Designing Audio Power Amplifiers book. Although treated there in the context of power amplifiers, most of the material applies equally well to the use of DC servos in phono preamps. There are some design subtleties that are important in applying DC servos that need to be understood.
Servo is a feedback loop and all of its components are in the signal path. This is especially true of the op amp(s) and the integrating capacitor(s). The op amp should be of audio grade, as it can inject noise and distortion into the signal path. Similarly, the integrating capacitor should be of high quality, polypropylene at minimum. Fortunately, the integrating capacitors used in DC servos do not usually have to be very large, so cost consideration of using high-quality capacitors are mitigated a bit.
Never use a DC servo that employs a non-inverting integrator. Unfortunately, this approach is popular with some designers. This circuit configures the integrator as a differential amplifier, using two integrating capacitors. In most circuits the use of the non-inverting integrator saves an op amp at the expense of an additional integrating capacitor. This is a mis-guided and poor trade-off in terms of both board area and cost. Worse yet, the common mode rejection of the non-inverting integrator depends on precise matching of the two capacitors, which is usually not available. Poor common mode rejection in the differential integrator can alter the servo frequency characteristics and allow unintended amounts of output back to the input stage.
Cheers,
Bob
The use of servos and +/- supplies allows coupling cap free design. This has an extra benefit of reducing the number of IN SERIES connections in the audio path. The 'problems' are put in a parallel path and can be minimized or avoided, instead.
It must be said that a capacitor is not just the ideal device represented by a symbol on a schematic. It has leads (sometimes magnetic) dissimilar metal junctions (sometimes lousy) and it has an imperfect dielectric material, in virtually every case, except vacuum.
Care must be taken with the servos, however, and they should be DECOUPLED as much as possible from the audio itself.
It must be said that a capacitor is not just the ideal device represented by a symbol on a schematic. It has leads (sometimes magnetic) dissimilar metal junctions (sometimes lousy) and it has an imperfect dielectric material, in virtually every case, except vacuum.
Care must be taken with the servos, however, and they should be DECOUPLED as much as possible from the audio itself.
Put that in big letters. With some attention, the requirements that Bob rightly brought up can be minimized- a properly implemented servo can get by with moderately good opamps and caps, nothing fancy needed.
In power amplifier input pairs I place power resistors next to the devices. These allows me thermal control of the offset from the bipolar input transistors! I also use lm34s to set the reference level!
You don't really need power resistors, I could imagine 1/8 Watt ones glued to TO92's maybe even in the current mirror so you could use duals for the input devices.
While we all run around designing circuits that are DC coupled in the interests of the ultimate in high fidelity, guys like Doug Self designing mixers are just as busy slapping 5534's and 5532's down with dozens of electrolytics in the signal path between the mic and the hard disc where the music is stored.
2 or 3 electrolytics in the signal path between your source (CD, phono or whatever) and the power amp will not alter the sound irretrievably for the worse. Depending on your pursuation, that was already taken care of in the recording studio, or it is, as I suspect, of no consequnce in the big scheme of things.
The differences in CD recordings (noise, sound stage, dynamic range, instrument placement, mic set-up etc) are huge and in our discussions on these pages, we seem to completely ignore this.
2 or 3 electrolytics in the signal path between your source (CD, phono or whatever) and the power amp will not alter the sound irretrievably for the worse. Depending on your pursuation, that was already taken care of in the recording studio, or it is, as I suspect, of no consequnce in the big scheme of things.
The differences in CD recordings (noise, sound stage, dynamic range, instrument placement, mic set-up etc) are huge and in our discussions on these pages, we seem to completely ignore this.
Last edited:
Scott,
In amplifiers such as Jan's Pax where current conveyors are extensively used, the mismatch of the positive rail and negative rail sources placed demands on the input stage to keep the output offset close to zero volts. By using thermal excitation the current sources can be tweaked in to match each other.
The current trend in solid state designs is to try and get complimentary parts to match as closely as possible.
So where you have the ability to actively monitor a device's parameters you can make some adjustment.
It is just another tool in the advanced designers tool box.
ES
I am using a modified Meridian 101 preamp that has no series capacitors in the signal path.
As a matter of fact, it is basically a double opamp LME49720 configured to have 16x gain placed after the log pot.
I have no DC offset on the output just by using a carefully built + - 15v PSU using special series regs.
The mods I did on the Meridian resume basically to the opamp swap and the new psu.... I did not alter the stock configuration so it never had any caps in the signal path (I mean I did not invent anything... I am only explaining what I found inside the preamp)
Now I am learning about DC servo and Bob explained (thank you Bob) it must be placed in the signal path... It seems we are trying to avoid caps but adding more items in the signal path.
(I always try to avoid opamps because they rely so heavily in negative feedback and have loads of components only needed for it´s safe functioning)
Can we use DC Servo using only discrete components ?
As a matter of fact, it is basically a double opamp LME49720 configured to have 16x gain placed after the log pot.
I have no DC offset on the output just by using a carefully built + - 15v PSU using special series regs.
The mods I did on the Meridian resume basically to the opamp swap and the new psu.... I did not alter the stock configuration so it never had any caps in the signal path (I mean I did not invent anything... I am only explaining what I found inside the preamp)
Now I am learning about DC servo and Bob explained (thank you Bob) it must be placed in the signal path... It seems we are trying to avoid caps but adding more items in the signal path.
(I always try to avoid opamps because they rely so heavily in negative feedback and have loads of components only needed for it´s safe functioning)
Can we use DC Servo using only discrete components ?
Last edited:
Yes, but you'll not gain any performance advantage. It's pretty easy to set up an IC opamp servo that is effectively out of the way at audio frequencies and requires no fancy components.
The distinction between series and shunt in signal paths as far as claimed audibility is... well... not logical. And as usual, totally unsupported by any evidence.
What do you mean by using an opamp out of the way at audio freq ?
Do you mean the opamp is operating at very high freq (or at dc level) and that can not be heard ?
About the distinction between series and shunt parts... of course all parts leave their own signature but IMO a series cap is restrictive and a shunt cap is aditive. (Soundwise)
Do you mean the opamp is operating at very high freq (or at dc level) and that can not be heard ?
About the distinction between series and shunt parts... of course all parts leave their own signature but IMO a series cap is restrictive and a shunt cap is aditive. (Soundwise)
IMO There are no perfect components... all of them leave or add some signature.
Kirchoff´s law "The algebraic sum of currents in a network of conductors meeting at a point is zero" works as an idea so we can visualize the effects of currents inside a schematic but what I meant was:
series cap is restrictive = these caps somehow limit bandwith and caps are not very linear so freq response is always affected.
for shunt cap (I meant decoupling cap) is aditive = these tend to add bandwith by reducing impedance at certain frequencies but at the signature cost
About the opamps in the signal line.... that might be bad. No two opamps sound the same.
Kirchoff´s law "The algebraic sum of currents in a network of conductors meeting at a point is zero" works as an idea so we can visualize the effects of currents inside a schematic but what I meant was:
series cap is restrictive = these caps somehow limit bandwith and caps are not very linear so freq response is always affected.
for shunt cap (I meant decoupling cap) is aditive = these tend to add bandwith by reducing impedance at certain frequencies but at the signature cost
About the opamps in the signal line.... that might be bad. No two opamps sound the same.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.