Justcallmedad said:
Richard,
I appreciate all the explanation regarding the regs, quite nice setup you have developed and I will be testing some of these ideas as soon I come back from a long trip to S. America.
Your work is really inspirational and few members care to share their work on shunt regs over the forums, I guess maybe, little has been done before.
Would be nice to hear from you when you are able to make some comparisons against other PSU you might have on your gear.
Actually folks, I LOVE remote control. I have remote controls all over the place, but we didn't want any sonic compromise in this preamp.
The latest Parasound JC-2 preamp designed by Bob, Carl, and me (CTC) has remote control, because many people demand it. We use a cheaper TKD pot with a motor control. It, to us, is as close as we can get, to the CTC preamp and still have all the features that most people rely on.
The latest Parasound JC-2 preamp designed by Bob, Carl, and me (CTC) has remote control, because many people demand it. We use a cheaper TKD pot with a motor control. It, to us, is as close as we can get, to the CTC preamp and still have all the features that most people rely on.
John,
My guess is that a servo with a long enough time constant (low F) to get out of the way of the music doesn't provide as close a control as folks want. If the time constant is shorter (higher F) then it encroaches on the music.
Understand that I am not attacking your use of servos. I personally have said many times in many threads that I respect your work and find it of great interest. You are one of the few people here who rates an "of course" as to whether I read your posts. I am simply putting forth a hypothesis. I have used servos before and very likely will again. Depending on how you define 'servo,' I have a circuit idea for these power JFETs I've been messing with that uses one for the bias/DC offset. If I can get the circuit to run the way I want it to without one, I don't use a servo. But the idea is there if I need it.
I personally favor long time constants, but that falls rather naturally out of my desire for very wide bandwidth with as little feedback as possible. I happen to like details, and imaging--just to cite one example--suffers with higher rates of feedback. From this you can correctly deduce that I come from the tube camp, but am trying to find ways to seduce solid state into giving me what I want.
Grey
My guess is that a servo with a long enough time constant (low F) to get out of the way of the music doesn't provide as close a control as folks want. If the time constant is shorter (higher F) then it encroaches on the music.
Understand that I am not attacking your use of servos. I personally have said many times in many threads that I respect your work and find it of great interest. You are one of the few people here who rates an "of course" as to whether I read your posts. I am simply putting forth a hypothesis. I have used servos before and very likely will again. Depending on how you define 'servo,' I have a circuit idea for these power JFETs I've been messing with that uses one for the bias/DC offset. If I can get the circuit to run the way I want it to without one, I don't use a servo. But the idea is there if I need it.
I personally favor long time constants, but that falls rather naturally out of my desire for very wide bandwidth with as little feedback as possible. I happen to like details, and imaging--just to cite one example--suffers with higher rates of feedback. From this you can correctly deduce that I come from the tube camp, but am trying to find ways to seduce solid state into giving me what I want.
Grey
As I already told it in my post #83, configurations used by John Curl for its DC servos don't look like those of the other audio designers except maybe for the Vendetta.
In the schematic of the Dennesen JC-80 and in the schematic of the PLD-2000 Parasound preamplifier, M.Curl uses 2 OP (1 single and 1dual) in an astute and very ingenious configuration not very easy to understand.
The Blowtorch also uses 2 OP. I think that the schematic (folded cascode) of Blowtorch is relatively simple with few components.
On the other hand, I think that its success and its stability of working is due largely to a very particular utilization of DC servos.
Maybe that Mr. Curl will accept to tell us a few more of them.
Darry
In the schematic of the Dennesen JC-80 and in the schematic of the PLD-2000 Parasound preamplifier, M.Curl uses 2 OP (1 single and 1dual) in an astute and very ingenious configuration not very easy to understand.
The Blowtorch also uses 2 OP. I think that the schematic (folded cascode) of Blowtorch is relatively simple with few components.
On the other hand, I think that its success and its stability of working is due largely to a very particular utilization of DC servos.
Maybe that Mr. Curl will accept to tell us a few more of them.
Darry
Switches -v- Relays.
John,
Going back to your comments, yes, I have listened (a great deal!) to these things, and they all affect the sound in some way or other.
However, unlike passsive (and some active) components, in my experience they don't specifically 'colour' the sound. i.e. affect the balance of sound, or accentuate any part of the audible spectrum, in quite the same way.
What I hear is a veiling of everything, in different degrees depending on the contacts' materials and construction of the switch or relay. This veiling or fogging of the signal is quite noticeable if a hard-wired bypass is used as a comparison, and results in a loss of details, dynamics, and sound-stage etc.
If one looks closely at say a Shallco switch, the input wiring is soldered directly to the opposite end of the silver plug which actually makes up one of the contacts, and there is a heavy double 'bridge' of very hard silver, the (4 in total) narrow edges of which then wipe along a circular track which forms the output contact.
The twin edges of this hard sprung bridge being under considerable pressure at each end, will, as you suggest, cut through any sulphides, or whatever, which do tend to develop in normal use in domestic (or commercial) environments, on either the multiple contacts or on the 'ring' of hard silver forming the other contact of the switch.
Apart from the two soldered joints (input & output) which every device would need, anyway, there are merely two other 'discontinuities'. i.e where this wiping 'bridge' makes contact with the inputs & outputs, but everything else is 'solid' metal (silver in this case, which helps!). Also, these two discontinuities are continually being physically 'cleaned' every time the switch is operated.
Compare this with less well-made switches and relays and generally there will be many metal-to-metal junctions in their internal construction, mostly utilising dissimilar metals which are riveted together. All of these joints will suffer over time as a result of (microscopic) corrosion which is always present when dissimilar metals are in intimate contact, especially where heating and cooling takes place in addition, which encourages movement between the parent metals due to their differing thermal coefficients of expansion.
So, even where hermetically sealed relays are used (some of which can sound reasonable initially) unless their internal construction is very unusual, there is an inevitable deterioration in performance over time, simply as a result of their having several different metals 'in series' with the signal path, some of which are chosen for their springing abilities, and some for resistance to arcing & wear etc.
There is no mystique here, of course, but simple very high quality engineering, which regrettably costs a lot to provide.
John,
Going back to your comments, yes, I have listened (a great deal!) to these things, and they all affect the sound in some way or other.
However, unlike passsive (and some active) components, in my experience they don't specifically 'colour' the sound. i.e. affect the balance of sound, or accentuate any part of the audible spectrum, in quite the same way.
What I hear is a veiling of everything, in different degrees depending on the contacts' materials and construction of the switch or relay. This veiling or fogging of the signal is quite noticeable if a hard-wired bypass is used as a comparison, and results in a loss of details, dynamics, and sound-stage etc.
If one looks closely at say a Shallco switch, the input wiring is soldered directly to the opposite end of the silver plug which actually makes up one of the contacts, and there is a heavy double 'bridge' of very hard silver, the (4 in total) narrow edges of which then wipe along a circular track which forms the output contact.
The twin edges of this hard sprung bridge being under considerable pressure at each end, will, as you suggest, cut through any sulphides, or whatever, which do tend to develop in normal use in domestic (or commercial) environments, on either the multiple contacts or on the 'ring' of hard silver forming the other contact of the switch.
Apart from the two soldered joints (input & output) which every device would need, anyway, there are merely two other 'discontinuities'. i.e where this wiping 'bridge' makes contact with the inputs & outputs, but everything else is 'solid' metal (silver in this case, which helps!). Also, these two discontinuities are continually being physically 'cleaned' every time the switch is operated.
Compare this with less well-made switches and relays and generally there will be many metal-to-metal junctions in their internal construction, mostly utilising dissimilar metals which are riveted together. All of these joints will suffer over time as a result of (microscopic) corrosion which is always present when dissimilar metals are in intimate contact, especially where heating and cooling takes place in addition, which encourages movement between the parent metals due to their differing thermal coefficients of expansion.
So, even where hermetically sealed relays are used (some of which can sound reasonable initially) unless their internal construction is very unusual, there is an inevitable deterioration in performance over time, simply as a result of their having several different metals 'in series' with the signal path, some of which are chosen for their springing abilities, and some for resistance to arcing & wear etc.
There is no mystique here, of course, but simple very high quality engineering, which regrettably costs a lot to provide.
Grey, (and others), servos can be problematic, IF the effective bandwidth is not low enough. Decades ago, I thought that it would be a good idea to make a servo also act as a hi pass filter. BAD IDEA! It really effected the sound and I found that a good coupling cap was actually better.
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. This isn't obvious, unless you look into it, how very low you have to go, in order not to effect things.
Also, servo gain should be the minimum necessary to do the job, but no more. The servo should be considered an 'impure', but necessary, element that should be buffered from the audio signal as much as possible.
It really effected the sound and I found that a good coupling cap was actually better. 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. This isn't obvious, unless you look into it, how very low you have to go, in order not to effect things.
Also, servo gain should be the minimum necessary to do the job, but no more. The servo should be considered an 'impure', but necessary, element that should be buffered from the audio signal as much as possible.
The point where you apply the DC-servo output is also important, I think...
I mean for example if you DC unbalanced a differential pair bias by applying too much DC feedback to correct the output offset, it can't be very good. But of course we suppose that this output offset is not so important and that all circuit is well tuned on this matter, and that the temp drift is well compensated...
I use a cut off about 0,5 hertz, more precisely a time-constant of approximately 3 seconds.
I mean for example if you DC unbalanced a differential pair bias by applying too much DC feedback to correct the output offset, it can't be very good. But of course we suppose that this output offset is not so important and that all circuit is well tuned on this matter, and that the temp drift is well compensated...
I use a cut off about 0,5 hertz, more precisely a time-constant of approximately 3 seconds.
If a circuit needs so much DC applied that it throws everything out of whack, I would tend to think that either the circuit topology needs more work or that the parts are insufficiently matched.
I've seen servos that run as high as 5 or 6Hz. To my way of thinking, that's clearly a sonic problem. Granted, the thing responds more quickly to DC offset, but it's getting into the low end.
As my math instructor used to say,"Boo, hiss, scowl, frown...yuck-icky-poo!"
(He was the sort of fellow who could say something like that without people questioning his masculinity. Not only that, but he could say it with a straight face, which is more than many can do.)
John,
Do you have any views on using multiple filter sections or does phase become a problem? I've never gotten around to trying it, in part because I don't trust the stability.
Grey
I've seen servos that run as high as 5 or 6Hz. To my way of thinking, that's clearly a sonic problem. Granted, the thing responds more quickly to DC offset, but it's getting into the low end.
As my math instructor used to say,"Boo, hiss, scowl, frown...yuck-icky-poo!"
(He was the sort of fellow who could say something like that without people questioning his masculinity. Not only that, but he could say it with a straight face, which is more than many can do.)
John,
Do you have any views on using multiple filter sections or does phase become a problem? I've never gotten around to trying it, in part because I don't trust the stability.
Grey
It is important where to return the servo to the input.
The further down the circuit path, the better.
The CTC connection is in the second stage, which is really the cascode of the amp. I use 4 separate connections. This tends to make the input to be less effected by the servo, which is probably a good thing. You might call my line stage a hybrid of servo and open loop only.
The further down the circuit path, the better.
The CTC connection is in the second stage, which is really the cascode of the amp. I use 4 separate connections. This tends to make the input to be less effected by the servo, which is probably a good thing. You might call my line stage a hybrid of servo and open loop only.
I also think, that the offset compensation must be done as downstream as possible in the audio circuit. In the case of a preamp or a line amp something like the schematic below could be interesting and of less “disturbance” for the audio circuit.
I didn’t really compare it yet, but now that I know that John Curl and J. Carr use this approach it could be good.
DC-Servo example.
Another way could be “a la” Charles Hansen on his V3 power amp, if I remember well he carries out the offset correction by the means of the folded cascodes.
John, yours could be something like that, but only different?
“I use 4 separate connections”, does it mean 2 connections for the offset to ground (1 per side) and the other 2 for the offset between outputs ? Do you use the same “flow” from the dc servos compound for the upper (P) and the lower (N) cascodes devices of one side ? Lots of questions !
What I need to best understand is why 2 offset corrections? I mean, if you compare and correct the DC offset at both outputs with respect to ground, then the offset between both outputs of the balanced circuit will be very close one to the other. Thus I guess that your configuration is better but for an unknown reason to me.
I didn’t really compare it yet, but now that I know that John Curl and J. Carr use this approach it could be good.
DC-Servo example.
Another way could be “a la” Charles Hansen on his V3 power amp, if I remember well he carries out the offset correction by the means of the folded cascodes.
John, yours could be something like that, but only different?
“I use 4 separate connections”, does it mean 2 connections for the offset to ground (1 per side) and the other 2 for the offset between outputs ? Do you use the same “flow” from the dc servos compound for the upper (P) and the lower (N) cascodes devices of one side ? Lots of questions !
What I need to best understand is why 2 offset corrections? I mean, if you compare and correct the DC offset at both outputs with respect to ground, then the offset between both outputs of the balanced circuit will be very close one to the other. Thus I guess that your configuration is better but for an unknown reason to me.
I would rather not go into details, but it should be obvious now, that Jcarr and I, as well as Walt Jung, and probably several others involved in hi end audio, take servos seriously, and don't just add one, without seriously attempting to minimize their potential problems. Some of us don't use servos at all, but I find that impractical for what I am trying to do, ie make noiseless polarity switching at the preamp output, rather than somewhere else in the audio chain.
Nobody here doubt about the interest of DC-Servos and I think that they have heard their defects also. I understand that you don’t want to go further into details, but at least can you enlighten me on my last question about the interest of the “2 offset corrections” ? I don’t think that it could reveal anything of your design.
You brought the discussion on this subject… It would be a pity to close it now ?
You brought the discussion on this subject… It would be a pity to close it now ?
JUSTCALLMEDAD
"'Nobody here doubt about the interest of DC-Servos"
Well, not sure.
As french, you may know Perrot's (= Hephaistos of L'audiophile) writings and some of his extremly pertinent remarks about the DC feedback in any amplifying unit.
See post 19 of mine here which reminds his point fo view.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=64930&perpage=10&highlight=&pagenumber=2
According to Perrot, the DC servo voltage sould not be fed back to either input of an amplifying device. You can easily find his correlated patent.
As far as I understand, the philosophy behind the Blowtorch preamp seems to have the best DC stability before applying DC servo feedback.
~~~~~~~ Forr
§§§
"'Nobody here doubt about the interest of DC-Servos"
Well, not sure.
As french, you may know Perrot's (= Hephaistos of L'audiophile) writings and some of his extremly pertinent remarks about the DC feedback in any amplifying unit.
See post 19 of mine here which reminds his point fo view.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=64930&perpage=10&highlight=&pagenumber=2
According to Perrot, the DC servo voltage sould not be fed back to either input of an amplifying device. You can easily find his correlated patent.
As far as I understand, the philosophy behind the Blowtorch preamp seems to have the best DC stability before applying DC servo feedback.
~~~~~~~ Forr
§§§
- Status
- Not open for further replies.