The DC Servo we all know and love
Yes/no, we know, but don't love.
For audio amplifiers, it is a solution to a problem that does not exist. There are always artifacts generated as well.
This response triggered me to look at Rod's site.
The article starts with -quo- A good part of this has come from the stupid idea that "The best cap is no cap." -unquo-
Thanks Mark, but those are two different things surely? There are going to be all kinds of differences between group and solo?
Looking at the timelines, it become one of two.
Either Marsh told Jung before 1965 about his idea of a DC servo, or he did not.
If he did, he might have invented it. But in that case, why wait until 1980 before publishing on the problem it is supposed to cure...evil capacitors in line with the signal?
I am sure Marsh is prepared to share with us the first time he met Walt, just to establish the possible veracity of his claim.
Either Marsh told Jung before 1965 about his idea of a DC servo, or he did not.
If he did, he might have invented it. But in that case, why wait until 1980 before publishing on the problem it is supposed to cure...evil capacitors in line with the signal?
I am sure Marsh is prepared to share with us the first time he met Walt, just to establish the possible veracity of his claim.
Richard,
There is the rub ... I would suspect the CMA would maintain bandwidth with gain change but looks that it would be on the verge of instability without major compensation. (not good in my experience).
Bandwdith (excessive) can be a double edge sword after all we are not producing a video amplifier (which has a totally different criteria). I can cite several designs that subscribe to this DC to light philosophy, that have not fared well in the market place.
Current understanding of good amplifier design is that open loop bandwidth exceeds your hearing buy a small amount before application of feedback. Amplifiers with excessive bandwidth can tend to be unstable. Driving a reactive load (real world speaker) can make this problem worse. Not many speakers represent an eight ohm resistive load.
Which brings up the question I asked before about the performance of your amplifier into a reactive load.
Jam
There is the rub ... I would suspect the CMA would maintain bandwidth with gain change but looks that it would be on the verge of instability without major compensation. (not good in my experience).
Bandwdith (excessive) can be a double edge sword after all we are not producing a video amplifier (which has a totally different criteria). I can cite several designs that subscribe to this DC to light philosophy, that have not fared well in the market place.
Current understanding of good amplifier design is that open loop bandwidth exceeds your hearing buy a small amount before application of feedback. Amplifiers with excessive bandwidth can tend to be unstable. Driving a reactive load (real world speaker) can make this problem worse. Not many speakers represent an eight ohm resistive load.
Which brings up the question I asked before about the performance of your amplifier into a reactive load.
Jam
Last edited:
There are circuits which can be DC-coupled AND have no dc drift or offset over time. (forget chopper-stab circuit). But, most of those are low gain circuits.
Take the "simple" HPA of mine as an example..... no dc servo, no caps. Stable over time re drift of dc offset. So, it can be done. Now, the context of Walt Jungs issue was that he was designing a high gain preamp (RIAA) and direct-coupled to DC would present a serious challenge for stable zero vdc on the output due to thermal drift etal. Thus, the recommended servo for it.
No where did he nor I suggest it for a power amp or a line level stage.
Those who have tried it on PA have told me that a multi-stage servo sounds best. I guess that means buffered more.
Drift should be a very slow voltage change... and not to be used to correct for a dc off-set. Slow as in days, weeks or months. So. A very long time constant would be used... just enough to steer the drift rate to a fixed point (zero or near zero dc).
Thus, you would not audibly hear any affect with such a low low low cut off freq.
I hope that helps clear up the "real story" behind my solution for Walts phono preamp allowing it to be dc-coupled... yes, no cap is "better" than any cap even my own cap design (patented REL-Cap).
The size and cost of an extreamly low loss, low DA cap is large and that is a secondary reason for the dc-servo.
THx-RNMarsh
Take the "simple" HPA of mine as an example..... no dc servo, no caps. Stable over time re drift of dc offset. So, it can be done. Now, the context of Walt Jungs issue was that he was designing a high gain preamp (RIAA) and direct-coupled to DC would present a serious challenge for stable zero vdc on the output due to thermal drift etal. Thus, the recommended servo for it.
No where did he nor I suggest it for a power amp or a line level stage.
Those who have tried it on PA have told me that a multi-stage servo sounds best. I guess that means buffered more.
Drift should be a very slow voltage change... and not to be used to correct for a dc off-set. Slow as in days, weeks or months. So. A very long time constant would be used... just enough to steer the drift rate to a fixed point (zero or near zero dc).
Thus, you would not audibly hear any affect with such a low low low cut off freq.
I hope that helps clear up the "real story" behind my solution for Walts phono preamp allowing it to be dc-coupled... yes, no cap is "better" than any cap even my own cap design (patented REL-Cap).
The size and cost of an extreamly low loss, low DA cap is large and that is a secondary reason for the dc-servo.
THx-RNMarsh
Last edited:
Richard,
There is the rub ... I would suspect the CMA would maintain bandwidth with gain change but looks that it would be on the verge of instability without major compensation. (not good in my experience).
Bandwdith (excessive) can be a double edge sword after all we are not producing a video amplifier (which has a totally different criteria). I can cite several designs that subscribe to this DC to light philosophy, that have not fared well in the market place.
Current understanding of good amplifier design is that open loop bandwidth exceeds your hearing buy a small amount before application of feedback. Amplifiers with excessive bandwidth can tend to be unstable. Driving a reactive load (real world speaker) can make this problem worse. Not many speakers represent an eight ohm resistive load.
Which brings up the question I asked before about the performance of your amplifier into a reactive load.
Jam
Jam,
You did not make a CMA. They are very stable and not comp as you describe. CMA are used all the way up thru GHz freq range with no oscillations. They are very stable. Study thier compensation method and look at some HF/RF CMA circuits as well to understand.
Second, did you forget what I showed you and told you at my house? The HF and how that is converted into audible IM if the amplifier cannot reproduce it cleanly? HF from smps and digital amps and DAC's. Your experience/examples are all related to VFA and CFA designs. Not applicable to the CMA.
I will give you a way to estimate if a circuit will be in CMA ...... look at the circuit Z and currents. If the circuit Z are low values and currents are high, and topology is right, you probably have a CMA. That isnt fool proof but close enough if you have to guess and not measure.
Again, I will use my original line stage to illustrate CMA behaviour.... First I went to a topology which would eliminate caps again. Then, I used low circuit Z values... the highest value used is 1K. 100 Ohms to 1K. And, then i ran it at high currents with a +/- 24dc supply.
The BW was very wide, rock stable and I had to run the output near clipping to see any distortion; At +/- 22 v p-p the thd was .005%. In CMA operation... low fb. This is late 1970's.
Try to keep up... this is 40 plus years ago.
Audio slowly got around to the CFA but I suggest trying CMA also. THx-RNMarsh
Last edited:
- I think a problem exists. DC offset in the output. Means the need to adjustments to minimize-it. In an industrial production, it has some cost and drawbacks.
- What are the artefacts you are talking about ?
On an objective and technical point of view, if there is no phase turn at 10Hz, and no increase of distortion (but the contrary) with the servo, could-you justify this assertment ?
https://www.diyaudio.com/forums/attachments/solid-state/787498d1570966205-pizzicato-200w-low-distortion-cfa-amplifier-pizzi-schema-gif
Here, the unusual* servo I use in this project.(See attachment)
The original audio signal is canceled in the comparator, before to be filtered and the remaining DC error applied.
Is here any luck we could notice a subjective and noticeable impact on the sound character ?
* No patent pending ? ;-)
Attachments
Last edited:
Thanks.
-RNM
I found it strange that no one commented. Could it be that some "EEs" (read my lips ;-), so fast to criticize everything and everybody, did not notice or understand ?
Last edited:
I was talking about the thread devoted to this project. They would sleep for six months and their grunts would be only noises meaning nothing that they make in their sleep? ;-)
You reassure me.
I feel that a great love story will start.
Attachments
Last edited:
Note that I took care to place myself in the first row, in the center of the photo.
Your claim was about "snake oil" and "business pitch" in my posts at the hydrogenaud.io forum.
And you failed to provide evidence for "snake oil" or any "business pitch" in my posts on hydrogenaud.io .
Instead you posted content about what other members imaginations. Don't you know the difference between real evidence and the usual banter when people are running out of arguments?
To refresh my example from the last time; if I would claiming that "Evenharmonics is lying on diyaudio" and after being asked for evidence, I simply would quote the opinion of other members that "Evenharmonics is lying on diyaudio", should that count as evidence for the fact of "Evenharmonics is lying on diyaudio" ?
- Status
- Not open for further replies.