Assuming there is a capacitor in the correction circuit
(and there always seems to be one) then we cannot
say that capacitors have been removed from the
signal path.
The presumption is that the "servo" capacitor can afford
to be of higher quality because it can be a much smaller
film type. Additionally, as I pointed out before, the
correction can be injected into more innocuous points,
like the diff pair current source, where you will see some
differential rejection.
Of course the only capacitorless circuit I have seen was
battery powered.
(and there always seems to be one) then we cannot
say that capacitors have been removed from the
signal path.
The presumption is that the "servo" capacitor can afford
to be of higher quality because it can be a much smaller
film type. Additionally, as I pointed out before, the
correction can be injected into more innocuous points,
like the diff pair current source, where you will see some
differential rejection.
Of course the only capacitorless circuit I have seen was
battery powered.
-------------------------------------------------------------------------peranders said:
Also I can identify much more easily the type of coupling cap in Pass Preamps (Aleph P) than direct coupled servo devices. The coupling caps on the Aleph must be of a much higher quality than the standard ones for truly good sound quality; I don't care what the REVIEWERS say.
peranders said:
Well, let's think about this. I will use the Leach Amp as an example, as I familiar with that. Here, the DC-gain capacitor (C6 and C7) is 220 uF (plus a low value film cap) isolated with a 1K1 Ohm resistor. At audio frequencies, about 1/21th of the output voltage is being fed back, so this is all that the capacitor sees. It does not see the full audio signal as, say, an input coupling capacitor would. Actually, the signal does not even pass through it, as the cap is in parallel, isolated by that 1K1 resistor. I admit that the signal can still be influenced by any distortions that the capacitor can generate, though ideally, at audio, it is a dead short.
For the servo amp, at least the with the integator approach I am familiar with, an attenuated audio signal is fed through an op amp which has a series-connected capacitor in the feedback. Any instabilities in the servo amp (and there should be none in a well designed servo amp) and any distortions in that capacitor can potentially impact the sound. As has been said, the distortions of the smaller capacitor should be less audible than the distortions of that big 220 uF capacitor, even if it is an esoteric one and even if it is bypassed with a lower value, high quality film type.
At least this is how my feeble mind sees things.
Anyway, I am intrigued by all of this and am becoming less prejudiced towards servo amps. I just need to justify in my own mind that they can not only be just as good as the NFB approach using "esoteric" capacitors, but better. The jury still seems to be out.
You have got some things backwards (sorry) but all caps which are involved in the signal path has always a very little AC (audio) voltage across them, even the input cap.
It's very important taht the feedback network is linear (concerns all parts inlcuding caps) otherwise you will get distortion.
The advantage to have a DC-servo in a power amp is obvious. The opamp has most likely MUCH better DC-parameters than discrete transistors. Normal offset voltage is 25-100 mV with 50 mV as a normal value. With almost any opamp you can get less than 1 mV offset regardless of the gain.
It's very important taht the feedback network is linear (concerns all parts inlcuding caps) otherwise you will get distortion.
The advantage to have a DC-servo in a power amp is obvious. The opamp has most likely MUCH better DC-parameters than discrete transistors. Normal offset voltage is 25-100 mV with 50 mV as a normal value. With almost any opamp you can get less than 1 mV offset regardless of the gain.
Normal offset voltage is 25-100 mV with 50 mV as a normal value
I think with a matched input diff pair that offset voltages in the single digit millivolt range are nt that diffcult to achive with monolythic diff pairs like the 2SK389. But that comes for using decent semicodutors and design techniques I guess.
Servos can use smaller value capacitor than an AC coupled circuit is the only real advatage. The diasadvantage is the sonics of the opamp in the servo, the need for possible addition supplies for the servo, and circuits interaction with interfacing to to output of the servo opamp. You now have an op amp with a capacitor in 2 feedback loops and the possible additional of a power supply for the opamps, instead a simple coupling cap to deal with. also you had better make sure the circuit powers up without a big low frequecy transient from the servo. I would rather buy a decent coupling cap. who needs a servo when most caps sound the same anyway, right?
I think with a matched input diff pair that offset voltages in the single digit millivolt range are nt that diffcult to achive with monolythic diff pairs like the 2SK389. But that comes for using decent semicodutors and design techniques I guess.
Servos can use smaller value capacitor than an AC coupled circuit is the only real advatage. The diasadvantage is the sonics of the opamp in the servo, the need for possible addition supplies for the servo, and circuits interaction with interfacing to to output of the servo opamp. You now have an op amp with a capacitor in 2 feedback loops and the possible additional of a power supply for the opamps, instead a simple coupling cap to deal with. also you had better make sure the circuit powers up without a big low frequecy transient from the servo. I would rather buy a decent coupling cap. who needs a servo when most caps sound the same anyway, right?
Nelson Pass said:
Precisely.
And this was why I was disappointed when you advised me in the "Lamm v.s Aleph" thread that you will be using coupling caps in the 'new' SOZ circuits.
It is just unfortunate that the 'original' SOZ circuit is so inefficient.
One only needs to try the experiment I mentioned (also in that thread) of temporarily shorting out DC feedback blocking caps in amplifiers, to hear just how much the sound improves, even if they do happen to be BGs or good quality filmcaps.
On two other amps I conducted similar trials on, I temporarily disconnected their DC servos, with much the same results.
It is therefore clear to me that one is better off without either of these devices, and, until something better comes along, SOZ (Mark 1) rules the waves!
Regards,
There is another side of a coin here as well. One can design relatively simply circuit which requires coupling cap. The same circuit can be implemented without the cap, yet at the cost of more complicated design which allows capacitor free interface. So the cap is eliminated, but the circuit is more complicated with more parts, both passive and active. What is better for the sound? I noticed that as long as you don't short the cap, you don't really know what you are missing. But I never managed to ad more parts to hear the same circuit and compare.
I noticed that as long as you don't short the cap, you don't really know what you are missing. But I never managed to ad more parts to hear the same circuit and compare. Peter Daniel said:There is another side of a coin here as well. One can design relatively simply circuit which requires coupling cap. The same circuit can be implemented without the cap, yet at the cost of more complicated design which allows capacitor free interface. So the cap is eliminated, but the circuit is more complicated with more parts, both passive and active. What is better for the sound?
Hi Peter,
I am not absolutely sure I fully understand your meaning here, so please excuse me if I have got it wrong.
Surely someone like yourself cannot be saying that 'ignorance is bliss' with audio gear, and that any potential improvements which you haven't yet experienced are not really worth striving for?
If so, it seems odd that you spend so much time making such super looking, and apparently good-sounding, gear as you have shown us all.
I always enjoy your posts very much, and, as I have said in the Forum before, my views seem to mirror your own, generally.
Perhaps from what I have read you are more interested in the actual mechanical/electronic construction (rather than in the electronic circuit design of the equipment you make) than I am, but you have built some very well-designed gear, and to an extremely high standard.
I therefore simply cannot believe that if there was a simple way of achieving an improvement in the sound of anything you construct, you wouldn't bother to try it out for yourself, purely on the basis that if you haven't already heard that improvement, you are quite happy to do without it!
Like me, you seem to like BGs, but as I have said many, many times, I hate their cost and ridiculous burn-in times, and however good they sound when you have gotten over these hurdles, they are still not perfect, unfortunately. (That's just blown any chances I had of any freebies from BG , even tho I seem always to be singing their praises!)
Therefore, wherever possible, I would prefer to do without them, and I simply cannot tolerate the awful results I hear from most other electrolytics.
Some 20 years ago I did a lot of listening tests on feedback amps, and in particular, direct coupled ones, but they all sounded considerably worse for the inclusion of the (usually) necessary DC blocking caps. I tried everything, including 'recasting' the circuits to permit the use of much lower value film caps (which needed commensurately larger feed-back resistors like say 150k- which added more noise than otherwise!), and nothing sounded anywhere near as good as a plain straight wire.
After a discussion with Ben Duncan, who is quite well thought of as a designer, writer, and commentator in the UK on audio matters, I was persuaded to try some servo circuits to avoid the need for these wretched caps.
Fortunately, Erno Borbely published some suitable 'servo' designs in TAA at about that time, so this also made it fairly easy for me.
Regrettably, although I thought I was then well on the road to audio nirvana, as I said in my earlier post, subsequent listening tests told me that this was just not so.
In my experience, being fussy about every single area in a circuit, and utilising the best techniques and marrying them with the best sounding components, does make a lot of difference to the resulting sound, so that is what I now always try to do.
One weak link in the chain can spoil the sound, or at least make it a lot less good than it otherwise could be, so I don't leave any stone unturned in my quest for improvements.
In this particular case, I was rather pulling Nelson's leg as I was already aware of the difficulties of doing a great deal to improve (overall) the SOZ circuit, without making some 'conceptual' topological changes. But, as you have implied, improvements in the resulting sound do not always follow as a result of adding more components, which frequently add their own 'shortcomings'.
Regrettably, in spite of my near to 40 years experience with DIY audio, I still lack the imagination needed to come up with any radical and new circuit ideas of my own, and I am therefore reliant on the generosity of 'free-thinking' designers like Nelson, if I wish to try out something really new.
Having a start, based on another's published design, I will then tinker about with circuit variations, and especially the most suitable-sounding components in the various areas of the circuit, until I am satisfied that I have got the best which is possible from that particular design.
I am not paranoid about it, but if I discover that there is a better 'sonic' result from any particular experiment (which is usually done initially on a temporary, 'lash-up' basis until the results are known), I will certainly not simply forget about the improvement I heard.
Unfortunately, in many cases there simply doesn't appear (at least to me!) to be a realistic or perhaps cost-effective way of making the improvement become more permanent, and feedback DC blocking caps are just one case in point.
I don't care for the idea of 'frying' my speakers with large and/or varying DC offsets, which even if they do little permanent damage to drivers, they do spoil the operation of moving coil drivers (in spite of what I continually read about this topic).
In most cases it doesn't take much DC voltage to displace a voice coil relative to the magnet gap, which effectively 'biases' its normal operation by then functioning in a 'lop-sided' manner, and because of this displacement (which also 'pre-loads' or stresses the cone's elastic surround) this worsens any distortion or non-linearities in its operation.
The effect is quite noticeable (even with short voice coils working in long gaps [like say ATCs] which I generally prefer), and the result is a less 'relaxed' overall sound.
In fact, too little attention is given to this particular aspect, I believe, especially when I often see comments about quite high offsets in amplifiers, and if like me you have tried introducing such an abberation deliberately, you will see what I mean.
This is merely another case where attention to detail will provide an overall improvement in sound (which may well not be readily noticed if one has always lived with it in a particular amplifier), and one which, in my view, is also well worth having!
Lesson and rant over!!!
Regards,
I was only implying that sometimes the coupling cap is a lesser evil comparing to a more complicated circuit in order to avoid such a cap.
It was hard for me to choose the proper capacitors for my Aleph X amp. They form a 6dB high pass filter and are definitely necessary in that location. I played for many hours swapping caps, reversing them, paralelling and eventually at the end of a day I was satisfied. I chose 0.1u Jensen copper foil paralleled with 0.02u Silvertone silver foil cap. Without the silvertone, Jensen sounded non impressive, it was the silver cap that added sparkle and life to the presentation. I also noticed that polarity was very important and when placed opposite direction the caps sounded like crap. Outer foil to the load worked better.
It was my personal amp and the cost of the caps was about $160 (just for stereo balanced inputs). I can't imagine I would be using this configuration if it was my commercial offering, its just too expensive. When I build a gainclone I use 4.7u BG N non polar cap ( $5 ) and the results are very pleasing. I mean, whoever listens to this amp, likes the sound. That's why I don't see a reason to try something more exotic and spoil the fun. If it works, don't touch it, they say.
It was hard for me to choose the proper capacitors for my Aleph X amp. They form a 6dB high pass filter and are definitely necessary in that location. I played for many hours swapping caps, reversing them, paralelling and eventually at the end of a day I was satisfied. I chose 0.1u Jensen copper foil paralleled with 0.02u Silvertone silver foil cap. Without the silvertone, Jensen sounded non impressive, it was the silver cap that added sparkle and life to the presentation. I also noticed that polarity was very important and when placed opposite direction the caps sounded like crap. Outer foil to the load worked better.
It was my personal amp and the cost of the caps was about $160 (just for stereo balanced inputs). I can't imagine I would be using this configuration if it was my commercial offering, its just too expensive. When I build a gainclone I use 4.7u BG N non polar cap ( $5 ) and the results are very pleasing. I mean, whoever listens to this amp, likes the sound. That's why I don't see a reason to try something more exotic and spoil the fun. If it works, don't touch it, they say.
I couldn't agree more, Peter, and it seems I hadn't properly grasped what you were trying to say earlier, as I half expected.
Glad to hear we are still on the same wavelength, as you had me worried for a while!
The motto is: "When in doubt, leave it out", or if that's not possible, do it the most simple way you can.
Or perhaps: "Anything you *don't* use, cannot be badly chosen or go wrong".
I can now get back to my lathe and polishing the screwheads of my latest enclosure, without any worries.
As I said in an earlier post to you, there is no reason why an amateur job should look any worse than what the pros come up with, but clearly I don't need to lecture you on that!
Nowadays, since S/S fixings have become readily available, I always try to use them in place of electro-plated jobs, since, as I mentioned, one can polish the heads up beautifully, and that is something I have never seen on a pro job. (This trick comes from years of old car and bike restoration, which I am now getting a bit old for, regrettably!!)
Regards,
Glad to hear we are still on the same wavelength, as you had me worried for a while!
The motto is: "When in doubt, leave it out", or if that's not possible, do it the most simple way you can.
Or perhaps: "Anything you *don't* use, cannot be badly chosen or go wrong".
I can now get back to my lathe and polishing the screwheads of my latest enclosure, without any worries.
As I said in an earlier post to you, there is no reason why an amateur job should look any worse than what the pros come up with, but clearly I don't need to lecture you on that!
Nowadays, since S/S fixings have become readily available, I always try to use them in place of electro-plated jobs, since, as I mentioned, one can polish the heads up beautifully, and that is something I have never seen on a pro job. (This trick comes from years of old car and bike restoration, which I am now getting a bit old for, regrettably!!)
Regards,
On pages 13 and 14 of the data sheet for the OP42 opamp from Analog Devices, you will find a DC servo conceptually similar to some that I have used for discrete circuitry - a major point being that it does not act directly on the signal amp's plus or minus inputs.
Here the DC servo circuit is being applied to an opamp, and if my experience with discrete circuits is anything to go by, I think that there is a good possibility that a servo implementation along these lines of thought will affect the sound less than a conventional non-inverting servo fed back into the opamp's inverting input, or for that matter, a coupling cap.
Thoughts or opinions welcome.
http://www.analog.com/UploadedFiles/Data_Sheets/569248160op42.pdf
Related question for Scott Wurcer - on the AD797, how do the null pins tap into the circuit? Or at least it would be useful to know the respective polarities of null pins 1 and 5, as I'd like to try an inverting servo.
regards, jonathan carr
Here the DC servo circuit is being applied to an opamp, and if my experience with discrete circuits is anything to go by, I think that there is a good possibility that a servo implementation along these lines of thought will affect the sound less than a conventional non-inverting servo fed back into the opamp's inverting input, or for that matter, a coupling cap.
Thoughts or opinions welcome.
http://www.analog.com/UploadedFiles/Data_Sheets/569248160op42.pdf
Related question for Scott Wurcer - on the AD797, how do the null pins tap into the circuit? Or at least it would be useful to know the respective polarities of null pins 1 and 5, as I'd like to try an inverting servo.
regards, jonathan carr
I have tried the same thing with LT1028/1115, using the offset pins. Works very good. I think it's important to choose an opamp which has good audio performance to at least 1 kHz => don't use LM324 
I have AD8620 as servo opamp in my Gainclone
http://home8.swipnet.se/~w-80046/hifi_files/qsx/qsxm3r0schema_p1.pdf
I have AD8620 as servo opamp in my Gainclone
http://home8.swipnet.se/~w-80046/hifi_files/qsx/qsxm3r0schema_p1.pdf
I have a strong hunch that LT1028 is excactly the same chip but a better selection.
http://www.linear.com/pdf/1028fa.pdf
Look at page 17 in the datasheet for LT1028
The filter at the output removes the unwanted zero which a non-inverting amp creates.
http://www.linear.com/pdf/1028fa.pdf
Look at page 17 in the datasheet for LT1028
The filter at the output removes the unwanted zero which a non-inverting amp creates.
Hoping that this is still on topic, I'd like to put some grains of salt into the equatation.
I don't really understand, how a small DC component can significantly worsen an amplifier. Consequently, I'm under the impression, that it's not wise to spend great efforts in fighting the last mV out of the circuit.
If you don't prefer to listen to single sine waves, there will be for the most time a "DC" component, when amplifying for example a 1kHz tone. Some slower tone, present at the same time. So, your audio chain should better be able to handle this situation without mixing the signals. But if this is the case, small DC shouldn't be a problem at all.
I hope I have expressed myself in an understandable way.
Regards,
Peter Jacobi
I don't really understand, how a small DC component can significantly worsen an amplifier. Consequently, I'm under the impression, that it's not wise to spend great efforts in fighting the last mV out of the circuit.
If you don't prefer to listen to single sine waves, there will be for the most time a "DC" component, when amplifying for example a 1kHz tone. Some slower tone, present at the same time. So, your audio chain should better be able to handle this situation without mixing the signals. But if this is the case, small DC shouldn't be a problem at all.
I hope I have expressed myself in an understandable way.
Regards,
Peter Jacobi
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.