DC servo

[uli]

Peranders:

Did you ever try to hear a possible difference in 'sound' in your SMD GC with or without this DC servo?
Many people claim this difference. As I always try to avoid caps in the signal path and thus use servos when I need a dc free environment, I am not sure anymore about my own 'objective' hearing.

Uli



PS: even my phono stage is completely dc coupled.

[Roushon]

this is a very good point raised. i will also be interested in seeing a discussion on it as i am going to build my GC ( parallel of two inverted LM3886 with separate buffer). many people suggested me not to use the dc-servo.

[peranders]

No, I haven't tested it yet. But as you know my DC servo is in world's class using a AD8620, not like LM324 or uA741 exactly.

I'm pretty sure that many of those "bad" DC servos are made of not very high classy opamps. The servo is an integrated part in the audio chain so it must be good, at least up to 1 kHz.

[uli]

up to this point my experiences with dc servos are just good ones. the only issue is the costs of the extra parts and the design, which in fact isnt really an issue. I started like many Europeans with that famous hi end preamp from "Elektor" in the late 80ies. Soon I swapped the outputcaps as there is a servo circuit in the phonostage already. I copied this circuit to the output opamps (old PMI OP27 better than BB) and the performance gain was enormous!
Recently I designed some kind of balanced opamp with diff in and diff out. This circuit although well sounding is very difficult to control without servo. During tweaking people often loose their objective standards and exactly this is my problem. Now Im seeking for other opinions.

Uli

[uli]

Originally posted by peranders
No, I haven't tested it yet. But as you know my DC servo is in world's class using a AD8620, not like LM324 or uA741 exactly.

I'm pretty sure that many of those "bad" DC servos are made of not very high classy opamps. The servo is an integrated part in the audio chain so it must be good, at least up to 1 kHz.

Youre right.

I use OP177 and MKP in the servo because I have some dozens of them although the AD8551 has better dc characteristics.

Ive seen servos with LF353 and elkos. It is very clear (at least for me) that such a circuit wont contribute anything good to the result

Uli

[Roushon]

what is your opinion about using LF412 as servo as in AN-1192? though i have got some AD8620. I am just curious.

thanks

[peranders]

LF412 is like a Volvo, not a Mercedes or BMW. I think it's OK.

[Roushon]

[uli]

Originally posted by Roushon
what is your opinion about using LF412 as servo as in AN-1192? though i have got some AD8620. I am just curious.

thanks

there are many possibilities to implement a dc-servo. One is to simply (?!) make an integrator in the feedback loop as often seen in manufacturers applications. You can add a RC combo at the output of this integrator to achieve a steeper curve. This has some advantages. The servo itself influences the audio circuit not as 'dramatical' as the standard integrator and thus you can use a broader range of devices. The drawback is the quality of the cap used in this RC combo. It should be of reasonable quality.


Personnally I wouldn't quote the 412 as a Volvo - more like a Skoda or Fiat, hoping that this PERSONAL opinion does not lead to another flame war about car quality!


Uli

[Roushon]

Hi Uli,
Thank you very much for your details explanation. As far as Audio electronics is concerned I am a newcomer. I am learning reading different threads, so details logical explanation helps a lot. I have one question. In the circuit design of LM3886 (or any such amplifier circuit) how is the offset voltage generated? Is it because of the chip or the choice and quality of the outside components. I mean would a better design (other than the circuit in the datasheet) around LM3886 without using any other op-amp will help to get rid of the offset voltage? I came to know T-network help to a certain extent. May be it is not a question at all, but please let me know...

With best regards.