To use DC servo or not?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Is it worth the extra effort of using a DC servo on an amplifier ? Is there any big advantage over using a DC blocking capacitor in the feedback network ?

I've looked at putting in a servo into my design (similar to the Leach Amp) and have messed about a bit in the simulator, but that's it. One thing that intrigues me is servo's that cancel the offset by playing with the LTP current sources.

Any comments from those far more experienced than me ? :)
 
One thing that intrigues me is servo's that cancel the offset by playing with the LTP current sources.

Steven used one and published in thread : 'is there anybody built a non-feedback amp' or something like this.
Then it must be good solution :)

Is there any big advantage over using a DC blocking capacitor in the feedback network ?

Obvious one is lower value of capacitor. Caps like 1u are of much better quality than big electrolytics.
 
DC servo kit from Taiwan

Hi all,

I believe I come across this thread 3 months too late. I am also very interested to hear about others opinions on this subject.

As matter of fact, I have found and purchased couple of this DC servo kits from a web site in Taiwan. The DIY community there has been heavily promoting this simple little circuit as a major improvement to sonic performance of any audio gear that require the use of large size (capacitance) capacitor as a mean of DC de-coupler or filter in the signal path. They also advertised that this device can be applied to power amp in order to eliminate the need of using high capacitance bipolar electrolytic caps as a low frequency filter in the feedback path or to pre-amp & DAC in order to eliminate the need of DC blocking caps on the output.

Attached please find the circuit diagram. I don't believe they are the inventor of this circuit but rather, as far as I know, the first DIY group generally implement it.

Well, I don't about pre-amp and DAC application but I would most certainly try it on my Leach Superamp. After all, these kits only costed 1/4 of the price of a very good NH-series "Black Gate" electrolytic capacitor.

So far, I have finished soldering the board and in the process of building a separate DC power supply for it because the DC rail on my power amp is too high for the kit to handle.

Will post further updates on this project soon.
 

Attachments

  • dc servo.jpg
    dc servo.jpg
    57.7 KB · Views: 3,169
I have never used a DC-offset servo.
I havent found it necessary.
With amplifiers having a gain of only 15-25 this is not often a problem.

I do not know if any degrading of performance
has been reported by a small DC-offset, say within +-0.100 Volt.

An input capacitor of good quality wouldnt hurt much.
 
Hi lineup,

There are potential huge problems with DC servo's degrading the sound. Most simple servo's use a non-inverting chip whose gain reduces to unity at HF. But the cheap FET chips used have very poor PSRR at HF and, if their rails are not adequately filtered at HF the PSRR lets through the spray of even harmonics generated by AB on the supplies of the amp - and feeds them through to the input (usual).

The solution is to split the output resistor from the servo to input and put in a C to ground. Cleans it right up.

Cheers,
greg
 
An input capacitor of good quality wouldnt hurt much.

lineup,

Thanks for your comment.

Please advise me, how would a input capacitor can improve the DC offset condition on the output of a Leach Power Amp when the root cause of the problem came from slight mis-match of the transistor or other commponents in the input stage??

Should I add a coupling capacitor on the input even though the original design does not require one??

amplifierguru,

Thanks for your comment. Yes, the OPA627 was a bit of overkill. The kits actually came with OPA134, but I have couple of OPA627 sitting around doing nothing, may as well use them for experimental purpose. I will even try out other OP-Amps in my stash too, such as AD797, and OPA637 etc.,

Just try to have some fun.

:D
 
jaycee said:
Any comments from those far more experienced than me ? :)
I have used DC-servos in almost every design I have made and it works very good as long as you do it right.

The opamp must have good audio properties up to 1 kHz (which almost any has) and also descent DC properties. It's also an advantage to use an opamp with small input bias currents because then you will have no problems with offset coming from that source.
 
amplifierguru said:
There are potential huge problems with DC servo's degrading the sound.
I would say that there are microscopic problems with DC servo's as long as you do it right, good power supply etc.

One way to reduce the potential HF problems is to insert a filter at the servo output. By this you create a very small hunch at 100 Hz or so but 0.1-0-3 dB is bearable.
http://www.sjostromaudio.com/hifi_files/qrv/qrv08r0schema_p3.pdf The whole design consists of 6 pages.
 
peranders,

Thank you for your information. It was very helpful. I am sure will insert a filter on the servo output. Just try to be on the safe side.

I prefer to keep the overall servo output resistor to be 20k in order to keep it in line with the resistor of other feedback paths of the Leach Amp circuit.

Can I just split the 20k into two 10k and then insert a 0.1uF cap in the middle?
 
Should I add a coupling capacitor on the input even though the original design does not require one??


Yes, I would.
There is no reason not to.
If you think you will have bad or even noticable performance
by adding, say a 4.7uF polypropylene,
I say you better test it.
There are 1000 of high quality amplifiers, using this little trick.
:)

Does the Leach original use DC-servo ?
If not there is no need to.

Mismatch causing offset.
Shouldnt it be possible to do matching? This is what I should do.
Matching is for me one important phase
in building an amplifier.
I always do this.
 
Administrator
Joined 2004
Paid Member
Hi jtcc1015,
Keep in mind that a 10K resistor with a cap on it may introduce a filter like effect at about 16 Hz. At high frequencies the impedance at this point just dropped below 10 Kohms. You should increase the resistance to a much higher value.

-Chris
 
Hi jtcc1015,

My suggestion -

....The solution is to split the output resistor from the servo to input and put in a C to ground. Cleans it right up.

This is an output filter, Peranders.

As Anatech says, 20K is too low. Splitting it in half with 10k then Cof say 10uF, then 10K back to your feedback input puts 10K across your FB resistor to ground which, if it's 1K will alter amp gain by 10%.

100K resistors would be better since the servo correction output can go to +/-14V.

You will find the OPA627 has a tighter spec on DC offset than the OPA134 but, in practice, there's little in it. Being both FET chips they both have adequately low input bias currents to allow 1M input filter resistors for long time constant.


Of course there is a far more elegant (better and easier) way to provide DC offset adjustment on dual diff'l input amps, which I use in the SKA and 300W version.

Cheers,
Greg
 
Hi Jacco,

Sorry mate, just can't give everything away - as Hugh would say, it's proprietary.

But if you draw a double differential with a commoned input, and a commoned feedback point and R divider from the output and sit and look at it for awhile looking for places where you can find a nice steady low Z defined voltage, it may come to you as it did me. Those who have bought my amp kits have it.

Merry Christmas all
remember those less fortunate

Greg
 
anatech,

I meant to use 10k and 0.1uF to set up low-pass filter on the output of the DC servo not 10uF. Therefore, the roll-off frequency for 10k and 0.1uF combination should had been 159.155Hz instead of 16Hz as what you had mentioned.

Did I get my calculation right?

My original objective of incorporating a Taiwan made DC servo kit on my Leach amp was to try to find out if it is as good as what those Taiwanese claimed. As matter of fact, I have been using my Leach Superamp in its original design for more than 2 years already, and I am quite happy with it. During the construction of it, I matched all the transistors, zener, and diodes as well as used good quality "Holco" resistors. As for capacitors, I used "Black Gate" for electrolytics, "Kimber Kap" for power supply de-coupler and high-frequency bypass, "Sliver Micca" for pF range position and sliver coated, "Telfon" insulated wires all the way. The output DC off-set with input shorted to ground was about +/-11mV only, which was well within acceptable range. In effect, I don't really need a DC servo at all. As I have mentioned, I simply wanted to experiment and, if possible, learn about the advantage and dis-advantage of using a DC servo. Therefore, I would treat it as a add-on circuit rather than altering the original Leach's design to accommodate it. By that, I will connect the input of the DC servo to the Amp's output (the speaker +ve), and the output connect to the feedback summing point of the Amp's input stage.

Since the DC servo output will be connected in parallel with the existing feedback path, my after thought will be not adding a high-frequency filter at all. Otherwise, part of the feedback signals with frequency higher than 159.155Hz will be shunted to the ground.

peranders,

very much appreaciate your help. I will look into the LTSpice. I have actually thought of adding a speaker protection circuit before I connect the DC servo because I understand the fact that an Op-amp circuit can have higher failure rate than a simple RC filter. Therefore, I am working on both upgrade kits simultaneously.

I can see it will get a while to get these projects completed as I am not going work on them during Christmas and New Year holiday.

Shall post further update soooooon.............

Merry Christmas to you all.
 
Is servo or not servo is the real problem ?
DC feedback is to maintain the DC output of the amplifier at 0V. Two mechanisms affect this DC : the DC offset at the output of the input differential pair and the DC offset due the the following stages of the amplifier. To get the lowest and most stable distorsion from an input differential pair, it should always work with strictlly equal operating points currents in both its devices. Any DC difference due to DC feedback will affect this operating point. So capacitor or servo feedback do not make great changes. A capacitor is much easier, a bipolar type is recommmanded for ultimate performances.

A more logical way to control the offset at the output of the amplifier is a servo which will act after the input stage. See Perrot's patent US 5 635 874.

Another way would be to connect the input stage to the following stages only at AC : an all new scheme bringing back coupling capacitors.

~~~~~~~ Forr

§§§
 
Administrator
Joined 2004
Paid Member
Hi jtcc1015,
I haven't checked my calc's, you may be correct. But that figure is even worse for the amp! You got the point, and that's all I wanted.

forr is hit on something I have noticed over the years. If an amplifier is balanced and the diff pair doesn't have to do much work to maintain a DC balance, the amp sounds much better. I didn't know why that was until having the time to study this in the last few years. A DC servo circuit may interfere with the diff pair, throwing the collector current balance out.

A better place for injection of the correction signal may be in the VAS. I still don't like this, but it may be better than upsetting the differential amp in the front end.

-Chris
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.