DC Servo using the OPA1656 CMOS Opamp

One thing about DC servo opamps that I had some issues with is at power-up.

Take the TL051. for example. It is specified for +/-5V minimum.
When the supply raills were lower than that at power-up, the output DC is undefined, which can be as much as 5V.
This undefined output state then charges up the feedback cap, which takes seconds to discharge later.
I noticed this when I was using OPA140 as servo, getting a 3V DC at staart-up.

In a power amp, it is probably just a thump.
When used in an input servo, that is a hell of a lot of start-up DC.


Patrick
 
I would not use the TL051 or the 71 for an input servo - the offsets are far too high. Re the noise, I usually use a post filter after the servo. If the corner frequency is set correctly you can reduce the servo 1/f noise by orders of magnitude.

Re the thump, don’t you mute your amps at power up?
 
Well I can see this is important to you.
Would you consider the ADA4000? About a buck?
ADTL082, 50 cents?

Jan
At a glance, TLV9362 (or a few other TLV parts) should satisfy your requirements at a moderate cost.

WRT servos in general, I agree with the Marks (4 and Johnson) that a/ They affect the sound and b/ They are in the audio path.
Years ago we did a lot of listening tests and ended up with an optically coupled servo similar to what Steve Dunlap used in the Krill amp.
That was before a lot of the newer super audio opamps and we were very critical chasing the nth degree of sound quality. In the larger scheme of things it's
pretty subtle and in most replay systems you just won't be able to hear the difference. If you are chasing ultimate audio quality, socket the servo opa and
get someone (preferably younger than you) to do critical listening... or you can just not bother and go with what objectively works.

TCD
 
How much input offset does that correspond to? Is that more or less than the maximum VGS mismatch of the JFET pair?

JFE2140 jfet pair has only 4 mV max offset - the best of jfet pair at all (LSK389 has 20 mV max)

you have to check whether the worst-case offset and the maximum expected deep subsonic input signal can drive the DC loop into clipping.

I use "worst case" option while Microcap12 analysis both for jfets offset, bjts and diodes parameters from 0 to 70 Centigrade and 2% resistors for DC, and 2% for resistors and 5% capacitors for AC. Therefore everything is optimized by Microcap. Using passive RC LPF before intergrator prevents both DC loop clipping @ LF maximum input and loosing main audio NFB loop gain - see here screenshots sukhov_gain_open loop_close loop.jpg and sukhov_stability_loopgain.jpg @ https://www.patreon.com/posts/mm-constant-loop-67857060
 
Last edited:
You use an LSK389C according to the attachment of https://www.diyaudio.com/community/threads/dc-servo-using-the-opa1656-cmos-opamp.388777/post-7089267 so I don't understand the relevance of the JFE2140. Anyway, you checked that the control range is large enough, so that should be OK.
Yes I use LSK389C @ TO-71 for earlier version "Су-XXI v1...v3" [ https://www.patreon.com/posts/vinilkorrektor-0-59167403 ] without DC servo but with Bi-polarized “nichicon MUSE” acoustic series instead @ output and ground of NFB loop [ https://www.patreon.com/posts/datashity-xxi-60583437 ] . You can see 2 green barrels here on PCB [ https://www.patreon.com/posts/pervye-nashikh-69590035 ] . The JFE2140 is better than LSK389 not only with much lower offset but also lower input capacitance and price. PS. Thanks for your attention to my project!
 
Last edited: