ChrioN

I'm toying arond with a precision dc circuit that requires something like a jfet op amp -
choppers and such are out of the question.

I've searched around and found nothing. How hard could it be?
Vos at unity gain is basically: Voltage at the output of the opamp - voltage at the input of the op amp.
Atleast in this case, because I don't have to worry about ibias and such.
So lets say we built a simple circuit with a couple of op amps in order to determine that difference.
We now know the offset voltage Then what? Input and output impedance should be kept alone.
Maybe applying a variable voltage to the feedback node? I have no clue.
« Last Edit: June 26, 2020, 12:38:18 PM by ChrioN »
"All PCB traces are curved to emulate the electrical response of point-to-point wiring" -Drip


Matador

I think you are looking for a 'DC servo':  for example:

https://sound-au.com/articles/dc-servo.htm

abbey road d enfer

I'm toying arond with a precision dc circuit that requires something like a jfet op amp -
choppers and such are out of the question.

I've searched around and found nothing. How hard could it be?
Vos at unity gain is basically: Voltage at the output of the opamp - voltage at the input of the op amp.
Atleast in this case, because I don't have to worry about ibias and such.
So lets say we built a simple circuit with a couple of op amps in order to determine that difference.
We now know the offset voltage Then what? Input and output impedance should be kept alone.
Maybe applying a variable voltage to the feedback node? I have no clue.
This is the exact description of a DC servo, as Matador said.
https://seventhcircleaudio.com/assets/J99B/J99BR11/docs/j99b_sch-194077baab99b50340100df094a2b74bacd4b939170c573984c975dc9930d807.pdf
Check what both sections of U3 do. One compensates offset for null output offset, the other for null input offset.
Who's right or wrong is irrelevant. What matters is what's right or wrong.
Star ground is for electricians.

moamps

This is the exact description of a DC servo, as Matador said.
https://seventhcircleaudio.com/assets/J99B/J99BR11/docs/j99b_sch-194077baab99b50340100df094a2b74bacd4b939170c573984c975dc9930d807.pdf
Check what both sections of U3 do. One compensates offset for null output offset, the other for null input offset.

Interesting design (with a drawback, IMO), an implementation of Jensen twin servo with another DOAs. As I see it, first servo sets the zero DC at output of the first DOA, second is used for second DOA. Both DOAs work as DC amplifiers so a servo is needed (two are maybe overkill).
The problem I see is that two decks of the switch should be BBM, last should be MBB. But all decks are BBM.

JohnRoberts

I'm toying arond with a precision dc circuit that requires something like a jfet op amp -
choppers and such are out of the question.
There are modern precision op amps (without choppers) that deliver good DC performance.
Quote
I've searched around and found nothing. How hard could it be?
How hard did you search...? I wrote about DC servos back in the 80s.
Quote
Vos at unity gain is basically: Voltage at the output of the opamp - voltage at the input of the op amp.
Atleast in this case, because I don't have to worry about ibias and such.

? with NF voltage at output should roughly equal + input voltage +/- input offset voltage.
Quote
So lets say we built a simple circuit with a couple of op amps in order to determine that difference.
differential..? but that differential amp can also have errors.
Quote
We now know the offset voltage Then what? Input and output impedance should be kept alone.
Maybe applying a variable voltage to the feedback node? I have no clue.
There is a well vetted simple servo approach using high quality film caps and op amp to buffer impedances. Perhaps moot in light of modern op amps with improved DC precision. (last time I used a DC servo was in the 80s but at the time it was a merchantable feature.)

I hope this helps.

=======
{TMI} now for today's TMI... My criticism of the DC servo(s) schematics posted is the topology. While this gets a little esoteric, I prefer a topology using a passive LPF RC feeding the servo op amp input. This way it is impossible to slew limit the servo op amp with step impulses. Even if the servo is fed from an existing circuit output, that output can be exposed to RF and high edge rate interference.  [/TMI]

JR

 
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

Interesting design (with a drawback, IMO), an implementation of Jensen twin servo with another DOAs. As I see it, first servo sets the zero DC at output of the first DOA, second is used for second DOA. Both DOAs work as DC amplifiers so a servo is needed (two are maybe overkill).
The problem I see is that two decks of the switch should be BBM, last should be MBB. But all decks are BBM.

I've always thought that the twin servo thing was complete overkill, the name sounds great thou "Twin Servo"

ChrioN

Quote from: JohnRoberts
There are modern precision op amps (without choppers) that deliver good DC performance.

There definitely are! But I want to be cocky for now and say even they're not good enough.

Quote from: JohnRoberts
How hard did you search...?

Again, not quite good enough I suspect. DC servos I off course know of, but I was thinking maybe there was another way (and for sure there are if you want to involve digital solutions. But I find elegant analog ones to be the most elegant.

Quote from: JohnRoberts
? with NF voltage at output should roughly equal + input voltage +/- input offset voltage.

I think we are saying the same thing.

Quote from: JohnRoberts
differential..? but that differential amp can also have errors.

I was thinking of a good chopper for that part.
"All PCB traces are curved to emulate the electrical response of point-to-point wiring" -Drip

JohnRoberts

There definitely are! But I want to be cocky for now and say even they're not good enough.
I can't guess how good you need...
Quote
Again, not quite good enough I suspect. DC servos I off course know of, but I was thinking maybe there was another way (and for sure there are if you want to involve digital solutions.
I once considered using flash based microprocessors to read a DC offset and drive DPOTs to correct. Unclear how often you want to update. For low noise analog designs you may not want a micro running in the background. 
Quote
But I find elegant analog ones to be the most elegant.
Elegant is not a technical term.
Quote
I think we are saying the same thing.

I was thinking of a good chopper for that part.
I haven't looked at choppers for a long time and never used one myself. There is some concern about them leaking switching noise, but I suspect modern ones are quite good.

Perhaps share what your application is and design criteria?

JR
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

ChrioN

Coming back to DC-servos - I always looked at them as something you used when you didn't want big dc offsets in your ac waveform = not something you use in dc circuits. I have very limited knowledge in these, I think I'll have to look them up further.


Quote from: JohnRoberts
I can't guess how good you need...[/qoute]
I can't say because I don't know yet. It would be nice to come close to a choppers 0.5-5uV.

Quote from: JohnRoberts
For low noise analog designs you may not want a micro running in the background. [/qoute]
If you do you're pcb layout right and use isolation ics, they're hardly noticable. Of course this becomes harder as circuit complexy grows.

Quote from: JohnRoberts
Elegant is not a technical term.
Typo, otherwise a wierd because the word comes up two times in a short sentence.

Quote from: JohnRoberts
I haven't looked at choppers for a long time and never used one myself. There is some concern about them leaking switching noise, but I suspect modern ones are quite good.

They are not quite good, some of the LT/AD ones are a amazing!

Quote from: JohnRoberts
Perhaps share what your application is and design criteria?

Basically an analog front end.
"All PCB traces are curved to emulate the electrical response of point-to-point wiring" -Drip

Coming back to DC-servos - I always looked at them as something you used when you didn't want big dc offsets in your ac waveform = not something you use in dc circuits. I have very limited knowledge in these, I think I'll have to look them up further.

DC servos are useless in DC circuits because the whole point of a servo is getting rid of the DC component, think of a servo as an alternative to a coupling capacitor. If you want to use something with a low offset without using chopper amps, you should try something like the LT6018 or LT1115 with 50uV offset voltage, thats basically as low as it gets without using choppers, also the noise voltage density is a stellar 0.9 nV/rootHz.  Do you need more DC precision than this? what for?

You mentioned you need FET opamps, is input bias/offset current an issue? are you going to be using medium to high resistances? you haven't told us your application.  FET opamps usually have much higher offset voltage than bipolars but better bias/offset current, the LT1112 has a slightly higher offset voltage (60uV) but a much lower input current 250pA, with its 0.3V/usec SR and 750 KHz. GBP it is a DC rather than audio opamp.  Is the DC component relevant in your design or not? if not, then a DC servo will do, but still, you have to take into consideration the offset voltage of the servo, how much offset voltage is allowed/tolerable at the output? all these questions would have more meaning if you tell us what you want to do.
« Last Edit: June 28, 2020, 09:42:10 AM by Dualflip »


Coming back to DC-servos - I always looked at them as something you used when you didn't want big dc offsets in your ac waveform = not something you use in dc circuits. I have very limited knowledge in these, I think I'll have to look them up further.


 'Designing Audio Power Amplifiers". by Bob Cordell has a chapter on the design and implementation of DC servos that's worth a read.
Covers the basics and also goes into more advanced schemes such as adding a 2nd pole.   



 'Designing Audio Power Amplifiers". by Bob Cordell has a chapter on the design and implementation of DC servos that's worth a read.
Covers the basics and also goes into more advanced schemes such as adding a 2nd pole.

+1 on Bob's book, a great one indeed!

+1 on Bob's book, a great one indeed!

This. 

 :)

clintrubber

Again, not quite good enough I suspect. DC servos I off course know of, but I was thinking maybe there was another way (and for sure there are if you want to involve digital solutions. But I find elegant analog ones to be the most elegant.

Curious to your application, or to which specs that need to be met.  Or an exercise in how far things can be brought?

I patented a 'digital' system for offset reduction in multi-channel amps, in the presence of signal.
In essence a digital DC-servo if you will, but no uP.
Intended for integration; for a discrete realization it's obviously unwieldly. 


 

Related Topics

  Subject / Started by Replies Last post
11 Replies
3104 Views
Last post October 20, 2004, 02:47:38 AM
by Samuel Groner
4 Replies
3922 Views
Last post February 11, 2005, 03:20:38 PM
by volta
1 Replies
1590 Views
Last post November 28, 2005, 02:35:42 AM
by PRR
5 Replies
2838 Views
Last post July 10, 2006, 02:43:58 AM
by gyraf