DC Servo question... - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Chip Amps

Chip Amps Amplifiers based on integrated circuits

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 22nd August 2007, 01:17 PM   #11
diyAudio Member
 
Onvinyl's Avatar
 
Join Date: Aug 2002
Location: Germany
Hi Darkfenriz,
this is a good point. How do you know, I did not find it in the datasheet.
Is the AD820 also known for this effect?
Rüdiger
__________________
"I can feel what's going on inside a piece of electronic equipment. I have a sense that I know what's going on inside the transistors." Robert Moog
  Reply With Quote
Old 22nd August 2007, 01:35 PM   #12
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Hi Dx,
Post4 shows an inverting DC servo.
It has no filter on the input so the opamp has to amplify the whole audio and sub audio signal to extract the wanted sub-audio correction.

The circuit as shown also has no filter on the output feeding into the NI input of the amplifier. This allows any noise from the opamp/PSU to be injected into the power amp input without any attenuation.
In addition any errors that the opamp makes in extracting that sub-audio correction also get injected into the amplifier input.

Scrap that example and use one that filters both the input to the servo opamp and filters the output before injecting it back into the power amp.

Both the inverting servo and the non-inverting servo work. I believe it is easier to get the non-inverting servo to work well.
Gootee's example uses opamp2b to extract the sub-audio correction. Then opamp 2a applies active filtering to attenuate any remnants of unwanted audio and distortion and noise.
However this circuits works as a DC servo with the active filter wrapped around opamp2a removed. Then just use the passive output filter.
R8 & C4 are the input filter passing below 1Hz.
R13 & C13 are the output filter passing below 3Hz, if the second opamp is omitted. retain R1=R13.
This passive output filter allows slightly more audio through to the amplifier, but it is at a very low level and if designed correctly should be inaudible.

Hope this helps.
__________________
regards Andrew T.
  Reply With Quote
Old 22nd August 2007, 02:26 PM   #13
diyAudio Member
 
Onvinyl's Avatar
 
Join Date: Aug 2002
Location: Germany
Hi Andrew,
don't you think R5/C4 in Post4 does the input filtering?
And output is filtered by the voltage divider consisting of R6 : R2||Rin?
If not, why?
Regards,
Rüdiger

EDIT: Meanwhile I found the other thread where gootee explains his servo. My Question regarding 'why an acitve low pass filter' is answered
__________________
"I can feel what's going on inside a piece of electronic equipment. I have a sense that I know what's going on inside the transistors." Robert Moog
  Reply With Quote
Old 22nd August 2007, 04:02 PM   #14
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally posted by Onvinyl
@gootee: what advantages do you see in having an active low pass filter after the integrator?

Rüdiger
Hi Rudiger,

I think that probably the only advantage is better/steeper filtering, so that almost no audio frequencies can get through and get back to the amplifier input.

- Tom

Edit: Sorry. I must have been typing while your last message was posted. For others who might be interested, that other DC Servo thread is here:

PA100 with DC SERVO?
  Reply With Quote
Old 22nd August 2007, 04:15 PM   #15
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally posted by Dxvideo
<snipped>

But in Gootee's example, he uses two stages DC servo, while I coulnt understand one stage DC servo, how could I adapt it to my s h i t t y preamp project?????

More help for that poor diyer pls...
Hi Dxvideo,

The "second stage" is only a lowpass filter.

You could use a simple RC passive lowpass filter, there, instead of the second opamp etc. But if you use a dual opamp, it might be worth using the other half for the filter.

For a passive RC filter, you might want to use two resistors, with a capacitor to ground between them.

- Tom Gootee

http://www.fullnet.com/~tomg/index.html
  Reply With Quote
Old 22nd August 2007, 06:53 PM   #16
Electrons are yellow and more is better!
diyAudio Member
 
peranders's Avatar
 
Join Date: Apr 2002
Location: Göteborg, Sweden
Blog Entries: 4
Quote:
Originally posted by AndrewT
...

Remember that the servo only NEEDS to handle frequencies below audibility.
This is not quite right. In fact the servo must work extremely good up 1 kHz which isn't a problem for most opamps.

If you don't are convinced, do a SPICE similation in subject.
__________________
/Per-Anders (my first name) or P-A as my friends call me
Super Regulator SSR03 Group buy
  Reply With Quote
Old 22nd August 2007, 09:13 PM   #17
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Quote:
Originally posted by Onvinyl
don't you think R5/C4 in Post4 does the input filtering?
I don't think so.
the cap, C4, allows the opamp to work as an integrator, i.e. a constant 6db/octave slope. But there are limits to that slope.
The resistor, R5, limits the current fed into the integrator.
The non-inverting version uses the same opamp & cap combination to achieve the same slope again with limits.

Quote:
Originally posted by Onvinyl
And output is filtered by the voltage divider consisting of R6 : R2||Rin?
If not, why?
no, the resistors are a dividing ladder. The cap to filter out the higher frequencies is omitted. However the cap, C2=56pF, will have a low pass filtering effect, but attenuating below 60kHz, rather than 1Hz or 2Hz.

Can someone enlighten us on what the relative low pass frequencies should be to obtain best performance from the dual input/output filters either side of the integrating opamp?
Should the input filter be higher or lower than the output filter?
Where in the frequency band should these filters operate to minimise the effect on the wanted audio signal?
How low is too low to make the correction too sluggish in response to output offset errors?
__________________
regards Andrew T.
  Reply With Quote
Old 22nd August 2007, 09:23 PM   #18
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Quote:
Originally posted by peranders

This is not quite right. In fact the servo must work extremely good up 1 kHz which isn't a problem for most opamps.

If you don't are convinced, do a SPICE similation in subject.
Hi Peranders,
I can understand the requirement for the opamp to operate well at frequencies well into the audio band. Maybe much higher than 1kHz to ensure the integrator correctly creates that constant 6db slope.
However, the complete servo is only there to correct the slow drift in output offset and if working perfectly will have zero effect on the wanted audio frequencies. As is usual, there is compromise. The DC correction must be fast enough to minimise the output offset, both in extent and in time.
Yet, sufficiently filtered to avoid injecting some of the audio signal back into the amplifier.
This speed thing is what prompted my question in my previous post.

Summarising, the opamp must operate well at audio frequencies. The DC servo must pass the minimum of audio back to the amplifier correction node.

ps,
I don't need to be convinced, I am already a believer.
__________________
regards Andrew T.
  Reply With Quote
Old 23rd August 2007, 09:24 AM   #19
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally posted by AndrewT
I don't think so.
the cap, C4, allows the opamp to work as an integrator, i.e. a constant 6db/octave slope. But there are limits to that slope.
The resistor, R5, limits the current fed into the integrator.
The non-inverting version uses the same opamp & cap combination to achieve the same slope again with limits.

no, the resistors are a dividing ladder. The cap to filter out the higher frequencies is omitted. However the cap, C2=56pF, will have a low pass filtering effect, but attenuating below 60kHz, rather than 1Hz or 2Hz.

Can someone enlighten us on what the relative low pass frequencies should be to obtain best performance from the dual input/output filters either side of the integrating opamp?
Should the input filter be higher or lower than the output filter?
Where in the frequency band should these filters operate to minimise the effect on the wanted audio signal?
How low is too low to make the correction too sluggish in response to output offset errors?

Hi Andrew,

Those are good questions. (All of the rest of what I say might be wrong, but... ) I don't have all of the answers to those questions, offhand (and maybe none of the answers). But someone could just use the downloadable, ready-to-run LTspice simulation file for the DC Servo (or modify it for some other servo design) and find out, maybe by trying pre-filters with e.g. 2X and 1/2X of the post-filter's cutoff frequency.

Regarding where to put the overall cutoff frequency to minimize the effects on the audio output signal: It sure SEEMS like lower/slower would be better, in that respect. I think I tried a pre-filter for the integrator, a few years ago, but don't remember much about it, any more. It might add its own offset, though (which might be able to be compensated-for by changing the integrator's R (and C) values, IIRC).

Note that, in my DC Servo Ltspice simulation, the one that is downloadable from http://www.fullnet.com/~tomg/gooteesp.htm , one could click on the .op button and add a spice directive to calculate the THD of the amplifier output signal, such as ".four {freq} 9 20 V(OUT)" (without the quotes). Then, after a transient run, one would select Edit-->View Spice Error Log, to see the THD results. That way, it could be determined whether or not a change to the servo system had an effect on the THD of the output signal. (Note that, when using the .four command, it might be best to make sure that the transient simulation's run-time results in an integral number of signal cycles, and that the average offset is no longer changing when the last 20 cycles occur.)

Regarding what might be TOO low/slow, making the servo's correction-speed "too sluggish": I don't know that answer, either. But...

In order to not pass any audio frequencies, OR any sub-audio frequencies that would modulate the speaker-cones' centering positions (which I've read could also be "a bad thing", sound-wise), the servo system's response speed must be "very" slow. So maybe a better question would be "How fast would be too fast?". I don't know. More trade-offs, I guess.

My original design was a little faster than the current one, but also wasn't quite as stable (Faster-responding servos COULD be made more stable, but maybe only at the expense of letting more higher frequencies through.). The current version takes almost 9 seconds to fully-correct a 1-volt DC output offset of a 100 Hz 22v p-p sine output signal, but gets 90% of the way there in about 3.6 seconds. Of course, 1 whole volt of amplifier output DC offset would be quite excessive; "a lot"! That was just for testing/simulation purposes. But I think that the time-constant stays the same, basically, for any reasonable DC offest value and signal frequency.

One would hope that the only time that the SPEED of the correction would really matter would be at power-on time. And if an amplifier has excessive offset at power-on, then maybe it would be a good idea to have a delayed turn-on for the speakers, giving the servo system time to get the DC offset down by 90% or so, first.

One other possibility (remote, I hope), where response-speed might matter, and slower-yet might be better, which I have not really investigated, might be when a large "thump" of very-low-frequency signal (or maybe even just the beginning of a VLF continuous "tone") hits the output, maybe especially if the output was relatively quiet, just before that (maybe such as at the beginning of a new song?! AK!!): If the frequencies were low-enough, relative to the servo's response speed (and MAYBE also if the change in amplitude were large-enough, even for higher frequencies?!), such an event could cause a transient change in the output of the servo system, causing a transient change in the DC offset of the amplifier output. That might be able to affect the sound from the speakers, by modulating the centering locations of the speakers' drivers' magnets, and which, if it occurred, would last for the same 3.6 to 9 seconds or so, in my circuit's case. (But I don't know if the magnitude of this effect might be significant-enough to worry about, or whether or not it might be easy to make it so we wouldn't HAVE to worry about it.)

[Edit: When it's late, and I'm tired, I tend to "ramble"... ] I guess that if I were really serious about trying to make a DC servo system that would respond fast-enough to large offsets, but was as inaudible as possible in the audio band when the offset was well-controlled, I might try to use an adaptive filter (and/or maybe even integrator) design, that would lower its own time-constant when the error was larger, and raise its time-constant when the error was smaller. The time-constant could be changed using a voltage-controlled resistance. (An FET circuit could be used for that part. But I often like to use a Vactrol, which is basically just an LED encapsulated with a photocell. You could even make your own. But I've usually used something like the VTL5C2 Vactrol, which changes its photocell's resistance from a few hundred ohms to about 2 Megohms when its LED current is varied from 40 mA to 0 mA.) It might be difficult to get the response speed to be VERY much faster than we already have it, unless the error signal could be derived in a faster way; maybe dual (+/-) peak detectors (with the same type of adaptive filters...)? But at least the lowpass filtering for the LOW-error condition could be WAY-maxed-out, while still retaining "at least" the existing response speed for larger-error conditions. (It CAN work very well: I used a similar technique for an adaptive filter for the output of a peak detector that was part of a feedback system that controlled the output amplitude of periodic signals in an LM1875-based audio-frequency amplifier in a measurement instrument, and was able to get the peak detector's output to slew up OR DOWN very quickly, when the amplitude was far from the desired level, such that I could make a 5v P-P input result in any desired output amplitude from 30V P-P to 1V P-P, stable within less than 200 ms, for any frequency from 60 Hz to 20 kHz, but was still able to get a ripple voltage of well-under 1 mV P-P at the peak detector's output once the error became small, even for 60 Hz sawtooth waveforms.)

Sorry to have blathered-on about all of that, for so long.

- Tom Gootee

http://www.fullnet.com/~tomg/index.html
  Reply With Quote
Old 23rd August 2007, 10:54 AM   #20
AndrewT is offline AndrewT  Scotland
diyAudio Member
 
Join Date: Jul 2004
Location: Scottish Borders
Quote:
Originally posted by gootee
...........Regarding what might be TOO low/slow, making the servo's correction-speed "too sluggish": ............. But.............One would hope that the only time that the SPEED of the correction would really matter would be at power-on time. And if an amplifier has excessive offset at power-on, then maybe it would be a good idea to have a delayed turn-on for the speakers, giving the servo system time to get the DC offset down by 90% or so, first..............
that sounds like support for the philosophy that the output offset should be minimised BEFORE activating the DC servo.

The question that follows is:-
would it be better to minimise offset at completely cold and let the DC servo zero out the warm up offset?
or
set the warmed up offset to zero and hope that the first turn on offset is low enough and short enough to be tolerable.

I prefer the first alternative.
But, what if one switches on a pre-warmed amplifier?
__________________
regards Andrew T.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Another DC servo question. Dxvideo Chip Amps 27 20th September 2008 05:38 PM
DC Servo question Dxvideo Chip Amps 27 15th July 2008 12:01 PM
To use DC servo or not? jaycee Solid State 40 17th September 2007 02:04 PM


New To Site? Need Help?

All times are GMT. The time now is 09:42 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2