| Stocker |
This is probably not going to generate much interest but here goes...
If you are designing an amplifier that has potential for DC offsets on the output;
And the offset can be trimmed out;
And you would like to keep it to (nearly) zero offset...
...and somehow you are motivated enough to design this circuit but not design away the offset voltage in your amp... ;)
How about a front panel LED that is an offset indicator? That is to say, when your output has a DC offset, there is an LED winking at you. Then a little knob beside the indicator trims the offset away until the LED goes out. I am thinking more along the lines of a headphone amplifier, but the concept works as well for a main system amp, no?
This would probably be a pretty simple cct to design using any run-of-the-mill comparator setup.
Comments? |
|
|
| millwood |
| would a DC servo driving an LED work? |
|
|
| sreten |
| Designing in a problem so you can design a solution is not a popular appoach ;) sreten. |
|
|
| sreten |
| quote: | Originally posted by millwood
would a DC servo driving an LED work? |
Better to use the DC servo to make offset a non-issue.
:) sreten. |
|
|
| boholm |
Yes, it is possible.
IMHO it is quite simple: any OPAMP with a healthy gain should do the trick. Example: If +/- 50mV is max. before indication (LED switches on) then the gain should be around 3V/50mV=60. I say 3V for the LED+resistor to switch on, but all of it should be experimented with, so that the indication comes in an acceptable maner. And if you use a bicolored LED then you will know which way to turn your trimmer.
I hope I make sense . . . and no, I havent tried it myself. Just a suggestion. |
|
|
| Stocker |
| quote: | Originally posted by millwood
would a DC servo driving an LED work? |
| quote: | Originally posted by Stocker
...and somehow you are motivated enough to design this circuit but not design away the offset voltage in your amp... ;) |
If you went to the effort to design in the DC servo, the whole thing would be unneccessary, n'est-ce pas? Although I suppose, If you are a gadget freak or tweaking is in your blood (surely no one here ! ) It could be another gee-whiz feature to show off...
| quote: | Originally posted by sreten
Better to use the DC servo to make offset a non-issue.
|
Yes, if you like a DC servo :D
Boholm: seems like if you are going to do it, do it all the way: go for <10mV if you are making the effort. Heck, go for 0 if you want. but yes, this was the idea. A high-gain opamp would definately be a must. A bicolor LED would be one way, and a great idea for amps with dual supplies, BUT it would always be on. If I did it I would go for a single LED so I could have the satisfaction of seeing it fade to black as I zeroed out the offset. :cool: maybe an integrator...
It would be :cool: :cool: if you put it behind a translucent face plate and you didn't see anything at all besides the adjustment knob when the LED winks out. |
|
|
| boholm |
| quote: | Originally posted by Stocker
Boholm: seems like if you are going to do it, do it all the way: go for <10mV if you are making the effort. Heck, go for 0 if you want. but yes, this was the idea. A high-gain opamp would definately be a must. A bicolor LED would be one way, and a great idea for amps with dual supplies, BUT it would always be on. If I did it I would go for a single LED so I could have the satisfaction of seeing it fade to black as I zeroed out the offset. :cool: maybe an integrator...
It would be :cool: :cool: if you put it behind a translucent face plate and you didn't see anything at all besides the adjustment knob when the LED winks out. |
Well, I just left a little headroom, so that it wouldn't be triggered too often if the amplifier had some fluktuations (spelling? meaning:D ) But yes, you could go so low - I would test it out before final assembly.
Why would it always be on?
And if you go for a single LED then you would only catch the DC going in one direction depending on how you soldered in your LED . . . . oh, maybe you actually mean two LEDs (how to say?) opposite coupled? (sorry my technical english - needs brushing off:blush: )
And the fading would look good, but it shouldn't fade too slow. Otherwise you could turn the trimmer too far and past 0. |
|
|
| janneman |
What you would want is an integrator that senses the DC, basically a servo but without the servo output used to null the offset (because you want to do that yourself, right). Then take this output and stick it into what is called a window comparator. Basically two opamps in open loop that switch to HI or LO depending whether the offset is below or above a threshold. Combine the two to drive a LED and you have a bi-directional offset indicator.
Look up window comparators in Walt Jung's opamp cookbook, or ask google to look it up on the net.
Jan Didden |
|
|
| thoriated |
| This is close to a scheme I described a few weeks back here using an LED to indicate offset & bias. For offset, the output would be fed into a full wave 'ideal' rectifier using op amps which then drives an op amp configured as a current source driving the LED whose brightness is approximately proportional to the current passing through it. You can, of course, add integration or a LP filter to the above to reject audio band energy. |
|
|
| Fred Dieckmann |
| Perfect. Concise with a good recommendation for changes and a good reference. Also going to a linear IC website and doing a search for comparator data sheets will most likely get a good circuit in the application notes. |
|
|
| janneman |
Fred:
Give a man a fish and he has a great meal.
Learn a man how to fish and he eats for the rest of his life.
Jan Didden |
|
|
| jwb |
| You gotta wonder, if you are going through the trouble of all those amps, comparators, etc etc etc why not just build the stupid servo? Worried about stability issues? |
|
|
| Stocker |
Originally posted by boholm
Well, I just left a little headroom, so that it wouldn't be triggered too often if the amplifier had some fluctuations
Excellent point.
Why would it always be on?
It wouldn't! I was thinking about another project I have been contemplating for a while! :o
if you go for a single LED then <snip>
Your idea of a bicolor LED seems like exactly what would be called for. That way you don't have to fool with an integrator at all.
And the fading would look good, but it shouldn't fade too slow. Otherwise you could turn the trimmer too far and past 0.
Well, you can always adjust it back... :D ...another excellent observation.
Janneman, I think Boholm nailed it with his one bi-color LED.
| quote: | | Originally posted by thoriated For offset,<snip> the LED whose brightness is approximately proportional to the current passing through it. |
If you set the thing up right a plain comparator would (might? )be able to vary the voltage so the brightness variance would be controlled by it.
| quote: | Originally posted by jwb
You gotta wonder, if you are going through the trouble of all those amps, comparators, etc etc etc why not just build the stupid servo? Worried about stability issues? |
Depending on the amp, perhaps. Then again some people seem to despise servos, and others just like to tweak. Nobody ever said this was the easiest, simplest, most cost effective, most automatic or most elegant solution available. Depending on how unstable our theoretical amp's offsets are, this could end up being a pain in the booty to keep adjusting...
It's just an idea... :) |
|
|
| janneman |
| quote: | Originally posted by Stocker
Janneman, I think Boholm nailed it with his one bi-color LED.
If you set the thing up right a plain comparator would (might? )be able to vary the voltage so the brightness variance would be controlled by it.
[/B] |
A comparator by definition is an on/off switch, controlled by a reference. For instance, you may decide you can tolerate 10mV offset, you would have a comparator (really an opamp without feedback) with one input at 10mV and the other input at the offset. Whenever the offset gets above the reference, the output of the comparator switches. Depending on which inputs (- or +) have the ref or offset, the switching goes pos or neg. So if you take two comparators with opposite polarity connections, and wire-or the outputs (combining them with two diodes) you have a window comparator of which the combined output switches ON if the offset is outside the window (+ or - 10mV in this example).
The example in the quoted part above is no comparator at all but a linear amp whose output follows the offset to drive a variable current through the LED. That's not what he wanted.
In any case, you need an integrator to extract the offset from the amp output, which has a huge AC component completely swamping the relatively tiny offset. So, you need 3 opamps (yes I know you can do it with 2, but let's leave it at this for now).
Jan Didden |
|
|
| Steven |
| quote: | Originally posted by janneman
In any case, you need an integrator to extract the offset from the amp output, which has a huge AC component completely swamping the relatively tiny offset. So, you need 3 opamps (yes I know you can do it with 2, but let's leave it at this for now).
|
Make it a damped integrator. A normal integrator has an (almost) infinite gain for DC. Remember in this case, i.e. not being a servo, there is no feedback, so the LED wil go on after a while, even with a fraction of a mV offset.
Steven |
|
|
| jwb |
| Does it even need to be an integrator? I'll show my ignorance by suggesting you could just low pass the amp's output with a f(-3dB) ~ 1Hz and pass that to your DC-detecting window comparator |
|
|
| janneman |
Indeed Steven, good point.
Jan Didden |
|
|
| janneman |
| quote: | Originally posted by jwb
Does it even need to be an integrator? I'll show my ignorance by suggesting you could just low pass the amp's output with a f(-3dB) ~ 1Hz and pass that to your DC-detecting window comparator |
No false modesty please!
Hmm yes, why not. The nice point of the integrator is its low output impedance to drive the comparators, but a low-pass filter for very low frequencies would need either a large cap, which may be relatively large and expensive, or a large R, which may influence the comparators. But other than that, yes sure.
Jan Didden |
|
|
| Stocker |
| quote: | Originally posted by janneman
A comparator by definition is an on/off switch, <snip>
The example in the quoted part above is no comparator at all but a linear amp whose output follows the offset to drive a variable current through the LED. That's not what he wanted.
In any case, you need an integrator to extract the offset from the amp output, which has a huge AC component completely swamping the relatively tiny offset. So, you need 3 opamps (yes I know you can do it with 2, but let's leave it at this for now).
Jan Didden |
:o that stuff on my face would be egg.
Thank you and you are of course right about the comparator/ linear amp business. I am being too sloppy with my terminology in this thread!
also, when you use a quad op-amp like a TL074 or LM324, there are 4 in one package... |
|
|
| hjelm |
| Wouldn't be a nice retro touch if you had a low pass filter and an analog voltmeter, + and - indication, instead of a LED? That way you have a relative feel for the actual offset and can trim it down. |
|
|
| moamps |
Dual LED offset indicator
+ offset red
- offset green
zero orange
set window with R9
Regards |
|
|
| boholm |
| I would prefer the LED not to light at all, when there is no DC. That's more logical to me. Agree or disagree? |
|
|
| Circlotron |
| If you need an indicator to tell you that you have offset because you cannot hear it... :confused: |
|
|
| janneman |
| quote: | Originally posted by moamps
Dual LED offset indicator
+ offset red
- offset green
zero orange
set window with R9
Regards |
That's the idea, but I think in your circuit the window is always +/- .6V, independent of the pot. That's a pretty wide window for offset. I think that if you want to make the window variable (although I see no need for it), you need two pots, each across the diode and feed the fractional diode voltage to the comparator inputs.
You may also want to add a small positive feedback across the comparator to introduce some hysteresis, to avoid oscillations around the switching point.
You may also want to put R7, R9 in series with the comparator outputs to limit the LED current.
Lastly, place diodes across the LEDs in opposite polarity to avoid destroying them with too much reverse voltage.
Jan Didden |
|
|
| Stocker |
| quote: | Originally posted by boholm
I would prefer the LED not to light at all, when there is no DC. That's more logical to me. Agree or disagree? |
Agree
| quote: | Originally posted by Circlotron
If you need an indicator to tell you that you have offset because you cannot hear it... :confused: |
haha... somebody found out the secret! :D |
|
|
| moamps |
Hi Jan,
thanks for yours correction.
| quote: | Originally posted by janneman
That's the idea, but I think in your circuit the window is always +/- .6V, independent of the pot. |
Drawing error. Corrected.
| quote: |
....you need two pots, each across the diode and feed the fractional diode voltage to the comparator inputs. |
Two pots aren't needed if window is symetrical.
| quote: |
You may also want to add a small positive feedback across the comparator to introduce some hysteresis, to avoid oscillations around the switching point.[/B] |
OK, but not so important. | quote: |
You may also want to put R7, R9 in series with the comparator outputs to limit the LED current. |
Led current is defined only with R7 and R9. In corrected schematic with R7+R10 and R9+R11.| quote: |
Lastly, place diodes across the LEDs in opposite polarity to avoid destroying them with too much reverse voltage. |
Done.
Regards |
|
|
| janneman |
Looks like a winner. Clever, that R9 stuff. There may be some assymmetry if the ref diodes are not matched, but that's acceptable I would think.
Jan Didden |
|
|
| peranders |
I'm sorry, but this won't work (I might be wrong).
1 LP-filter, preferable up to 3th or even 4th order, cut frequency 0.01-1 Hz. Those filters most also have low offset voltages. This is dependent how low DC voltage you want to detect, how advanced the filter should be.
2 Fullwave rectifier or two halfway rectifiers, one for positive and one for negative signals
3 Comparator with HYSTERESES!
4 Indicators
Filtering out a couple of millivolts from several volts or tenths of volts is a tricky business. Don't forget that DC is low frequency AC! |
|
|
| janneman |
Per-anders,
You may remember that this comparator stuff is preceded by an integrator like the servo loop normally used to null the offset. Please look up the previous posts.
Jan Didden |
|
|
| thoriated |
| Some amps may also DC offset slightly when producing a larger AC signal. |
|
|
| moamps |
| quote: | Originally posted by peranders
I'm sorry, but this won't work (I might be wrong).
|
Hi,
I have this on breadboard. Works reasonably good.| quote: |
1 LP-filter, preferable up to 3th or even 4th order, cut frequency 0.01-1 Hz. Those filters most also have low offset voltages. This is dependent how low DC voltage you want to detect, how advanced the filter should be.
|
Too complex.| quote: |
Fullwave rectifier or two halfway rectifiers, one for positive and one for negative signals |
Too many additional IC's.| quote: |
3 Comparator with HYSTERESES!
|
...will destroy windows margine.| quote: |
Filtering out a couple of millivolts from several volts or tenths of volts is a tricky business. Don't forget that DC is low frequency AC! |
What about simple (one IC) DC servos in amps? Doesn't look so tricky.
Regards |
|
|
| moamps |
| quote: | Originally posted by peranders
Don't forget that DC is low frequency AC! |
Oximoron?:D |
|
|
| peranders |
| quote: | Originally posted by janneman
Per-anders,
You may remember that this comparator stuff is preceded by an integrator like the servo loop normally used to null the offset. Please look up the previous posts.
Jan Didden |
OK, but even so the integrator must be extremely slow and/or you need extra filters. |
|
|
| janneman |
| quote: | Originally posted by peranders
OK, but even so the integrator must be extremely slow and/or you need extra filters. |
Not different from 'normal' servo loops. The fact that they are used in many 10.000's of amps is some indication that most probably it works. I concede that often the implementation isn't very good, but we have beaten that particular horse to death in this thread I would think.
Jan Didden |
|
|
| peranders |
| OK, I'll back off a little but if you take signal from the DC servo intergrator you will not get an exact measure of the offset. The servo output is only a signal inside a feedback loop and tell you not much really. |
|
|
| janneman |
| quote: | Originally posted by peranders
OK, I'll back off a little but if you take signal from the DC servo intergrator you will not get an exact measure of the offset. The servo output is only a signal inside a feedback loop and tell you not much really. |
You obviously have not taken my advice to read the former posts. The servo loop is not closed, because he wants to regulate the offset by hand. I said "an integrator LIKE the servo loop NORMALLY used to....". There is no closed servo loop and the integrator output is the offset voltage, at least above the low pass freq. And indeed that should be low enough so that there is no lf audio to slowly vary the apparent offset.
Can I go now?
Jan Didden |
|
|
| thoriated |
| If one is not sure the amp being nulled may not dc offset slightly in the presence of some musical or test signals, best to dc null with no input signal applied or with a mute function enabled at the same time. |
|
|
| peranders |
| quote: | Originally posted by janneman
Can I go now? |
yes, :) |
|
|
| thoriated |
| This is meant to be implemented with a quad op amp. The first section is a differential integrator which will probably change - I drew it this way because I wanted to be able to select this circuit as a bias monitor as well as a offset monitor circuit. The second and third op amp sections and associated components comprise a high speed 'ideal' full wave rectifier. The final op amp stage acts as a bipolar current source to linearly control the LEDs intensity. 'Off' would correspond to 0 volts offset, in this case. I threw in D5 and R16 to limit sensitivity more than several tenths of to a couple volts away from the null range. All component values hypothetical:) |
|
|
| Stocker |
| quote: | Originally posted by peranders
yes, :) |
LOL :D
guys you are making my day(s) by actually putting some effort into this whim of mine.
Thanks everybody! :wave2: :grouphug: |
|
|
|
