[Slightly different - threads ?] Let's...

Hi,

How about a Meta-sticky-something for bringing up different circuit topologies
one has encountered or made up oneself ?

Most circuits are combinations of the well-known snippets (long tailed pairs,
current mirrors, attenuators, etc etc), but now and then some circuit pops up that
makes one looks twice. It can be elaborate or simple, the latter even
more desirable since these are often the most elegant.

Imho it'd be interesting to post them discuss them and check/learn/discuss why
the designer did it differently... wasn't there a 'usual' approach possible ?
What's the benefit of the variant ? Or was it done simply to be different,
perhaps even at the cost of being worse than the usual way ?
Or hey, that's nothing special at all ! Or: all true, but old hat !

All will be fine.

To start & clarify this all a bit further, I've re-titled the 'single node servo'-
thread from a while ago as #1 in the series:

http://www.groupdiy.com/index.php?topic=293247


And here's #2:

http://www.groupdiy.com/index.php?topic=27049


Just an idea, let's see if there's interest.

As I did for #2, might be best to put new ones in the Drawing Board.


Cheers,

Peter

That would be a great Meta, one that I would find very enjoyable to browse.

PRRs Different Differential is a good one.

http://www.groupdiy.com/index.php?topic=4510&highlight=different


As a side note, it is good to see that people are getting some use out of the microphone schematic archive I put up. I got ALOT more schematics that where sent to me during my 6 or so month hiatus from computers that will be going up in a day or two, just need to finish converting the schematics to more proper formats.

adam

[quote author="adamasd"]That would be a great Meta, one that I would find very enjoyable to browse.[/quote]
Hi Adam,

Nice, thanks for the response. How can I make it a Meta ? Or does a moderator need to do that ?

PRRs Different Differential is a good one.

http://www.groupdiy.com/index.php?topic=4510&highlight=different

Click to expand...

That's sure a nice entry.

I'd like to label the entries and this one would classify for #0 I'd say (the mother of all... :wink: ), but I doubt if that remains doable (if there will be many new entries coming in that is...)
But let's try. If anyone sees an unusual topology or a circuit(section) that does something remarkably different yet elegant and he/she remembering this thread then look up the last entry-number, do +1 & post.

As a side note, it is good to see that people are getting some use out of the microphone schematic archive I put up. I got ALOT more schematics that where sent to me during my 6 or so month hiatus from computers that will be going up in a day or two, just need to finish converting the schematics to more proper formats.

adam

Click to expand...

Round circle, yes, got the Oktava from your website, thanks :thumb: Looking forward to those new files, but take your time.

Bye,

Peter

It sort of looks like no one else is interested in this idea, which is a shame, its one of the best ideas I have seen here in awhile.

To make a meta I think you just copy the header info from an existing meta. Open a meta and go to View Page Source then copy the code from the beginning of that one to the first message of this thread.

Even if no one else is interested I will certainly keep posted links to this thread, maybe interest will be build over time.

website is being updated as we speak, new files will be available in 10 or 15 minutes. There will be more stuff coming up in the next couple weeks as well, still have a fair amount of files to sort through.

adam

Perhaps the problem is the possible scope and then organization of such an effort.

I view circuit design a lot like writing. Components are like the words with caps and resistors like nouns and adjectives, active devices could be verbs. The laws of physics, would be our grammar (albeit better enforced) defining our sentence structure.

To start designing circuits we need more than a list of words, or dictionary with words and definitions, we need a sense of what particular sentences (circuits) mean.

To learn how to write, we first read and study the writing of others. Circuit design is no different where we study the design (writing) of others. We do this not so we can plagiarize these sentences, but so we can understand the power of certain component combinations for future applications.

When I was building my personal vocabulary I voraciously studied every schematic I could get my hands on. The ideas for design in the electronic magazines were a favorite until I noticed them starting to recycle old design ideas. I still take pleasure from a good read (schematic).

The human brain is very good at organizing such diverse piles of information. An encyclopedic glossary of such circuit elements or building blocks could be useful if we agreed to a coherent indexing method. I wonder if the PTO hasn't come up with a search engine for going through their data base of invention schematics looking for similar constructs. Say something like search all "NPN+diode+PNP", etc.

JR

PS: On the specific topic of audio-DC servos, while I feel these are generally misunderstood (they don't really remove caps from the audio path), they can be useful.

**I used one in a phono preamp where the DC error at very output was used to vary a current source loading the very first input stage.

** I got a patent for a class A VCA approach that used a servo to trim out DC offset so VCA could be built without needing selected parts or manual adjustment in production.

PPS: On the specific topic of differentials, a few that come to mind are.

** 3 transistor differential. Third transistor can work against other two to manage net DC of two differential audio outputs managed by other two. Two of the transistors were for the + and - audio input, while the third transistor was for DC reference versus the other two... Another use, the third transistor could just be held static at average of other two, to just reduce the transconductance of the working pair while delivering more total output current when saturated (perhaps to make a higher slew rate opamp input stage).

** a 2 transistor differential but with a somewhat unusual application was used as a bicolor LED level/threshold indicator (for an old CX record decoder I designed that required user level calibration). The bicolor LED was one of the 3 leg types. The differential pair steered current between the red and green LEDs for above and below threshold, with both colors active at threshold. A simple intuitive visual feedback for level setting.

** another odd 2 transistor differential was used to self regulate class A current in a class AB power amp output stage. By putting a resistor in series with only one of the two differential transistors, a well defined input offset voltage could be established. Another relationship that could be used is put two transistors in parallel for one side of the differential pair, This established a 6 dB difference in current density, which also translated to a predictable nominal input offset voltage following the log relationship for Vbe. That offset voltage was used to set nominal bias current of output stage. This benefits from precision devices like a transistor array.

*** a N transistor differential. This was a "greatest of many" detector. Visualize 16 transistors with all emitters connected together and only one resistor to supply. Each transistor has an LED in it's collector leg. Which ever transistor dominated the compound differential captured all the current and lights up just it's one LED. This is the guts of the 1/3rd or 2/3rd octave graphic EQ feedback indicating LED drivers where the loudest bandpass lights up (also patented).

--- notable differentials I recall from others
** I recall a phono preamp (by Tomlinsom Holman) where he crafted a differential pair using one FET and one bipolar transistor (non-inverting gain stage front end).

** I also recall an obscure low noise design (from Marshal Leach) that used an emitter connected NPN + PNP. Kind of like PRRs cite but without the degeneration. It was powered from a single 9V battery with as I recall a swinging power supply. I don't think it was ever reduced to a commercial product.

I have already posted some of my examples noted, in other threads. There are surely many many many more worthy cites from others so this would indeed be an encyclopedic effort. perhaps worthy of a Wiki, if not already done. A black hole for available time.

[quote author="JohnRoberts"]
** I also recall an obscure low noise design (from Marshal Leach) that used an emitter connected NPN + PNP. Kind of like PRRs cite but without the degeneration. It was powered from a single 9V battery with as I recall a swinging power supply. I don't think it was ever reduced to a commercial product.
[/quote]

See a recent allusion to a modified version of this in here, in a thread exchange with Samuel G.

Paul Rossiter wrote up a "new and improved" version of the Leach circuit for audioXpress a while back. It's a very high-quality piece given the highly variable standards of aXp, although I was driven to contact him about some difficulties with one part of the analysis. The floating supply is essential; interestingly, it doesn't matter what its impedance is---a current source works the same as the specified battery!

This would be an evil circuit to throw at someone in a job interview or final oral exam.

Some of the improvements he credits to suggestions from his friend and ex-professor Ed Cherry, author with Hooper of the highly recommended Amplifying Devices and Low-Pass Amplifier Design, which has among other things a unified treatment of hollow- and sand-state devices. It is also infamously expensive---you would think by now that more ex-library and estate copies would be showing up, as it dates to 1968!

Paul told me Ed would be pleased that his book was still being read.

+1

I am a fan of Cherry's work. His several AES papers on amplifier design are must reads for those interested in such things. A text from him should be a good read.

I regret I have slacked off on my pursuit of design with little critters. Nowadays when we pay more to place the little buggers than they cost, it's more effective to push digital 1s and 0s around.

JR

John, we need to get you back from the dark side of the force.

I'll send you that article anyway.

[quote author="adamasd"]It sort of looks like no one else is interested in this idea, which is a shame, its one of the best ideas I have seen here in awhile. [/quote]
Thanks, the idea just popped up here. But the real thanks of course to the people that 'caused' those circuits that gave rise to the idea.

To make a meta I think you just copy the header info from an existing meta. Open a meta and go to View Page Source then copy the code from the beginning of that one to the first message of this thread.

Click to expand...

It may not be as simple in the end, but I just compared a few and got the impression it's nothing more than just putting META in the thread-title.

Even if no one else is interested I will certainly keep posted links to this thread, maybe interest will be build over time.

Click to expand...

:thumb:

website is being updated as we speak, new files will be available in 10 or 15 minutes. There will be more stuff coming up in the next couple weeks as well, still have a fair amount of files to sort through.

adam

Click to expand...

I browsed it, it's getting nicer & nicer :sam:

[quote author="JohnRoberts"]Perhaps the problem is the possible scope and then organization of such an effort.[/quote]
Nice connection with writing you made :thumb:

I have already posted some of my examples noted, in other threads.

Click to expand...

If these come to mind again please feel free to link to them here. And the list you gave of circuit-examples looks nice as well - providing rainy Sunday afternoons are in abundancy at where you live, q&d drawings would be interesting additions.

There are surely many many many more worthy cites from others so this would indeed be an encyclopedic effort. perhaps worthy of a Wiki, if not already done. A black hole for available time.

Click to expand...

:wink:
Would indeed be an endless endeavour, something nobody over here should be wanting to do I'd say. It's not realistic to go for complete here, but each entry is a +1, and that'd already be nice.

PS: On the specific topic of audio-DC servos, while I feel these are generally misunderstood (they don't really remove caps from the audio path), they can be useful.

Click to expand...

I don't regret that you bring this up... I recall a discussion about this here a while ago, it may have even been 'at the old place'. I got involved up to the point that a battle about is/isn't in the circuit-path was about to begin, but since isn't was the opinion of a known & respected micpre-designer I didn't took the risk of being thrown out :wink:
As an alternative I concluded that the more or less blind respect of at least a part of the population wasn't founded on firm reasons. But enough about that.



Regards,

Peter

is

[quote author="JohnRoberts"]is[/quote]
:?:

[quote author="clintrubber"][quote author="JohnRoberts"]is[/quote]
:?:[/quote]

[quote author="clintrubber"]
[quote author="JohnRoberts"]
PS: On the specific topic of audio-DC servos, while I feel these are generally misunderstood (they don't really remove caps from the audio path), they can be useful.[/quote]

I don't regret that you bring this up... I recall a discussion about this here a while ago, it may have even been 'at the old place'. I got involved up to the point that a battle about is/isn't in the circuit-path was about to begin, but since isn't was the opinion of a known & respected micpre-designer I didn't took the risk of being thrown out [/quote]

With servos there still is a capacitor in the audio path, albeit buffered by an opamp.

JR

Agree.

In a similar sort of misconception I have encountered a number of folks who somehow think series elements as opposed to shunt elements in a circuit are more suspect.

Servos are just a sneaky way to do a highpass filter. For a given circuit you may be able to employ a smaller and maybe higher-quality capacitor(s) than the one(s) that would otherwise be inserted in the signal path, and maybe that allows you to have a more consistent response at the circuit output node with respect to loading effects*. But you have to look carefully at the noise from the error amp and integrated thermal noise from the resistor(s).

What occurred a while back and I've never seen, but may well have been done: have a sample-hold for the needed correction voltage. Update at prolonged times of zero signal, unless another comparator detects something outside of a permissible window. If one uses an infinite-hold-time S/H (a/d to memory to d/a) you don't have to worry about hold capacitor droop.

Then one could truly say that the servo is out of the signal path most of the time.


*but there is still an effect from loading, since the servo system loop gain will be affected.

[quote author="JohnRoberts"][quote author="clintrubber"][quote author="JohnRoberts"]is[/quote]
:?:[/quote]

[quote author="clintrubber"]
[quote author="JohnRoberts"]
PS: On the specific topic of audio-DC servos, while I feel these are generally misunderstood (they don't really remove caps from the audio path), they can be useful.[/quote]

I don't regret that you bring this up... I recall a discussion about this here a while ago, it may have even been 'at the old place'. I got involved up to the point that a battle about is/isn't in the circuit-path was about to begin, but since isn't was the opinion of a known & respected micpre-designer I didn't took the risk of being thrown out [/quote]

With servos there still is a capacitor in the audio path, albeit buffered by an opamp.

JR[/quote]

Ooops, silly me... Yes, IS ! :green:

[quote author="bcarso"]
What occurred a while back and I've never seen, but may well have been done: have a sample-hold for the needed correction voltage. Update at prolonged times of zero signal, unless another comparator detects something outside of a permissible window. If one uses an infinite-hold-time S/H (a/d to memory to d/a) you don't have to worry about hold capacitor droop.

Then one could truly say that the servo is out of the signal path most of the time.
[/quote]

precisely what I have been postulating for some time as well. has a digital servo been implemented in a commercial audio product yet? you can have a programmatic time constant. for example, very fast at startup to speed up settling time, then slower, and finally it can freeze until it needs to act again. one advantage of this is that highly asymmetrical waveforms won't make the servo react and will therefore be theoretically reproduced more faithfully. stacatto oboe, vocals, etc, could make a traditional analog servo "bounce" (though I havent tried to verify any commercial designs that do this, I believe most would)

mike p

While I doubt it would be embraced by audio phools I have been thinking about merging my small micro processor work with a tweaked out mic preamp. Things like DC offset voltage trims, CMR trims, input stage current density (for low noise), etc... could arguably be performed as offline operations at start up or periodically during prolonged idle times.

I would be inclined to apply some of these features interactively with the operator. Things like optimizing noise for a given microphone would require a learn mode and could use multiple memories to save and restore presets. I could even imagine benefits from tweaking CMR for each mic to account for real world variance.

I find some of this academic when you realize you will never encounter a microphone and/or following path as pristine, but it is a fun exercise to take one finite circuit block to a new higher level at least in principle.

Unfortunately such an effort may not resonate with a high end that expects dramatic audible improvements. Taking something from .001 to .00001 is generally not that dramatic. Dramatic audible changes usually don't come from improved linearity.

JR

Note: I am inclined to go ahead and have the preamp perform the A/D operation so I can ignore what happens after that conversion. Of course that raises the degree of difficulty since I haven't figured out how to roll my own uber convertor (yet).

[quote author="mikep"][quote author="bcarso"]
What occurred a while back and I've never seen, but may well have been done: have a sample-hold for the needed correction voltage. Update at prolonged times of zero signal, unless another comparator detects something outside of a permissible window. If one uses an infinite-hold-time S/H (a/d to memory to d/a) you don't have to worry about hold capacitor droop.

Then one could truly say that the servo is out of the signal path most of the time.
[/quote]

precisely what I have been postulating for some time as well. has a digital servo been implemented in a commercial audio product yet?[/quote]
Not that I know of, but it's at least been patented:

edit: shameless link removed, has been up for long enough now


Regards,

Peter

"has a digital servo been implemented in a commercial audio product yet?"

I think there have been hints of this from time to time---maybe a Krell piece a while back?

As far as JR's concern that a dramatic difference wouldn't be evident from uC controlled real-time tweaks, I would agree---but to a certain extent having The Story is enough to get reviewers excited and listening. I think studio users are generally a bit more discerning, and would welcome the interactive features.

Speaking of Krell, the remarks about somehow series components being more audible and deleterious than shunt come up when reading some of their latest ad and review copy about equipment touting current drive outputs. The claim is now there's no concern about cable loss (not that there is ever much to begin with in home audio unless something is terribly wrong!---but that's another story) because the current drive takes out all the series impedance. What's not mentioned is the effect now of the shunt impedance.

Also it's very hard to make a really good current output stage. And for other uses, feeding multiple destinations is not easy without some very odd input stages.

Of course the industrial control folks are quite familiar with using current as the variable, with the 4-20mA loop stuff out there for years and years, so the idea is also hardly new.

[quote author="bcarso"]And for other uses, feeding multiple destinations is not easy without some very odd input stages.[/quote]
In principle the system in the link above is able to do that, see Fig.3.