[tangent]

I'm reworking the article and going back over this thread, and there are a few things I don't understand from your first post, Jeffrey.

Quote:

the op-amp is in the inverting configuration

The reference is a positive voltage, and the output swings positive. Feedback is to the negative input. Where is the inversion?

Quote:

only the NPN transistor in the op-amp output stage is used to modulate the pass transistor

Doesn't the PNP sink current in the output stage of the op-amp? We turn the NPNs on in these headphone amps by pulling current from the op-amp's output.

 

[jeffreyj]

Quote:

Originally posted by tangent I'm reworking the article and going back over this thread, and there are a few things I don't understand from your first post, Jeffrey. the op-amp is in the inverting configuration The reference is a positive voltage, and the output swings positive. Feedback is to the negative input. Where is the inversion?

Sounds reasonable, but I think you are getting tripped up from the op-amp being powered by a single positive supply. You are used to thinking of inversion as a positive input voltage turning into a negative output voltage, but the true definition is that a rising input voltage causes a falling output voltage. The fixed input on the op-amp is the (+) one, so this must be an inverting amplifier, and, indeed, when the voltage applied to the (-) input rises, the output voltage falls (goes closer to ground).

In actuality, the output voltage of the op-amp changes very little; it is the current being sunk to ground that accomplishes regulation in this particular circuit, but I am trying to keep things clear.


Quote:

only the NPN transistor in the op-amp output stage is used to modulate the pass transistor Doesn't the PNP sink current in the output stage of the op-amp? We turn the NPNs on in these headphone amps by pulling current from the op-amp's output.

Mea culpa - you are correct. I totally flubbed that one up and can't imagine how I got the idea that an NPN would be appropriate as the closest it could come to the negative supply (ie - ground) would be one diode drop whereas a PNP would be able to come within a few millivolts (depending only on Vce[sat] which is proportional to Ic).

So, questioning "authority" has paid off once again, eh?

 

[tangent]

Quote:

the true definition is that a rising input voltage causes a falling output voltage.

In a noninverting AC amplifier, the output goes through the feedback network to the inverting input, where a rising signal relative to the noninverting input causes a falling output. Also, in a noninverting configuration, gain is (Rf/Rg)+1. Sounds exactly like what is happening here.

In an inverting amplifier, the signal to be amplified goes to the inverting input, and the noninverting input is tied to some reference.

I think you might be looking at the fact the +IN is tied to a reference, which is usually a practice seen most often in inverting amplifiers. But that's for the AC case, where you need a reference for the signal. In this DC amplifier, the reference is the signal.

Quote:

I totally flubbed that one up

I hope you don't think I'm taking shots at you now. I'm just asking questions for clarification.

 

[tangent]

Another confusion:

I'm reading through the fourth article in the 1995 series, and Galo recommends three changes to lower dropout voltage. I don't see why a single one of them helps.

1. Use a 2V green LED in the current source. Galo says this "enhances" the output swing of the op-amp but doesn't explain how. All I see is that the collector gets pulled about 0.4V above the base when the voltage across the pass transistor is 1V. This is not enough to forward bias the CB junction, but it's close. (For that reason, using a 2.2V or higher LED might be disastrous here.) I don't see what this buys. Another effect is that the current source's output increases unless you also increase R5, and Galo mention doing that. They can't just want a higher current through the CCS, since you can just change R5 to get that. Somehow pushing the base 0.4V further below the input helps things, but I'm at a loss as to why.

EDIT: This will also bring the C and E about 0.2V from each other. Perhaps they're trying to pinch the CCS off, which they can't do when the CCS is 0.4V closer to the input voltage? Why would that help?

2. Galo recommends lowering R4 to 10K to 12K.. This will increase current through the LED, and it drains base current away from the CCS transistor more easily. Somehow this is good for 0.1V or so of improvement.

3. Galo recommends using a transistor with a lower saturation voltage for the CCS transistor. This confuses me since saturation voltage goes up as collector current goes up, which it did due to the first change.

 

[jeffreyj]

Quote:

Originally posted by tangent In a noninverting AC amplifier, the output goes through the feedback network to the inverting input, where a rising signal relative to the noninverting input causes a falling output. Also, in a noninverting configuration, gain is (Rf/Rg)+1. Sounds exactly like what is happening here.

You are absolutely correct - the error amp by itself is in the non-inverting configuration; scratch my earlier statement, then, about the benefit of using the error amp in the inverting configuration.

It is only by the shunting of the pass transistor's drive current to ground that inversion of the output occurs, and *that's* what tripped me up. From a too broad perspective I looked at the circuit and said, "if Vout is higher than Vref, the pass transistor tends towards cutoff," which is how a series pass regulator (with NPN transistor and a positive supply) always works. A shunt regulator is configured oppositely, so that when Vout > Vref, the error amp output increases, turning the shunt element on *harder*.

Too bad for me, because while there is a series pass transistor doing the throttling, here, the control method is decidedly that of a shunt regulator...

Now, this was by all accounts a stupid mistake for someone like me to make. But the fact is, the more experience you have in a particular field the more likely you are to gloss over the details - in this case, the mistake I made was to only consider what the circuit must do from a general point of view: if Vout is too high then the pass element must turn off.

So, I feel pretty dumb right about now, but also vindicated because, hey, who besides me is a more vocal proponent of questioning authority around here?

 

[jeffreyj]

Quote:

Originally posted by tangent Another confusion: I'm reading through the fourth article in the 1995 series, and Galo recommends three changes to lower dropout voltage. I don't see why a single one of them helps. 1. Use a 2V green LED in the current source. Galo says this "enhances" the output swing of the op-amp but doesn't explain how. All I see is that the collector gets pulled about 0.4V above the base when the voltage across the pass transistor is 1V. This is not enough to forward bias the CB junction, but it's close. (For that reason, using a 2.2V or higher LED might be disastrous here.)

There is no real advantage to using a higher Vf diode as the op-amp will easily be able to adjust for absolute variations in Vf (of the LED) and Vbe (of all three transistors involved - CCS, pass and internal PNP sink) due to temperature, etc. Considering the CCS by itself, then, yes, a higher Vf LED will result in less percentage variation of the output current with changes in temperature (Vbe varies -2.1mV/C). If the temp changed 25 degrees then Vbe would change by ~50mV, but this is out of a nominal Ve of 1.0V (w/ Vf = 1.6V), or 5% total. If Vf = 2.0V, then Ve = 1.4V, and a 50mV change would be... 3.6% - not exactly an improvement worth writing home about, especially since the actual current going to the base of the pass transistor is controlled by the op-amp, anyway, not the CCS.

Increasing the voltage at the base will reduce the compliance range of the CCS, which you have rightly surmised, but I am not sure why you think that there is a CB junction anywhere in the circuit which is in danger of being forward biased... at least as long is Vin is more positive than ground!


Quote:

2. Galo recommends lowering R4 to 10K to 12K.. This will increase current through the LED, and it drains base current away from the CCS transistor more easily. Somehow this is good for 0.1V or so of improvement.

More current through the LED will only serve to minimize any variations in hfe of the CCS transistor. But who cares, since what determines the ultimate accuracy of the output voltage is the LM329 reference, not the CCS. More current through the LED otherwise results in a slight increase in its Vf, just like any other diode, which serves to lower the compliance of the CCS.


Quote:

3. Galo recommends using a transistor with a lower saturation voltage for the CCS transistor. This confuses me since saturation voltage goes up as collector current goes up, which it did due to the first change.

Well, a lower Vce[sat] would help to offset the increase in minimum overhead voltage otherwise required for Ie to be held constant, but presumably the input to this circuit will already be "pre-regulated" by a three-terminal regulator, so the abolute compliance voltage of the CCS is not so important. The absolute accuracy of the circuit is wholly dependent on the accuracy of Vref and of the output sampling resistors, R2 and R1. The op-amp will compensate for all other variables within its gain-bandwidth product.

Finally, I should clarify that I have been talking about the 1995 version of the Jung regulator. I just now jumped over to your site and saw that the 2000 version is also there and I don't remember it being there when I first commented in this thread...??? At any rate, I've had enough for right now.

 

[JohnFerrier]

To take tangent another direction, the schematics lack a detail that I think make the "Super Regulator" attractive to the integrated alternatives: the sense lines. I think it is one of the features of this circuit is to precisely control the voltage at the target circuit (independent of wiring and connector losses). The sense connections are aways somewhere, the SuperRegulator circuit allows one to move the connections to the load. (For the integrated devices the ground sense connection is internal to the component.) This may be a smaller point, but is the remaining reason, I would chose a dozen components rather than one.



JF

 

[ppl]

OMG this thread is about as vocal and argument filled as the similar discussion of this Circuit over at diy audio. They have over 25 posts on the subject and old Walt himself steps in to clarify some of the misinformation on this circuit. He has offered to email the complete article to interested persons however to save him lots of requests perhaps someone that has a mailbox capable of accepting a 14 Meg pdf file should have Walt email it to them and then redistribute it. Also be note that this is a copyrighted article and you must get Walt’s permission to post it on a web site. I would suggest you start at the first page located at http://diyaudio.com/forums/showthrea...5&pagenumber=1

Then about 22 or so pages later Walt steps in to clarify some misconceptions
http://diyaudio.com/forums/showthrea...&pagenumber=21


I must say in defense of this thread that it at least stayed somewhat on topic, while over at DIY audio they go off on so many side tangents (No Pun intended) from Nobel Prize winners to the claims of so called flat earthier people.

I am glad that Warren is reconsidering the use of this type of regulator over the LM-317 of his first power supply attempt. As he and most others on this forum Know I am not fond of IC regulators. I understand that some newer types from Linear technology have addressed the inherent stability issues of once chip monolithic regulators however the bandwidth is still quite low and this creates a rising output impedance vs frequency, Moreover pg.8 lower right hard corner graph shows the same overshoot and ringing that older LM series regulators have. Missing on the data sheet of the LT-1963 is output impedance vs frequency. This data was present on the old LM series regulators so perhaps it’s absence in the LT series indicates less than optimal performance in this area. By looking over the data sheet of this device I would suspect the output impedance vs frequency to not be any better than the older regulators.

A variation on the Jung regulator concept can be found in the technical manual to one of the many the commercial versions available. Hears just one from ALW Audio. http://www.alw.audio.dsl.pipex.com/M...g%20rev2.1.pdf
http://www.alw.audio.dsl.pipex.com/jung_schematic.htm

Another version somewhat related to the ALW version ready to plug into the socket that normally contains a IC regulator.
http://www.dipchipelec.com/apoxjr.htm





Analog Devices have an exalent article By Walt Jung on important aspects of High performance Voltage regulators and talks about issues of a low noise voltage reference in addition to a overall good general overview of the subject. Par2
http://www.analog.com/UploadedFiles/...ower_sect2.pdf

Also page 41 of this Analog device App Note gets into Voltage references and the advantages of rail to rail opamps in low dropout regulators if your thinking AD-8065 your getting warm.
http://www.analog.com/UploadedFiles/...33Section9.pdf
http://www.planetanalog.com/showArti...cleID=12802404


On pages 2-6 of this app note the effect of thermal distortion or drift is addressed with regard to Resistors and how they are mounted in addition to the power dissipation of the feedback resistors causing errors this is most critical in precision DC circuits and is also applicable to Audio Amplifiers. Wonder why one of the reasons I like use high value feedback resistors this article is also a great reference for all precision circuits.
http://www.analog.com/UploadedFiles/...60FFSect10.pdf

I don’t see many of you on msn’s DIY site however you will find this article of interest
http://groups.msn.com/DIYElectronics...wersupply.msnw

Discussion of capacitors used in regulators while covering all types of topologies Linear and Low dropout types are addressed.
http://www.planetanalog.com/showArti...cleID=12802147

Low noise op amp article describes the effect current consumption has on noise performance.
http://www.planetanalog.com/showArti...cleID=12802635

Theory and operation of Linear regulators available in an App note from National semiconductor. IMHO Must read.
http://www.national.com/an/AN/AN-1148.pdf#page=1

What’s all this Vbe stuff anyway is another entertaining article by Bob Pease describes the changes in Vbe of Bipolar transistors. Ya Most people Know this however some do not and others are not quiet sure of the relationship.
http://www.national.com/rap/Story/vbe.html
http://www.national.com/rap/Applicat...570,24,00.html history Abstract of Band-gap references Also a Must read.


I meant to also include one last link that may be slightly off topic however it concerns Mentoring and it is a good read also. Sorry for the omission but hear it is. Another great read by Bob Pease is located at
http://www.national.com/rap/Story/0,1562,7,00.html




http://www.analog.com/UploadedFiles/...60FFSect10.pdf

 

[tangent]

Quote:

while there is a series pass transistor doing the throttling, here, the control method is decidedly that of a shunt regulator...

Yes, I see what you mean.

Quote:

So, I feel pretty dumb right about now

Oh. You're human? Hmmmm...

It's a poor teacher who never learns from the students.

Quote:

I am not sure why you think that there is a CB junction anywhere in the circuit which is in danger of being forward biased...

The base of the CCS transistor is fixed, right? And the emitter is fixed above it as well. So, when the circuit's dropout voltage is being pushed to its limit in the 1995 circuit, the output of the error amp is going to be lifting C up past B. AoE says you can put current through that junction if C is a diode drop or more above B. They also say that if C is forced close enough to E, the transistor shuts off, so it seems to me this would do strange things to the regulator. That's why I'm latching onto all these strange behaviors...they are the only changes I can see near the dropout of the regulator.

Quote:

But who cares, since what determines the ultimate accuracy of the output voltage is the LM329 reference, not the CCS.

We're not especially interested in accuracy, but rather that whatever voltage we get stays steady. A "15V" regulator where each sample could range from 14.5V to 15.5V and which has 0.001% ripple is more helpful in audio than a regulator with 1% ripple whose DC component is 15.000V.

I know I'm lecturing to the teacher here. I only point this out because I think what they're worrying about is ripple voltage getting through the CCS. That's more work for the error amp to take care of, which lowers total ripple out of the regulator.

Quote:

I just now jumped over to your site and saw that the 2000 version is also there and I don't remember it being there when I first commented in this thread...???

Yes, I just put the 2000 circuit up recently.

I won't even get into 2000 circuit questions yet. There's plenty about it that puzzles me, but I think I'll stew on it for a while before asking questions.

Quote:

the schematics lack a detail that I think make the "Super Regulator" attractive to the integrated alternatives: the sense lines

I describe them in the text.

I don't agree with the complex way the article schematics are drawn; they get into layout issues which are better left to the layout. I favor a simpler schematic that glosses over details to aid quick comprehension.

Anyone who wants to make regulators won't make regulators as good as if they read the articles instead. That's just fine with me. The only need I'm filling here is a gentle introduction to the concepts. I'm not trying to replace the articles.

Quote:

OMG this thread is about as vocal and argument filled as the similar discussion of this Circuit over at diy audio.

There is confusion here, but also a willingness to learn. That thread you pointed me to is a bitter bicker-fest, with very little learning going on. You've just gone and reminded me why I still don't read diyaudio.com.

Quote:

He has offered to email the complete article to interested persons

I've read all the TAA articles. This thread is about the questions I have after reading the articles. As opposed to the diyaudio.com thread, where 95% of the posters hadn't read a single one of the articles.

By the way, the June 1996 Electronic Design article and the 2000 Audio Electronics articles are up on his web site. And I also gleaned some links from the diyaudio.com thread. All of this is in my article now.

Quote:

Analog Devices have an exhalent article By Walt Jung

Some of these look like it's a chapters from a book rather than standalone app notes. Are they perhaps from some of the ADI application manuals from the early to mid 90's? Are the other parts still available? The 2002 _Op Amp Applications_ references these older manuals frequently but I don't have copies. If one can download the chapters, I'll go get them all. I did a search on Analog.com and couldn't even find a way back to the link you gave.

Quote:

On pages 2-6 of this app note

Can you fix that link, please?

Quote:

Discussion of capacitors used in regulators

Good one! I had read Simpson's earlier article on this topic reproduced as an appendix in Pease's book, but this looks much more in-depth.

 

[ppl]

tangent i tried several times at getting the link to work and it never got the complete url transfered so i emailed you the pdf file sorry.
http://www.analog.com/UploadedFiles/...60FFSect10.pdf

I again tried to post the link and for some stupid reason it is not working. Oh well maybe you can post a followup and figure out how to link to that doc. Hopefully you will get the doc intact in the mail.

 

[phaedrus]

It looks like a break tag wanders into the url when the forum parses it. Is this the right article -- the hyperlink is just as broken, but you can paste it into your browser starting www, ending .pdf:

www.analog.com/UploadedFiles/Associated_Docs/48511532174844452507435855319349073202222160FFSect 10.pdf

 

[ppl]

yes that is the correct link however it still dose not work. Anyway i emailed the pdf doc i had on file to Tangent

 

[tangent]

It's a limitation of the web board software, I guess. I"ve made a redirect page for it:

http://tangentsoft.net/audio/misc/sect10.html

That will bounce you straight to the link target.

 

[peranders]

The problem with the link is that this forum software adds a line break, near at the end.

Why this extremely long URL's?

 

[PeterR]

nm