X/SuperSymmetry - Am I Missing Something?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've been looking at Pass' X/SuperSymmetry amplifier stuff, wondering what the patent is actually for. (I've even looked at the patent!) So far, I haven't found it. But, just in case I've overlooked it, I thought I'd ask here: what am I missing?

Here's my understanding of it so far.

The amps in question are basically differential-input, differential-output amps, with some kind of long-tailed pair (or simple variation thereof) for the inputs. And negative feedback is applied from the outputs to the inputs, by feeding each output back to the corresponding, inverting (relative to that output) input.

That seems to be it.

So, what's the patent actually for? What have I missed? Surely the patent can't be for what I've described (in summary) above. (Then again, given the reputation of the US PTO...)

?
 
I felt like you do, and I would still admit, I am not the one but... 1st, The normal diff pair does not use the F.B.. Due to the feedback though, any "output" signals appearing at the gates form some gain controled signal "processing". The signals at the gates are level shifted to the Sources and hence transfered to the other Source and level shifted to it's gate.:D :D :D At the same polarity that it showed up at the first gate... Therefore, these signals can possibly appear common mode and not be "processed" (amplified). Distortion appearing at the first drain and being feedback to the gate is not the same as music being summed at that point with the feedback. Therfore it is possible to "process" the input signals and treat the distortion like a common mode un processed signal. Nelson say's an order of magnatude I beleive better distortion figures... Wow, I will now wonder if I have made a total something of myself or maybe someone will help me out...

Thanks
 
flg said:
I felt like you do, and I would still admit, I am not the one but...

Thanks, but I already knew that that's how they work. It's just what long-tail-pair-style inputs do. And when negative feedback is applied to each (inverting) side? Why, it still does that - quelle surprise!

Let me put it another way.

Imagine you've got an op-amp with differential outputs. Being an op-amp, it's also got differential inputs. Suppose, also, that the outputs have a bit of distortion. Now, if you were to apply feedback to reduce that distortion, how would you do it? (Note that the inputs and outputs are both differential, which means that either input can be regarded as inverting, depending on which output you choose to regard as non-inverting ;) )

How would I do it? Well, I'd probably stick feedback resistors between the outputs and the inputs, and add some input resistors, too. The inverting input would be connected, by a feedback resistor, to the non-inverting output, and the non-inverting input would be connected, also by a feedback resistor, to the inverting output. The two feedback resistors would be the same as each other, as would the input resistors.

Does that seem reasonable? Does that seem, perhaps, obvious and trivial (once you understand what a differential-output op-amp is)?

Now, suppose, internally, the op-amp just happens to have a long-tail-pair-style input?

:eek:

What of Pass' patent now? How is it not a patent on the trivial, the obvious, the old-as-the-hills?

What am I missing?
 
Not knowing what is actually in your diff opamp, I would only repeat, I'm not such a good teacher. But, If you have read the patent, I believe the description covers what this acrticle also covers regarding typical opamp configurations... It may not address your particular idea ... Using an opamp to acheive a similar topology would likely be outside the scope of the patent ...

http://www.passlabs.com/downloads/articles/susy.pdf

This should explain why typical circuits are not symetrically canceling with balanced and opposite signals... vs just using feedbak... It is not meant to be anything complicated... Just very usefull... And patentable...:D :D :D
 
well... I'm sure this'll be far from exact but let me take a stab at this.

First, IMO a patent serves two purposes.
1) allows the owner exclusive rights to license or produce the particular technology for 10 years.
2) claims the owner as the inventor of the technology.

Now, obviously #2 is important to the petitioner but #1 is the important one.

The truth is, you're right, and there are plenty of patents out there that seem obvious or "duh". But lots of times it's only obvious after the fact that the obviousness comes out and the point really is: the patent was specific enough to get approved.

So if you look at the patent for X/SuSy, yeah it's negative feedback, but it only works for perfectly matched pairs, and the patent is specifically for cascoded LTPs with the feedback taken from the LTP, not after a VAS. And that's a pretty specific criteria. That's why it's patentable. If you were to apply X feedback to a Cascoded matched LTP with a say... a matched Source follower after it, well that'd be a different circuit and you could most likely market it (as long as your lawyer is better than NPs :) )
--
Danny
 
Netlist said:
Did you look at the article at passlabs?
There you will also find the patent # in case you want to understand the full details.

Yes, I've read that article, and (as I mentioned) had a look at the patent.

I also had a bit of a search through this forum, too.

And I've also had a brief look at the SuperSymmetric Son of Zen thing on (what was it?) PassDIY (or whatever it's called).

Perhaps I should spell it out: I have no difficulty seeing how distortion on one output ends up on the other output, such that it differentially reduces that distortion.
 
The nice thing about this topology is that many people seem to implement a lot of variants. From extremely simple (which most of us like) to more complex schematics have arisen all over the passlabs forum. I'm not sure if an x-tube amp has been done but I'm sure it must be perfectly possible.

/Hugo:)
 
Well, its easy to say all the words now. We have look at all the schematics, the smart tricks, from all over the smart brains all over this world, and say what's so special about it, because it makes sense and it is just the way to achieve the targetted result, many else uses that same trick, it is the only way to do it, etc...etc....etc.....

But imagine when you have no, or much less schematics to study, what year is that patent? Is internet already available in that year? And you come with that idea from yourself, resulting something special result, will you think "Ah, I will not patent it, although I make amps for my living"?

You should see patent #6.882.225, it is dated 10 April 2005, is he inventing something not known before 2005? Why do you think he make the patent, cost so much?
 
Some Patent Basics

Perhaps it would be a good idea to review some patent basics. (You might find it helpful to have a look at the US PTO's website, particularly the patents section.)

For an invention to be patentable, it must be: new; non-obvious; and useful. (Okay, there's more to it than that, but that's the essential gist of it.) Usefulness is not in dispute here, but novelty and non-obviousness are. (That's 'novelty' in the sense of 'newness', not in the sense of, 'Oh, look what just fell out of my Christmas cracker!')

Electronic amplifiers? Not new.

Differential inputs? Not new.

Differential outputs? Not new.

Long-tailed pairs, and the like? Not new.

Implementing an electronic amplifier, with differential inputs and differential outputs, as a long-tailed pair? Not new. (Input stages of op-amps, for example.)

Applying negative feedback to electronic amplifiers? Not new.

The idea of applying negative feedback to an amplifier with differential inputs and differential outputs? Obvious.

The idea of applying negative feedback to such an amplifier when that amplifier just happens to be implemented with a long-tailed-pair-style input stage (which might just happen to be the only stage in the amplifier)? Obvious.

So, what's Pass' patent for? What's the new, non-obvious thing/feature/method/etc that's required for the patent to be valid? What am I missing?
 
Sounds like Pass bashing to me........................:rolleyes:
 

Attachments

  • xy020.jpg
    xy020.jpg
    13.3 KB · Views: 790
jam said:
Sounds like Pass bashing to me........................:rolleyes:


It does seem that way. Fortunately NP takes it well :cool:, he seems like a VERY reasonable guy. I know nothing of patents, so I'm just a by-stander on this one .:clown: But:


Many (20?) years ago I thought...hey, why can't smoke detectors be linked by RF so when one goes off, they all do! This is now being implemented just as I had invisioned. I haven't looked but I'm sure there is some sort of patent on the thing. But why??

RF..not new
smoke detectors..not new
smoke detectors that all trip at once..not new
The exact implementation of all technologies to gain the desired result..new perhaps?(I guess this woukld be why)

I dunno...
 
You may say a shoe is just a shoe.
Think up a new shape of shoe and it still is a shoe.

My partner came up with a shoe that is shaped to remain in line with the ergonomical axis of the human body at every position of movement and applied for a patent.
One sporting shoes Nickelodeon is already in violation with the patent at his Asian production fascilities, the other shoe Adidi-Adida is in negotiations for buying/using rights of the patent.
No idea why, it is just a shoe.

Mr Pass has a mind that comes up with a lot of ideas, and has done so for a very long time.
Despite his many patents many still seem to enjoy copying his shoes.
No idea why, the guy wears Flamingo shoes !
 
Sometimes, very obvious things can hide beautiful-novel goal, if you think about it deep enough:D

Differential outputs? Not new.
From what I see, the essence of that patent is about that..........,
BUT it is not just "differential outputs", it is about using the differential outputs...............
WITH very short signal propagation (well, folded cascode is counted as 1 stage).................
TO achieve what?..............

To achieve maximum distortion cancelation with differential outputs.

You make X amp (or differential outputs amp to you) with ordinary 3 stages or more (common differential+2xVAS+2xoutput stage+driver(s) stages), you won't get the same thing with what this patent intended.:D
 
Member
Joined 2001
Paid Member
Certainly one novel aspect of the design is its handling of common-mode error. Designers of previous differential shunt feedback amplifiers, of which this is an example, recognized that the balanced feedback network alone could not reduce residual common-mode error. As a result, these designs included (or were amended to include) extra corrective circuits to reduce any CM residual.

The Pass circuit exploits a feature of balanced shunt feedback amps that apparently was not previously appreciated, which is that the existing shunt feedback network tranfers CM error across the amp's line of symmetry in such a way that CM error is reduced when evaluated differentially at the output.

If I understand correctly, the degree of CM error reduction achieved is greater than what would be expected from a consideration of the loop gain, due to the iterative and mutually reinforcing nature of the feedback. Relying on this mechanism for distortion reduction, rather than on extra loop gain or corrective circuits, allows for a simpler and potentially more stable gain topology for a given level of performance.

It could be argued that this approach to handling residual CM error will not work as well as expected on paper when the load has an asymmetric impedance characteristic, as it well might in a power amp application where the load is a loudspeaker. But since in a simple circuit the CM residual would then appear mostly as increased second harmonic content, the subjective effects are likely to be pretty benign.
 
azira said:
well... I'm sure this'll be far from exact but let me take a stab at this.

First, IMO a patent serves two purposes.
1) allows the owner exclusive rights to license or produce the particular technology for 10 years.
2) claims the owner as the inventor of the technology.

According to the US Constitution:-

The Congress shall have Power ... To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries[.]

So, (in the US) the purpose of patents is "To promote the Progress of Science and useful Arts".

The truth is, you're right, and there are plenty of patents out there that seem obvious or "duh". But lots of times it's only obvious after the fact that the obviousness comes out and the point really is: the patent was specific enough to get approved.

Unless I'm missing something about what the patent is actually for, it was obvious before Pass did it. Negative feedback was being applied to amplifiers long before Pass filed for his patent in 1994. Differential-input, differential-output, long-tailed-pair-based amplifiers (such as long-tailed pairs themselves) existed long before 1994, as well. The idea of applying negative feedback to such amplifiers simply fails the non-obviousness test.

(When I first learned of the existence of amplifiers with differential outputs as well as differential inputs, one of the first things I asked myself was: how would negative feedback be applied? I'd already learned the basics of op-amps when I was in what Americans would call 'high school', including the basic stuff of negative and positive feedback. It wasn't at all difficult for me to work it out. This was, oh, fifteen years ago, or thereabouts, if not more.)

So if you look at the patent for X/SuSy, yeah it's negative feedback, but it only works for perfectly matched pairs, and the patent is specifically for cascoded LTPs with the feedback taken from the LTP, not after a VAS. And that's a pretty specific criteria...

Irrelevant. That application of negative feedback to a differential input/output amplifier, as applied to the amps in question, is obvious, regardless of the internals of the amplifier itself. It doesn't cease to be obvious simply because the amplifier itself (before feedback is applied) happens to be implemented in some particular way. It doesn't cease to be obvious, it doesn't cease to be old and well-known, just because the amplifier itself happens to use folded cascodes internally, happens to have a long-tailed-pair-style input stage, or happens to have no further stages after the input stage.

It's like building walls from bricks and mortar. The bricks are cemented together by the mortar. It's nothing new; it's far from non-obvious. Does it matter if the bricks are made out of a new kind of fired clay? No, because they're still bricks, and the idea of cementing them together with mortar is nothing new. Does it matter if it's a new kind of mortar? No, because the idea of cementing bricks together with mortar is nothing new. Does it matter if the bricks were made using a process that doesn't involve firing them at all? Again, the idea of cementing them together, to build a wall, doesn't cease to be obvious and innovel.

As for the amplifier itself, before negative feedback is applied, it seems to be just a simple variation on long-tailed pairs (if it's a variation at all). Long-tailed pairs, and simple variations thereof, are nothing new. Implementing them with MOSFETs? Nothing new. Cascode? Nothing new. Using long-tailed pairs, and the like, as differential-input, differential-output amplifiers? Nothing new.

So, what's the patent for? What am I missing?

If you still don't 'get it', think of the amplifier itself (without feedback) as a black box. (This is the sort of thing engineers do all the time, by the way.) All you know about this black-box amplifier is that it's got differential inputs, and differential outputs, and that it's got a particular, open-loop gain for differential signals, and a particular, open-loop gain for common-mode signals. That's it. You don't know how it's internally implemented, whether it's got a long-tailed pair input stage, or anything like that.

Now, knowing that the application of negative feedback to amplifiers (such as differential-input op-amps) is as old as the hills, surely the idea of applying negative feedback to this black-box amp is also obvious. And it's easy to work out how to do it, too. Very easy. The idea of doing it is obvious, and actually doing it is just a matter of simple, so-called 'high school' algebra - same as with ordinary op-amps.

As it's a black box, the internals can't possibly affect whether or not the application of negative feedback is old or obvious. Folded cascodes and long-tailed pairs do not have magic, psychic properties that render engineers blind to the otherwise obvious and well-known. The patent simply can't be for the application of negative feedback to such amplifiers (unless the patent is invalid and should not have been granted in the first place). It would have to be for the amplifier itself.

The amplifiers themselves seem to be just simple (old and obvious) variations of long-tailed pairs. They respond to negative feedback in the same kind of way that long-tailed pairs do anyway - because that's essentially what they are! There doesn't seem to be any basis for a patent there. The folded cascode outputs don't seem to change that, either. The absence of further stages doesn't make a difference, because such long-tailed-pair-style stages are already amplifiers themselves, anyway.

So, what am I missing? What's the patent for?
 
Pass-Bashing?

jam said:
Sounds like Pass bashing to me........................:rolleyes:

Yes, it sounds like Pass-bashing to me, too :xeye:

It's not supposed to be Pass-bashing, though. What it is supposed to be is a process of elimination, in the hope of homing in on what it actually is that the patent's for. That's why I keep asking: what am I missing?

So far, it seems to me that it must be something to do with the amplifier itself (not including negative feedback), such that when negative feedback is applied to the amplifier, it's groovy (improved distortion reduction, etc). Trouble is, Pass' description of how it works just fits with my understanding of long-tailed-pair-style amps and negative feedback, anyway :(

But patents are (supposed to be) for the non-obvious, and it's that non-obvious something that I'm looking for. The trouble with non-obvious things is that they aren't always obvious, which is why it seems sensible, to me, for me to ask, 'What am I missing?', rather than to just leap to the conclusion that the patent's invalid and shouldn't have been granted.

:)
 
Hi, Simon,

But patents are (supposed to be) for the non-obvious, and it's that non-obvious something that I'm looking for. The trouble with non-obvious things is that they aren't always obvious, which is why it seems sensible, to me, for me to ask, 'What am I missing?', rather than to just leap to the conclusion that the patent's invalid and shouldn't have been granted.

From the quote above, I got a feeling you already have a "hint" about what "should" be the answer to your question of "What am I missing" in SuSy.
Why? Because you seems to refuse explenations like Joe Berry's Common Mode distortion, and still ask "What am I missing"?

To really know the power of SuSy, it is not as easy as reading or drawing schematics, comparing one to another. You got to BUILD it, compare it to something that you THINK it won't differ much to.

Then you will know the answer to your question of "What am I missing?"

If you keep comparing schematics and don't build the real amp(s), you will keep asking this without knowing the answer. SIM cannot play music with real speaker, it won't help much here.

Sometimes the effect of small difference(s)/details can be shocking in audio result. In schematics it just shown as 1cm additional drawing line, or slightly different connection, or different resistor value, you won't take it important.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.