First, it is not a mundane example of a differential pair - if it was
you could look it up in Horowitz and Hill. Only in 2000 was
such a topology incorporated in a chip, 6 years after the
patent issued.
The patent issued without any objections as far as I know, and
the examiner was the well-known Stephen Mottola, the primary
reference was Walker et al, and the patent issued in a few
months.
It's development was not the casual mating of two separate
amplifiers as you say:
"He started off with two, uncoupled amplifiers, used together for
the amplification of differential signals. Then, he joined them
together at the waist (as it were), adding a resistor between the
MOSFET sources. Superficially, it doesn't look like a long-tailed
pair, because it doesn't have a single tail. Instead, the tail's split
in two, in parallel, with the connecting resistor in between."
First off, the circuit was always a differential pair. Use of dual
current sources for bias was simply a convenience, so don't
tell me my process.
What I was thinking for about 15 years was the answer to the
question - "In an error correction amplifier (such as Walker's
or Mike Wright's) who corrects the corrector?"
Seen in hindsight, it might look trivial, but then so does a paper
clip. True, it doesn't do all that much for just the diff pair by itself,
but incorporated in a power amplifier with additional distortion
after the diff pair, the circuit cross couples error on one side to
roughly equal error on the other side, resulting in cancellation.
If you look at the output of any side singly, it bears a strong
resemblance to the open loop performance of the circuit.
I can see from your comments that you have never built such
an amplifier. As a concept it accomplishes more with less
circuitry and feedback than any other design I know, and it is
the only 2 stage circuit I've seen that can deliver .01% at a
kilowatt. (No, I don't count cascodes and current sources as
gain stages, and neither should you.)
"Pass' 'invention' turns out to be nothing more than a simple,
insignificant change to what must undoubtedly be very well-
known prior art."
Gee, thanks a lot. Why don't you give me a paper cut and
put some lemon juice on it.
you could look it up in Horowitz and Hill. Only in 2000 was
such a topology incorporated in a chip, 6 years after the
patent issued.
The patent issued without any objections as far as I know, and
the examiner was the well-known Stephen Mottola, the primary
reference was Walker et al, and the patent issued in a few
months.
It's development was not the casual mating of two separate
amplifiers as you say:
"He started off with two, uncoupled amplifiers, used together for
the amplification of differential signals. Then, he joined them
together at the waist (as it were), adding a resistor between the
MOSFET sources. Superficially, it doesn't look like a long-tailed
pair, because it doesn't have a single tail. Instead, the tail's split
in two, in parallel, with the connecting resistor in between."
First off, the circuit was always a differential pair. Use of dual
current sources for bias was simply a convenience, so don't
tell me my process.
What I was thinking for about 15 years was the answer to the
question - "In an error correction amplifier (such as Walker's
or Mike Wright's) who corrects the corrector?"
Seen in hindsight, it might look trivial, but then so does a paper
clip. True, it doesn't do all that much for just the diff pair by itself,
but incorporated in a power amplifier with additional distortion
after the diff pair, the circuit cross couples error on one side to
roughly equal error on the other side, resulting in cancellation.
If you look at the output of any side singly, it bears a strong
resemblance to the open loop performance of the circuit.
I can see from your comments that you have never built such
an amplifier. As a concept it accomplishes more with less
circuitry and feedback than any other design I know, and it is
the only 2 stage circuit I've seen that can deliver .01% at a
kilowatt. (No, I don't count cascodes and current sources as
gain stages, and neither should you.)
"Pass' 'invention' turns out to be nothing more than a simple,
insignificant change to what must undoubtedly be very well-
known prior art."
Gee, thanks a lot. Why don't you give me a paper cut and
put some lemon juice on it.
Simon G Best said:
But what you don't necessarily get with a LTP is the ability to modulate common-mode current to the degree needed to develop a CM correction signal on the output. If you bias the LTP with a constant current source you won't get any CM current modulation at all, and - unless there is another mechanism at work that I'm not aware of - you will not see *enhanced* CM cancellation at the output.
A single "long-tail" resistor bias may work well for a complementary JFET front end, where supply isolation is not so much an issue, and I think that's just how it's done in some Pass products. But for a single diff pair, the best approach is probably the one shown in the patent: using a CCS for bias and a resistor to ground whose value is chosen to allow for sufficient CM current modulation.
Nelson - now that you are on the thread, please correct me if I go astray on these explanations.
Nelson Pass
"He started off with two, uncoupled amplifiers, used together for
the amplification of differential signals. Then, he joined them
together at the waist (as it were), adding a resistor between the
MOSFET sources. Superficially, it doesn't look like a long-tailed
pair, because it doesn't have a single tail. Instead, the tail's split
in two, in parallel, with the connecting resistor in between."
Mr Pass i have developed a few longtail pair power amp DESIGNS , they are basicly lpt circuits, are they violating your pattent/copyrights
secondly Simon G doesnt know a thing about the benefits of
power ltp
cheers
"He started off with two, uncoupled amplifiers, used together for
the amplification of differential signals. Then, he joined them
together at the waist (as it were), adding a resistor between the
MOSFET sources. Superficially, it doesn't look like a long-tailed
pair, because it doesn't have a single tail. Instead, the tail's split
in two, in parallel, with the connecting resistor in between."
Mr Pass i have developed a few longtail pair power amp DESIGNS , they are basicly lpt circuits, are they violating your pattent/copyrights
secondly Simon G doesnt know a thing about the benefits of
power ltp
cheers
Hi, Simon,
This is not about technical theory, but just about what happened in the "field". At first, I also have always think to my-self, why NP doesn't just use a single LTP instead of 2 bridged common emitors with each current sources. In this thread we can see the theory background. (thanks to NP and Joe Berry)
But in the real CCT (without knowing the theoritical background), using a single LTP (like you make the "transformation" from 2 bridged single common emitors to single LTP), will drive you NUTS
. Something like Grey Rollin's AlephX is exactly what will make you nuts (Don't worry, there is a cure for it now). DC offset (absolute and common offset) is like a wild horse. (unless you can buy super-matched twin mosfets that is mirror image in everything in 10mA bias to 50V operation )
Making it 2 bridged common emitors like the original patent is VERY-VERY-VERY clever from my point of view
This is not about technical theory, but just about what happened in the "field". At first, I also have always think to my-self, why NP doesn't just use a single LTP instead of 2 bridged common emitors with each current sources. In this thread we can see the theory background. (thanks to NP and Joe Berry)
But in the real CCT (without knowing the theoritical background), using a single LTP (like you make the "transformation" from 2 bridged single common emitors to single LTP), will drive you NUTS
. Something like Grey Rollin's AlephX is exactly what will make you nuts (Don't worry, there is a cure for it now). DC offset (absolute and common offset) is like a wild horse. (unless you can buy super-matched twin mosfets that is mirror image in everything in 10mA bias to 50V operation )Making it 2 bridged common emitors like the original patent is VERY-VERY-VERY clever from my point of view
Pass Rules!
It is my humble opinion that the X-circuit is one of the most important advancements in audio design to come along in decades.
Like lumanauw has stated, it is a pure stroke of genius...............
...as simple as you might think it is.
It is easy to cast stones but till you have listened to one of these amplifiers........I have, and I can honestly say that is one of the best designs I have heard, if not the best.
Regards,
Jam
It is my humble opinion that the X-circuit is one of the most important advancements in audio design to come along in decades.
Like lumanauw has stated, it is a pure stroke of genius...............
...as simple as you might think it is.It is easy to cast stones but till you have listened to one of these amplifiers........I have, and I can honestly say that is one of the best designs I have heard, if not the best.
Regards,
Jam
Attachments
Jam,
This forum is "All Things Nelson Pass"
For someone who struggles to understand complex circuits, to me NP's designs are a very welcome sight...like a cool drink on a hot day; like a rainbow on a rainy day (from Seattle area); like an oasis in the desert...should I go on?
Thanks again NP! Finally I could somehow understand how an amp circuit go about its work.
This forum is "All Things Nelson Pass"
For someone who struggles to understand complex circuits, to me NP's designs are a very welcome sight...like a cool drink on a hot day; like a rainbow on a rainy day (from Seattle area); like an oasis in the desert...should I go on?
Thanks again NP! Finally I could somehow understand how an amp circuit go about its work.
Joe Berry said:
OK, your explanation does confuse me a little. I think the X
circuit is simpler than this - the errors of one side show up on
the other side in phase so that they cancel as much as
possible. If you look at one side alone, the performance is
nothing to write home to mom about.
mastertech said:
If I had invented the ltp, I certainly would be more pompous than
I already am.
I cannot speculate about what Simon knows, but the questions
he raises are legitimate enough. I just think he could be more
polite or better yet, say it on another forum where I won't see it.
Nelson
"If I had invented the ltp, I certainly would be more pompous than
I already am.
I cannot speculate about what Simon knows, but the questions
he raises are legitimate enough. I just think he could be more
polite or better yet, say it on another forum where I won't see it.
"
cheer up Mr Pass you cant win them all, He certainly knows a thing or two deosnt he?
cheers
"If I had invented the ltp, I certainly would be more pompous than
I already am.
I cannot speculate about what Simon knows, but the questions
he raises are legitimate enough. I just think he could be more
polite or better yet, say it on another forum where I won't see it.
"
cheer up Mr Pass you cant win them all, He certainly knows a thing or two deosnt he?
cheers
Nelson Pass said:
Indeed. It's not just any old differential pair, but a long-tailed pair.
According to your "DISCLOSURE OF THE INVENTION" in your patent, that is what you did:-
Yes, indeed that's what those two current sources were for. But, when the two amplifiers are coupled together, that pair of current sources are equivalent to a single tail of a long-tailed pair. That might not be what you'd intended, or what they're there for, but that's the result.
No kidding! That's what long-tailed pairs do.
I bet, when "incorporated in a power amplifier with additional distortion
after the diff pair," there's a second stage with high input impedance...
Looking at the schematic on the patent cover page, I see that it's basically a transconductance amplifier (with differential outputs), with negative feedback added. Remove the negative feedback (resistors 22, 23, 36 and 37), and you've got the raw transconductance amplifier itself (which happens to be that long-tailed pair I keep banging on about). As I'm sure you know, a transconductance amplifier is equivalent to a voltage amplifier with enormous open-loop gain, and enormous output impedance.
So, what happens when you add a second stage? A second stage with high input impedance? Perhaps very high input impedance? Well, unless the input impedance of the second stage is infinite, it'll lower the effective voltage gain of the first stage. But the remaining open-loop gain could still be huge. When negative feedback is applied (over the whole lot), the reduction in distortion will be truly impressive.
(Oh, but that doesn't involve the cross-coupling
)Anyway, if you're still sure that your amplifier was not the same as a long-tailed pair, can I ask you to give your understanding of how long-tailed pairs work? In particular, long-tailed pairs with constant current sources for the tails, so that they're very long tails? It would be interesting to see why you think they don't work in the same way, why you think they're different from what you came up with.
Simon, you have managed to be very annoying.
What do you want? Get a workaround to Nelson Pass' patent? Bash the patent office? Bash Mr. Pass?
Go invent something for yourself, and patent it. You have just to combine different prior arts together.
I would appreciate if this thread would be closed. It does make me sick to see how a great contributor to the DIY world has been treated here.
Tino
What do you want? Get a workaround to Nelson Pass' patent? Bash the patent office? Bash Mr. Pass?
Go invent something for yourself, and patent it. You have just to combine different prior arts together.
I would appreciate if this thread would be closed. It does make me sick to see how a great contributor to the DIY world has been treated here.
Tino
Simon,
Perhaps that's the point:
Maybe you don't understand the issue or maybe you do but don't grasp the subtlety that others seem to. Some people are more capable in certain areas than others,
Sincerely, from what you have said to this point I don't think further discussion is going to help. Some people think you are indeed missing something and have tried to explain, you don't think you are missing something..and your points that it is a subtle issue seem correct.
The results of the patent seem to be an amp that has novel quality-it does seem to be different sounding beast
What more is there to discuss? I suggest everyone just give it a rest...
Perhaps that's the point:
Maybe you don't understand the issue or maybe you do but don't grasp the subtlety that others seem to. Some people are more capable in certain areas than others,
Sincerely, from what you have said to this point I don't think further discussion is going to help. Some people think you are indeed missing something and have tried to explain, you don't think you are missing something..and your points that it is a subtle issue seem correct.
The results of the patent seem to be an amp that has novel quality-it does seem to be different sounding beast
What more is there to discuss? I suggest everyone just give it a rest...
Simon G Best said:
Perhaps that's the something you're missing - It's not the
"long tail pairs", it's what you do with them.
In any case, it's clear that an ltp is not required to implement the
concept, as illustrated in the example of using the so-called
"current feedback" opamps to create an X amp. I don't see
your description of an ltp in that circuit.
Perhaps you should build up a nice 2 stage amplifier as you
have described and document the performance for us. As
I said, you don't have to count the cascodes or current sources.
jacco vermeulen said:
No, that's flawed logic. It proves the circuit can be implemented and work, but it does not prove the circuit is fundamentally different from any other circuit. Now, don't misunderstand me. I don't say the circuit isn't novel, I am just saying the argument I quoted is wrong and adds to the confusion.
Now the circuit itself
So for the circuit itself, there are many claims and a lot of confusion. I admit I haven't read the patent, but as I understand, the main idea of SUSY is the elegant idea of crosscoupled feedback. The use of an LTPish input stage is not necessary, as Nelson himself has pointed out.
Then, this non mandatory LTPish input stage seems to cause a lot of confusion. Since there is feedback into the emitter/source nodes, it is obiouvlsy not an ordinary use of an LTP, at least. Whether an LTP would still qualify as an LTP if we inject feedback into the emitter/source of either or both transistors, I don't know. So, if we also remove this feedback, we are left with two transistors, each of which has a current source (both presumably having the same value) at the emitter/source, and the emitters/sources are joined by a bridge resistor. What is not clear to me is if it is disputed that this circuit is functionally equivalent to an LTP (which is obviously is in theory and assuming perfectly matched components). I am too lazy to figure out if behaves differently in the case of mismatched transistors.
Nelson Pass said:
According to your first patent claim, what you're doing with it is applying feedback on both sides
So?
Perhaps you should build up a nice 2 stage amplifier as you
have described and document the performance for us. As
I said, you don't have to count the cascodes or current sources.
Alas, I lack appropriate test equipment at this time
However, I can easily put something together in gEDA, and run some spicy simulations, if you like - Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.