You are correct. The complementary emitter-coupled pair is not a cascode.
The input impedance of a cascode is simply the input impedance of the common-emitter amplifier, 2*beta*Vt/Ic. The small signal impedance of the emitter-coupled pair is (beta_n*Vt/Ic)+((beta_n+1)*Vt/Ic). Note that the emitter-coupled pair has the small-signal model of a standard diff pair.
In circuit analysis, the diff pair and the Rush pair are both emitter-coupled pairs.
Sorry, I should have mentionned that I deleted it. My tought was badly expressed.
Looking at transistor modes, I find that the terminology is insufficient and inadapted to subtle analysises of circuits.
For example, what does a common emitter transistor become when a degenerative resistor is added to the emitter ? Is it still a common emitter with a smaller transconductance ? I prefer to say it is a split load configuration.
Similar imprecisions happen with common base configurations.
Let's imagine a very standard non-inverting amplifier having a single transistor input stage, non-inverting input (in+) at base, inverting input (in-) at emitter.
At some time,
- the in+ signal applied at the base is null (in fact always slightly noisy) but still connected to its source having a very low impedance as usual.
- a signal coming from the output of the amplifier (as does the counter-electromotive force of a driver in a noisy environment) is applied to in- which is the emitter of the transistor.
The in+ input transistor having its base AC connected to ground by a very low impedance and receiving feedback at its emitter could be considered as being common base.
However when a useful signal is applied at the in+ transistor base, a feedback voltage is applied at the emitter. The transistor is certainly not in common base mode anymore.
Si it looks like the input transistor works in two modes (even three if we look at the emitter replying the AC base voltage) depending of the signal input being null or not. Really ?
I find such an analysis shows some weakness when speaking of common base mode.
Last edited:
Yes, that is why I would not use a singleton input stage as an example of a common base transistor, it is a much larger sin than to call the Rush pair a cascode. Still, if we are discussing the amplifier's response to a current from the load, IE it's output impedance, then there is a case for treating the input stage like a common base stage, because the base voltage is no longer changing. I can hear the words "for the sake of analysis, we will treat the input transistor as a common-base stage...". Such trickery is commonplace, we all adapt a situation for the purpose of learning or doing work.
Agreed on the imprecision, that is why I don't get too worked up when someone uses the wrong word or terminology. Often I can see why they did it, and that informs me of their purposes, their level of skill, and their blind spots that I can help to fill in. When the misunderstandings are corrected, the terminology tends to follow anyway. Correct terminology can follow from correct understanding, but terminology in itself likely can't correct a misunderstanding.
Agreed on the imprecision, that is why I don't get too worked up when someone uses the wrong word or terminology. Often I can see why they did it, and that informs me of their purposes, their level of skill, and their blind spots that I can help to fill in. When the misunderstandings are corrected, the terminology tends to follow anyway. Correct terminology can follow from correct understanding, but terminology in itself likely can't correct a misunderstanding.
Last edited:
Thank you very much - very good advice:
Rush Cascode (NTP = "no-tail pair") or tailless pair = Rush's feedback current amplifier
for details go to §13.9.1
"Other Stage Configurations" on page 718-722 in the book "Amplifying Devices and Low-Pass Amplifier Design"
go to the pdf attachment.
The genuine paper released already 1964 (57 years ago) is the follow (obviously not available on the web):
Charles J. Rush: New Technique for Designing Fast Rise Transistor Pulse Amplifiers
(Argonne National Lab., nl.) Rev. Sci. Instr., 35: 149-56 (Feb. 1964)
https://aip.scitation.org/doi/pdf/10.1063/1.1718766
What is it about ? (ABSTRACT)
A unique fast amplifier configuration is presented. With this configuration a design for a given gain and rise time can be accomplished using transistor parameters normally given on the manufacturer's data sheet.
A five‐stage amplifier having a gain of 1580 and a rise time of 3.05 nsec was built using the technique presented here.
Who can upload this paper ?
In this case there are additional links:
Yet another DOA
EP1625656B1 - Circuit for improved differential amplifier and other applications
- Google Patents
CROWN BROADCAST Fanfare FT1-AP в Екатеринбурге
Audio amplifier circuit_electronic circuit schematic: low distortion audio amplifier - Programmer Sought
P.S.: Maybe the reason for the exceptional good sound of NAD's 3020 RIAA phono amp is even the use of Rush Cascode (NTP = "no-tail pair") as input stage.
Attachments
Last edited:
Yes, it was already brought up.
But since it keeps coming back, I had a look at the patent. The patent is not for a differential. It specifies one transistor as having it's base connected to ground in a common base configuration, for the purpose of isolating the input node from the output.
So Hugh's description is correct if we view it as a description of the Sziklai patent.
It's even possible that Sziklai didn't patent the differential version because it already existed - and we just haven't found the origin yet.
So it is not really a stretch to credit Rush with it's use as a differential stage. It seems we can still call it a Rush pair for now.
But since it keeps coming back, I had a look at the patent. The patent is not for a differential. It specifies one transistor as having it's base connected to ground in a common base configuration, for the purpose of isolating the input node from the output.
So Hugh's description is correct if we view it as a description of the Sziklai patent.
It's even possible that Sziklai didn't patent the differential version because it already existed - and we just haven't found the origin yet.
So it is not really a stretch to credit Rush with it's use as a differential stage. It seems we can still call it a Rush pair for now.
I did it. Send me an MP. However the circuit shown in the paper is not a Rush circuit as currently discussed here.
By the way, I thought the name of the inventor of the NPN-PNP coupled emitters circuit was Christopher Rush. I may be wrong, Ed Cherry attributes it to Charles Rush.
This circuit, just as the NAD3020 phono preamp, have two inputs, non inverting and inverting, which clearly demonstrates that a Rush circuit is not a cascode.
See also :[ The Rush Cascode: Possible Wiki page ]
The following may look senseless.
I investigated the various transistor modes and was particularly puzzled by the common mode base.
I slowly came to the conclusion that the AC connection to ground of a transistor base does not systematically make it working in common base mode.
My old professors taught me to look for the inputs and outputs. In a common base circuit, the base may not be connected to AC ground but it will be common to both the input loop and the output loop. It may not be AC ground--it may be another input or output.
However, by principles of superposition, you would treat other inputs as AC grounds, and ignore other outputs, and then analyze those separately.
Or maybe there is no output. Maybe it is just a load.
The emitter-coupled pair is by definition an emitter follower driving a common base stage. That it can be used as a differential pair is academic. The complementary emitter-coupled pair was invented by Sziklai et al and not by Rush.
That is Sziklai's definition, but there is no "one true dictionary" with all others being false.
It was invented by Sziklai as far as we can tell but the patent doesn't show whether he ever used it as a differential or even realized it's potential as such. But he must have been a clever guy, surely we can find it as a differential in one of his schematics (where are they anyway?). If not, Rush is the next contender.
It was invented by Sziklai as far as we can tell but the patent doesn't show whether he ever used it as a differential or even realized it's potential as such. But he must have been a clever guy, surely we can find it as a differential in one of his schematics (where are they anyway?). If not, Rush is the next contender.
Attachments
It's not just Sziklai's definition; it's the universally acknowledged and correct definition of an emitter-coupled pair: an emitter follower driving a common base stage. See "Electronics" by G. Waterworth, ISBN 0-333-45871-0, page 81.
Last edited:
A clear example of an emitter follower driving a common base stage: a complementary emitter-coupled pair.
Does that mean that we cannot call a Rush differential an emitter-coupled pair because one base is at the input and the other base is at a feedback network, and thus neither can be considered in common base configuration?
If a definition is overly strict, no one will respect it. If it is useful however, people will adopt it casually.
If a definition is overly strict, no one will respect it. If it is useful however, people will adopt it casually.
Last edited:
It is often interpreted so, even used by great names. It is wrong.
Think of the exact role of the input stage of a feedback amplifier.
It is the same, be it using a single transistor or two coupled emitters of transistors.
Forr, that an emitter-coupled pair can be used as a differential amplifier to, for example, subtract the feedback signal from the input signal in a feedback amplifier, does not invalidate the definition that such a pair is essentially a common-collector stage driving a common-base stage. Even a single common-emitter stage can be used as a differential amplifier to subtract the feedback signal from the input signal in a feedback amplifier, but that does not nullify the fact that it remains a common-emitter stage.
There are two known versions of a differential amp:
A: a complementary version as the input stage of NAD's 3020 MM phono stage and
B: the mostly used LTP version.
Both versions do work in two modes together with an additional VAS stage:
1) non inverted mode (like NAD's 3020 MM phono stage)
2) inverted mode.
In the inverted mode both known differential amp versions works as an emitter-coupled cascode, but the Rush-version is only the complementary version in that operating mode.
In general the inverted mode provide much more better audio - already in the early 80s either Mr. Bob Stuart or Allen Boothroyd (Meridian) mentioned this in an interview with the long-defunct (and for me the best) German Hi-Fi magazine "HiFi-Exklusiv". - check out also post #14 under
Non-complementary class AB linearized amp
Are there papers concerning the advantages of the inverted mode on amplifier stages from Mr. Bob Stuart or Allen Boothroyd ?
BTW - there is a third version of differential amp, consist of only a single transistor - go to
Simple Class A Amplifier
Fig. 3, Tr4.
First input (e. g. for input signal) is on base, second input (e. g. negative feedback-input) is on emitter (unfortunately not helpful for dc amplifying).
Attachments
Last edited:
I agree with you saying there is an emitter follower in an emitter coupled pair.
This emitter buffers the feedback network. Its base is usually called the inverting input.
It does not change the status of the other transistor whose base is the non-inverting input.
I partly agree. I should precise that both signals are voltages and that the single emitter is in split mode due to its emitter degenerating resistor which is shared with the feedback network.
So my question is : why should the insertion of the buffer between the feedback network and the emitter of the input transistor should change the mode of this transistor from common emitter to common base ?
Note that the emitter voltage of the input transistor is only very slightly modified when the buffer is added (however with a single transistor, its emitter resistor has a great effect on the feedback open loop gain).