john curl said:
This was not my intention... My response was mainly about the gain.
john curl said:
Ignore the input stage in my posted circuit, i was only comparing the vas, one using "rush cascode" the other using diff amp.
(okay, the thread got off topic)
When Kanwar introduced the rush cascode he mentioned the low psrr as main benefit at the cost of difficult biasing (like Christer already said). Also, it has very low 3rd harmonic.
I had following distortions for 12v output:
rush: 2nd : 200mv, 3rd: 1mv
diffamp: 2nd: 1mv, 3rd: 10mv
About the noise, i can only tell about the diff amp version (haven't built rush cascode), it is very silent, noise is only audible by pressing the ear to the tweeter.
In the attachment is freq response for the rush thing, green is gain in db, red is phasehift. (phasehift minus the phaseshift caused by input filter)
Mike
Attachments
john curl said:
And it is appreciated by me, and surely many others. A good teacher knows when to ask challenging questions to make people try to think for themselves. You have done it on previous occasions too when there has been someting interesting to take note of in a circuit or more general theory about circuits.
So, John or anybody else, is there somthing to what I said about potential thermal runaway, or am I thinkng wrong somewhere?
Hi, Mike,
Differential VAS is smart, it overcome the shortness of ordinary single transistor VAS, it magnifies the difference/dV of input differential legs, instead of absolute voltage drop of input differential legs.
I think the usage of Rush cascode is quite tricky. Like the example of your post#31, the performance of this Rush VAS will depend very much on the input (drop of the 680ohm resistors) and how stiff (how good/clean) you make the 2V3 reference towards +rail point.
What do you think about this modification, Mike? Trying to squeeze the capability of this Rush VAS.
The 680ohm resistors of the differential legs is divided into 2x340ohm and the center is bootstrapped towards emitors (220uF).
And between emitor-emitor of Rush Cascode we put R here.
Will this make the performance better?
Differential VAS is smart, it overcome the shortness of ordinary single transistor VAS, it magnifies the difference/dV of input differential legs, instead of absolute voltage drop of input differential legs.
I think the usage of Rush cascode is quite tricky. Like the example of your post#31, the performance of this Rush VAS will depend very much on the input (drop of the 680ohm resistors) and how stiff (how good/clean) you make the 2V3 reference towards +rail point.
What do you think about this modification, Mike? Trying to squeeze the capability of this Rush VAS.
The 680ohm resistors of the differential legs is divided into 2x340ohm and the center is bootstrapped towards emitors (220uF).
And between emitor-emitor of Rush Cascode we put R here.
Will this make the performance better?
Attachments
Mike,
the Rush cascode is usually used in the input stage, in the same way as a diff pair, with one base being the input and the other base being the feedback input. In Kanwars circuit one of tha bases is held at constant voltage, so it is more like an ordinary cascode. That won't give any more gain (speaking of transconductance), since the input voltage will be "split" over two Vbe drops. As far as I can see that means we get half the transconductance compared to an ordinary CE stage, with or without a cascode. In this case it is a diff pair where each half is (maybe) a Rush cascode, but that doesn't change anything. Such a diff pair should have half the transconductande of an ordinary one, with or without cascodes, I think.
Forr,
Yes, as we discussed in the other thread, the name Rush cascode is not very appropriate, I think, but seems to be a common name for the circuit. I do think the folded cascode is an appropriate name though. However, although some of us like things to be well structured and also having the most appropriate and well-descriptive name, the world isn't like that. Both ordinary words in the language and technical terms appear when somebody finds a need to give a name to something. Often they don't choose a very good name for it, but the name starts spreading. Sometimes new and better terms are proposed and manage to replace the earlier one, sometimes not. Such examples are plentiful in all languages, but often we are so used to them that we don't think of it. Do the british still put on ther MacIntosh and their Wellingtons when it is raining?
the Rush cascode is usually used in the input stage, in the same way as a diff pair, with one base being the input and the other base being the feedback input. In Kanwars circuit one of tha bases is held at constant voltage, so it is more like an ordinary cascode. That won't give any more gain (speaking of transconductance), since the input voltage will be "split" over two Vbe drops. As far as I can see that means we get half the transconductance compared to an ordinary CE stage, with or without a cascode. In this case it is a diff pair where each half is (maybe) a Rush cascode, but that doesn't change anything. Such a diff pair should have half the transconductande of an ordinary one, with or without cascodes, I think.
Forr,
Yes, as we discussed in the other thread, the name Rush cascode is not very appropriate, I think, but seems to be a common name for the circuit. I do think the folded cascode is an appropriate name though. However, although some of us like things to be well structured and also having the most appropriate and well-descriptive name, the world isn't like that. Both ordinary words in the language and technical terms appear when somebody finds a need to give a name to something. Often they don't choose a very good name for it, but the name starts spreading. Sometimes new and better terms are proposed and manage to replace the earlier one, sometimes not. Such examples are plentiful in all languages, but often we are so used to them that we don't think of it. Do the british still put on ther MacIntosh and their Wellingtons when it is raining?
Luamanuw,
Just remember that if you use a diff pair in the VAS, you must have enough gain so the diff voltage at its inputs is still very small. While a diff pair is very linear for small differential voltagas, it quickly becomes very nonlinear when the voltage increases. Of course, emitter degeneration helps to extend the linear region at the expense of less gain. The same holds for diff pairs at the input, of course.
Of course, those who don't care about non-linearities after the input stage and think that a heavy dose of global feedback is a cure all can ignore what I said.
Just remember that if you use a diff pair in the VAS, you must have enough gain so the diff voltage at its inputs is still very small. While a diff pair is very linear for small differential voltagas, it quickly becomes very nonlinear when the voltage increases. Of course, emitter degeneration helps to extend the linear region at the expense of less gain. The same holds for diff pairs at the input, of course.
Of course, those who don't care about non-linearities after the input stage and think that a heavy dose of global feedback is a cure all can ignore what I said.
http://www.diyaudio.com/forums/showthread.php?postid=28775#post28775
I just tried modelling the "rush circuit" in LTSpice, and gave up!
Edit: - NAD3020 circuit - sorry throwaway remark.
I just tried modelling the "rush circuit" in LTSpice, and gave up!
Edit: - NAD3020 circuit - sorry throwaway remark.
Attachments
Hi, Christer,
I've made something similiar, and it's difficult to stabilize like JC clued http://www.diyaudio.com/forums/showthread.php?s=&threadid=37866&highlight=
Kanwar makes 2 kind of cct, one is cascoded differential VAS, the other is this Rush cascode VAS. I'm not sure which Kanwar's cct you looked at, but the Rush cascode one (emitor meets emitor) will have big gain.
That's a clue of big gain stage
I've made something similiar, and it's difficult to stabilize like JC clued
http://www.diyaudio.com/forums/showthread.php?s=&threadid=37866&highlight=AndrewT said:
I suppose raincoats are taking over, which illustrates may point that sometimes newer and more appropriate terms appear and manage to replace the old ones.
I only used the british examples because english is our common language here on the forum. There are plenty of such things in all languages, I think, not the least in Swedish. Long ago, express trains were called snälltåg in Sweden, and that would literally mean a train that is nice (not nice in the sense of being nice to travel with but nice as when you talk about a nice person, who doesn't behave bad). So, did we have bad trains, or rude trains too? No, somebody had made a mistranslation of the german term Schnellzug, meaning a fast train. How such a stupid translation can spread and even be used by the railways is beyond me.
lumanauw said:
Take an ordinary CE stage. The transconductance of it is gm. Add a cascode to it, and the transconductance is still gm, since the cascode CB stage has no current gain. If you instead use a Rush cascode where you hold one of the bases at fixed voltage, you get a transconductance of gm/2.
Now take an ordinary diff pair. The current in either transistor is Vd*gm/2, where Vd is the differential voltage between the bases. Instead, take a Rush cascode used as a diff pair, with a differential voltage Vd between the bases. Same thing, the current is Vd*gm/2.
Then take an ordinary diff pair and add cascodes. Just as for the single CE stage, the cascodes do not change the transconductance, so the current in either half is Vd*gm/2. Swap this for the "diff pair" where each half is a Rush cascode with the upper bases held at fixed voltage. Each half has a transconductance of gm/2, in contrast to gm for the ordinary diff pair. Hence, the current in either half is Vd*gm/4.
Please correct me if a made any mistakes. And of course we are talking small signal analysis here, to not make things even messier.
Hi, Christer,
A little miscommunication, you are thinking about Rush Cascode (RC) for input stage, while I'm thinking about RC for VAS stage, but.....
You're right One thing that I don't realize is that in RC, the dV will be divided for 2 dVBE, because there are 2 BE junctions in series (like in drawing 3, d=20mV will be divided by 2, that is d=10mV for each BE junction in series RC).
It seems the double RC (drawing 3) will have the same gain as drawing 1 (simple CE).
Differential VAS (drawing 2) will have double gain.
PS : Sorry, I'm not EE, I try to understand these things by stone-age iteration. I imagine for full swing, the VAS will have to be driven by dV=20mV. You EE have the simpler way, one word "gm"
So, what is the advantage and disadvantage of Rush Cascode? It don't have current limitation (limited by CCS), bigger transient current capability.
Differential VAS (drawing 2) has good PSRR because it only sense dV, disregarding much of rail ripple.
How about looking at RC not from gain POV, but from other POV, like PSRR?
A little miscommunication, you are thinking about Rush Cascode (RC) for input stage, while I'm thinking about RC for VAS stage, but.....
You're right
One thing that I don't realize is that in RC, the dV will be divided for 2 dVBE, because there are 2 BE junctions in series (like in drawing 3, d=20mV will be divided by 2, that is d=10mV for each BE junction in series RC). It seems the double RC (drawing 3) will have the same gain as drawing 1 (simple CE).
Differential VAS (drawing 2) will have double gain.
PS : Sorry, I'm not EE, I try to understand these things by stone-age iteration. I imagine for full swing, the VAS will have to be driven by dV=20mV. You EE have the simpler way, one word "gm"
So, what is the advantage and disadvantage of Rush Cascode? It don't have current limitation (limited by CCS), bigger transient current capability.
Differential VAS (drawing 2) has good PSRR because it only sense dV, disregarding much of rail ripple.
How about looking at RC not from gain POV, but from other POV, like PSRR?
Attachments
Hi Christer,
I know the Rush circuit for very long, the first time I remember seeing it was in Cherry's NDFL amplifer and then, I saw it used some times with different technologies (tubes, mosfets and bipolars). Untill recently, nobody called it a cascode.
I am sorry to see that a recent misinterpretation (where every circuit using a common base transistor is thought to be a cascode) has lead so many people on this forum to use the expression "Rush cascode".
CC-CB diagram of Kyrgeo's post #1 may also be called a twin emitter coupled circuit (thanks to Mikeks for "emitter coupled"). Here is another emitter coupled (or CC-CB) circuit which is far easier to correctly bias and understand. The left or the right differential pair will be very familiar to everybody. Has anybody ever called it a "cascode" ?
I know the Rush circuit for very long, the first time I remember seeing it was in Cherry's NDFL amplifer and then, I saw it used some times with different technologies (tubes, mosfets and bipolars). Untill recently, nobody called it a cascode.
I am sorry to see that a recent misinterpretation (where every circuit using a common base transistor is thought to be a cascode) has lead so many people on this forum to use the expression "Rush cascode".
CC-CB diagram of Kyrgeo's post #1 may also be called a twin emitter coupled circuit (thanks to Mikeks for "emitter coupled"). Here is another emitter coupled (or CC-CB) circuit which is far easier to correctly bias and understand. The left or the right differential pair will be very familiar to everybody. Has anybody ever called it a "cascode" ?
Attachments
Forr,
a name is just a name, not necessarily descriptive of what it stands for, although it is often nice when they have that property. My real name actually is Christer, but does that tell you who I am or what I look like? I do agree, though, that one should attempt to find good descriptive terms when possible, and sometimes make an effort to replace less fortunate terms.
What about simply starting to say "the Rush circuit", or"the Rush pair"? Then most people who know about it should understand, I think, and there is no pretending that it is a cascode circuit.
As for the folded cascode, I still think that is a reasonable name, but there is never a single universal truth in such matters. We can only have opinions. Names can also be more or less reasonable depending on from which point of view you are looking at them. Suppose someone had invented the folded cascode before the ordinary one? Then maybe that would have been called a cascode, and the usual one be called an unfolded cascode? Better start barking at all those people who mix up cascade circuits with cascode circuits. That is much more confusing.
I'm trying to restrain myself, but... no, I think it is time to qoute Chomsky:
It is a bit contrived, and was surely meant to, but he is making a point by overemphasizing in that way.
a name is just a name, not necessarily descriptive of what it stands for, although it is often nice when they have that property. My real name actually is Christer, but does that tell you who I am or what I look like? I do agree, though, that one should attempt to find good descriptive terms when possible, and sometimes make an effort to replace less fortunate terms.
What about simply starting to say "the Rush circuit", or"the Rush pair"? Then most people who know about it should understand, I think, and there is no pretending that it is a cascode circuit.
As for the folded cascode, I still think that is a reasonable name, but there is never a single universal truth in such matters. We can only have opinions. Names can also be more or less reasonable depending on from which point of view you are looking at them. Suppose someone had invented the folded cascode before the ordinary one? Then maybe that would have been called a cascode, and the usual one be called an unfolded cascode? Better start barking at all those people who mix up cascade circuits with cascode circuits. That is much more confusing.
I'm trying to restrain myself, but... no, I think it is time to qoute Chomsky:
It is a bit contrived, and was surely meant to, but he is making a point by overemphasizing in that way.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.