What is the name of this topology?

[seanvn]

What is the name of this RC coupled differential amplifier topology.
I have tried using google but that was unsuccessfully. I just need the right keyword for it so that I can do some research.
I have used this circuit very successfully in a direct conversion receiver both as a RF and AF amplifier. It does seem to have very unique characteristics. I have seen a complementary feedback pair (CFP, Szilaki pair) version of it used in a high performance preamplifier. I just simplified it but it still works very well.
Can you tell me anything about it, including distortion levels, noise etc.

Sean O'Connor

 

[AndrewT]

Q2 operates with two outputs. The Pr1 output is from the common emitter amplifier, gain is set by R8 & R4 ~1times.
The Pr2 output is from the common collector amplifier, the gain is ~1times, This is usually called an emitter follower.
The combined out of phase outputs from the single input is sometimes called a phase splitter. This is very common in tube (valve) amplifiers.

The second stage around BC108 is a common base amplifier. It works to very high frequency and the gain is set by R9 & R5. I think this is also ~1times. This is often referred to as a cascode.

You have two amplifiers:
phase splitter and cascode

 

[seanvn]

It is not so simple. The outputs at Pr1 and Pr2 are equal in magnitude but 180 degrees different in phase. The gain is about 150 when R7=0, it drops to 2 when R7=R4,R5.
If you make R7 much greater than R4,R5 then it ceases to be a differential amplifier.
I would suggest simulating it or building it to see how odd it is.
Anyone else any ideas?

 

[seanvn]

So basically T1 is not acting as acting as a phase splitter as you might easily imagine. There is actually voltage amplification at R8 (equal to that at R9). I should say I used different transistors just to see if transistor mismatch had any effect. The answer to that is no. Curious and curiouser.

The original circuit I took it from is here:
http://sound.westhost.com/project66.htm

 

[DRC]

With C1 as an effective short, this is just a LTP variant - I think this is quite popular in IC amps.

dc

 

[sawreyrw]

AnderwT,

The gain of the second transistor is gm*R9 where gm=40*Ic and Ic is the DC operation point.

Rick

 

[jcx]

with a constant ref on the Q2 base it isn't working as a diff pair

could be considered a "folded" Rush Cascode

 

[Mooly]

with a constant ref on the Q2 base it isn't working as a diff pair

Correct... it's visually drawn to look like an LTP but has nothing to do with that topology. The Rod Elliot circuit is however a differential amp.

The circuit here is just a two stage design with the second device as a common base amp which are very common in RF amps/tuners etc as a first stage to match to the aerial.

http://users.ece.gatech.edu/mleach/ece3050/notes/bjt/cbamp.pdf

 

[DF96]

This is an emitter-coupled amplifier. The valve analogue is called the cathode-coupled amp. It is a form of differential amp. Differences from a normal LTP are: not necessarily balanced, often quite a 'short' tail, often AC coupling, not necessarily the same transistor type. To a first approximation all BJT behave in exactly the same way: exponential response. That is why changing the transistor had little effect. A normal LTP uses a matched pair in order to maintain DC balance.

The ECA will typically have lower even-order distortion than a common-emitter amp. As the output is in phase with the input it might have greater risk of oscillation due to capacitive feedback.

 

[seanvn]

Thanks for the information. So if you call the 2 transistors T1 and T1 the points T1-R4 and T2-R5 are virtual ground. Hence if C1 is large enough at the signal frequency the emitter load each transistor sees is (R4 parallel R7) and (R5 parallel R7) to a first approximation. Hence there will be over unity amplification at each collector.
Any ideas about Miller capacitance, noise level. A pointer to a research paper would be helpful but sometimes there are no papers! Hey, I know that.

 

[DRC]

I still think the analysis will be, more or less, the same as an LTP where the tail resistor is R4/R5 in parellel and Re is R7/2. Assuming C1 is large enough, and of course, the dc analysis is different.

dc

 

[AndrewT]

no.
The separate tail sinks defeat the LTP action.

An LTP uses the constant current draw of the tail to influence how the other half amplifies it's input signal.

 

[DRC]

An LTP uses the constant current draw of the tail to influence how the other half amplifies it's input signal.

thats what R7 + C1 do !

 

[tiefbassuebertr]

This is an emitter-coupled amplifier. The valve analogue is called the cathode-coupled amp. It is a form of differential amp. Differences from a normal LTP are: not necessarily balanced, often quite a 'short' tail, often AC coupling, not necessarily the same transistor type. To a first approximation all BJT behave in exactly the same way: exponential response. That is why changing the transistor had little effect. A normal LTP uses a matched pair in order to maintain DC balance.

The ECA will typically have lower even-order distortion than a common-emitter amp. As the output is in phase with the input it might have greater risk of oscillation due to capacitive feedback.

I can also call that emitter coupled cascode (another form of the good known normal cascode or folded cascode).
check out US Patent 6,600,367
http://www.patentstorm.us/patents/6600367/description.html
http://www.pat2pdf.org/patents/pat6600367.pdf
and
Information Bridge: DOE Scientific and Technical Information - Sponsored by OSTI

This topology I get often automaticly, when I transform a power amp from the non inverted mode to an inverted mode.

But an additional fact is of interest:
After replace R4 and R5 each through current sources (two independend current sources instead commonly used one pcs. for both halves) and R7 through a potentiometer, you get the possibility for easy variation of the open loop gain/damping factor from your whole power amp stage.

 

[DF96]

no.
The separate tail sinks defeat the LTP action.

Get rid of R7 and you have a short-tail pair. The tail is just R4 and R5 in parallel. Obviously this doesn't go down to DC because of the capacitor.

 

[cbdb]

A diff amp needs a diff input. T2s base is ac grounded. You have 2 outs, the first from a CE amp, the second from a 2 stage amp (cascode) CC then CB (common base).
Thats the way it should be analized, not as a diff amp which is possible but thats over complicating things for no reason. (draw the second transistor rotated 90 degrees clockwise, the usual way to draw a CB amp, and maybe it will be easier to understand)

 

[tiefbassuebertr]

A diff amp needs a diff input. T2s base is ac grounded. You have 2 outs, the first from a CE amp, the second from a 2 stage amp (cascode) CC then CB (common base).
Thats the way it should be analized, not as a diff amp which is possible but thats over complicating things for no reason. (draw the second transistor rotated 90 degrees clockwise, the usual way to draw a CB amp, and maybe it will be easier to understand)

Yes, the kind of drawing often determines, how well you understand the right working of a certainly circuit.
Independend of this - despite of the fact of grounding the inverted input it is still a diff amp - then the different between non inverted input and GND.