Author Topic: Operation of a longtailed pair with collector current mirror  (Read 11091 times)

0 Members and 1 Guest are viewing this topic.

Offline whysoTopic starter

  • Newbie
  • Posts: 1
  • Country: de
Operation of a longtailed pair with collector current mirror
« on: April 29, 2016, 09:10:06 am »
Hey everybody,

I seem to not quite get how this setup show below actually works. Now I have read quite a lot about how a longtailed pair works,
going from one with just two collector resistors and a large emitter resistor as a approximated current source,  to one with a "real" current source for common mode rejection.

Quote
Another improvement to the basic circuit is to employ a current mirror within the collector circuit of the transistors. This enables the differential collector current signal to be converted to a single ended voltage signal without the losses of the resistor while also increasing the circuit gain. The high effective collector load provided by the current mirror enables voltage gains of 5000 or more to be achieved provided there is no external load placed on the circuit.
(http://www.radio-electronics.com/info/circuits/transistor/long-tailed-pair.php)

I can not quite get an intuitive understanding for this.
Let me tell you the way I see it:



First there is a input difference, lets say the base voltage of T1 is a bit higher than that of T2. This increases the collector current of T1 to IC0 + ?.
Since T3 and T4 set up a current mirror, the collector current throw T2 will be forced to that of T1 (IC0 + ?).
Assuming IE is roughly equal to IC for both T1 and T2,  we have a total current of 2 * ( IC0 + ? ) at the constant (!) current source at the emitters of  T1 and T2. So this is the point I am totally confused about,
the current mirror and constant current source kind of set up a contradiction.

I am pretty sure I am missing out on a important point, can you guys help me please? ^-^

Also, can someone clarify that statment for me:
Quote
This enables the differential collector current signal to be converted to a single ended voltage signal [...]
 

Online Performa01

  • Super Contributor
  • ***
  • Posts: 1594
  • Country: at
Re: Operation of a longtailed pair with collector current mirror
« Reply #1 on: April 29, 2016, 10:16:11 am »
The statement about converting in a single ended voltage signal is simply wrong, which is probably causing all the confusion.

The constant current source sets up a total current I.
The sum of currents in T1 and T2 must be equal to I, so  I1 + I2 = I;
We can also say  I2 = I - I1;
The currents in T3 and T4 are equal to I1: I1 = I3 = I4;


The main thing is, the output is very high impedance (as it's basically just two collectors, hence current sources). So it delivers current, not voltage. More specifically, the output current will be the difference between I2 and I4.

On a perfectly balanced differential pair, the output current is zero, because I1 = I3 = I4 = I/2.
And I2 = I - I1 = I - I/2 = I/2;

Since I2 and I4 are equal, there is no "extra current" left for the output, hence the output current is zero.

With maximal imbalance, we get e.g. I1 = I3 = I4 = I;
And I2 = I - I1 = I - I = 0;

Now I2 is zero, but I4 = I, so the output sources a current equal to I.
The opposite would happen if I2 = I and I1 = 0, then the output sinks a current equal to I.

Because of the high impedance output, we can connect a high impedance load, where minimal current changes would give very high voltage changes, thus getting very high voltage gain. Even if the load impedance were the same as the resistor on the collector of T2 (in a diff. pair without current mirror), the gain would still be twice as much, because the signal of T1 is not wasted anymore, as the current mirror enables the output to not only sink current, but also source it.


 

Offline 3db

  • Frequent Contributor
  • **
  • Posts: 331
  • Country: gb
Re: Operation of a longtailed pair with collector current mirror
« Reply #2 on: April 29, 2016, 10:34:03 am »
Hey everybody,

I seem to not quite get how this setup show below actually works. Now I have read quite a lot about how a longtailed pair works,
going from one with just two collector resistors and a large emitter resistor as a approximated current source,  to one with a "real" current source for common mode rejection.

Quote
Another improvement to the basic circuit is to employ a current mirror within the collector circuit of the transistors. This enables the differential collector current signal to be converted to a single ended voltage signal without the losses of the resistor while also increasing the circuit gain. The high effective collector load provided by the current mirror enables voltage gains of 5000 or more to be achieved provided there is no external load placed on the circuit.
(http://www.radio-electronics.com/info/circuits/transistor/long-tailed-pair.php)

I can not quite get an intuitive understanding for this.
Let me tell you the way I see it:



First there is a input difference, lets say the base voltage of T1 is a bit higher than that of T2. This increases the collector current of T1 to IC0 + ?.
Since T3 and T4 set up a current mirror, the collector current throw T2 will be forced to that of T1 (IC0 + ?).
Assuming IE is roughly equal to IC for both T1 and T2,  we have a total current of 2 * ( IC0 + ? ) at the constant (!) current source at the emitters of  T1 and T2. So this is the point I am totally confused about,
the current mirror and constant current source kind of set up a contradiction.

I am pretty sure I am missing out on a important point, can you guys help me please? ^-^

Also, can someone clarify that statment for me:
Quote
This enables the differential collector current signal to be converted to a single ended voltage signal [...]


Iv'e found that asking even a simple question on this forum leads to loads of waffle and confusion.
Have a look on Alan's aka w2aew youtube channel.
You might find these links of use.




3DB

 

Offline danadak

  • Super Contributor
  • ***
  • Posts: 1875
  • Country: us
  • Reactor Operator SSN-583, Retired EE
Re: Operation of a longtailed pair with collector current mirror
« Reply #3 on: April 29, 2016, 01:20:34 pm »
Extensive discussion and analysis here -


http://www.ebay.com/itm/Operational-Amplifiers-Design-and-Applications-NoDust-/161884652925?hash=item25b113c57d:g:UnEAAOSwhcJWQ3Vl


There were 3 more books in the series for a total of 4 in the set.

Regards, Dana.
Love Cypress PSOC, ATTiny, Bit Slice, OpAmps, Oscilloscopes, and Analog Gurus like Pease, Miller, Widlar, Dobkin, obsessed with being an engineer
 

Offline orolo

  • Frequent Contributor
  • **
  • Posts: 352
  • Country: es
Re: Operation of a longtailed pair with collector current mirror
« Reply #4 on: April 29, 2016, 01:43:50 pm »
A current mirror can not PERFECTLY regulate current, it has a limited gain. For a current mirror, it TRIES to balance the current, that means, if it has a very high gain, to compensate the current imbalance, it can generate quite some high voltage difference (hoping to balance the current), and hence it can be used as input gain stage for a diff amp.
This exactly . At the small signal level (hybrid-pi model), the current source from a transistor is in parallel with the ro (Miller) resistance, which is usually very high. Roughly, if the right transistor of the differential pair tries to draw a small current dI more than the left one, that current has to flow through the ro of that side of the mirror, and the voltage difference at that emitter will be -dI * r0, for a very high ro.
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 7934
  • Country: us
  • Retired, now restoring antique test equipment
Re: Operation of a longtailed pair with collector current mirror
« Reply #5 on: April 29, 2016, 02:42:53 pm »
Note that the output (where the collector of a PNP in the mirror connects to a NPN in the pair) is not at a well-defined potential.  Connecting a load with DC continuity to a specific voltage will determine the quiescent voltage at that point.
 

Offline LvW

  • Frequent Contributor
  • **
  • Posts: 282
  • Country: de
Re: Operation of a longtailed pair with collector current mirror
« Reply #6 on: April 29, 2016, 02:53:50 pm »
I am pretty sure I am missing out on a important point, can you guys help me please?
Also, can someone clarify that statment for me:
Quote
This enables the differential collector current signal to be converted to a single ended voltage signal [...]

Let me try an answer in two sentences:
* Applying a differential voltage between the input nodes causes an inbalance between both collector currents.
* However, because the current mirror forces the two currents to be equal, the current difference is forwarded to the next stage and further amplified.
   This is how such a stage works. By the way: The small remaining inbalance within the current mirror (base current) can be compensated by a suitable base current of the next  stage.

That means: Indeeed, the differential signal is converted into a single-ended signal.
(I don`t know why in one post - reply'2 - this feature was classified as "total wrong")

« Last Edit: April 29, 2016, 02:58:03 pm by LvW »
 
The following users thanked this post: 3db

Online Performa01

  • Super Contributor
  • ***
  • Posts: 1594
  • Country: at
Re: Operation of a longtailed pair with collector current mirror
« Reply #7 on: April 29, 2016, 05:43:17 pm »
That means: Indeeed, the differential signal is converted into a single-ended signal.
(I don`t know why in one post - reply'2 - this feature was classified as "total wrong")

This enables the differential collector current signal to be converted to a single ended voltage signal

is wrong, because it's a current, not a voltage.
 

Offline LvW

  • Frequent Contributor
  • **
  • Posts: 282
  • Country: de
Re: Operation of a longtailed pair with collector current mirror
« Reply #8 on: April 29, 2016, 06:30:51 pm »
This enables the differential collector current signal to be converted to a single ended voltage signal
is wrong, because it's a current, not a voltage.

I suppose, you misunderstood the meaning of this sentence.
The difference in currents can be converted into a single-ended voltage - of course not within the first stage.
This was never claimed by anybody. Of course, the next stage performs this conversion.
 

Offline bson

  • Supporter
  • ****
  • Posts: 2265
  • Country: us
Re: Operation of a longtailed pair with collector current mirror
« Reply #9 on: April 29, 2016, 06:59:06 pm »
It's easier to understand if you imagine the input transistors instead are JFETs that are always "on" (until you go well into the negative).  All that changes with gate voltage in a JFET is the channel resistance, which causes the current mirror to swing the voltages over the two LTP inputs to maintain current symmetry.  If one in the T1-T2 pair presents a lower resistance that one will get a lower voltage to throttle the current.  The total current is of course set by the current sink near the bottom.  Note that it's important to not sink the output; any current loss there will be reflected as a voltage drop and hence nonlinearity.  Sinking 100uA on it if the current sink is set to 5mA means 1/50 distortion.  Another JFET for the VAS can be a good idea, with a gate resistor to control gate charge rates.

BJTs work the same except are sensitive to base current rather than voltage.  They need to be preceded by resistors if you want them to respond to voltage instead of current.  For a BJT LTP like yours the resistor input network is a critical piece.  There's also the issue of input base protection.  It's critical that the current sink is faster (has higher bandwidth) than T1, otherwise T1 will see transient Vbe's at rail voltage and won't last for long.  This means T1 needs to be given a reasonably high value base resistor or even a little RC rolloff.
 

Online Performa01

  • Super Contributor
  • ***
  • Posts: 1594
  • Country: at
Re: Operation of a longtailed pair with collector current mirror
« Reply #10 on: April 29, 2016, 07:09:23 pm »
I suppose, you misunderstood the meaning of this sentence.

Fair enough.
I just had the feeling it was exactly this sentence in conjunction with the schematics (that didn't show any load) what got the OP confused.
 

Offline rfeecs

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
Re: Operation of a longtailed pair with collector current mirror
« Reply #11 on: April 29, 2016, 09:30:13 pm »
This circuit is a voltage amplifier.  Voltage in and voltage out.  Not current in or current out.

The thing about converting to single ended is explained pretty well in the Wikipedia article:
https://en.wikipedia.org/wiki/Differential_amplifier#Collector_current_mirror
Quote
Collector current mirror[edit]
The collector resistors can be replaced by a current mirror, whose output part acts as an active load (Fig. 3). Thus the differential collector current signal is converted to a single ended voltage signal without the intrinsic 50% losses and the gain is extremely increased. This is achieved by copying the input collector current from the right to the left side where the magnitudes of the two input signals add. For this purpose, the input of the current mirror is connected to the right output and the output of the current mirror is connected to the left output of the differential amplifier.
The current mirror inverts the right collector current and tries to pass it through the left transistor that produces the left collector current. In the middle point between the two left transistors, the two signal currents (current changes) are subtracted. In this case (differential input signal), they are equal and opposite. Thus, the difference is twice the individual signal currents (?I - (-?I) = 2?I) and the differential to single ended conversion is completed without gain losses.

It's most easy to look at this from a small signal point of view.  That is a small variation about an operating point.

For the circuit with resistive loads, when you go from differential input to single ended output, you lose half the voltage gain (compared to differential in / differential out).  The current mirror restores that by mirroring the current from the left side to the right which causes the two signals to add, so it restores what you lost, making it a better conversion from differential to single ended.

You not only get twice the voltage gain, but also higher gain because of the high output impedance of the current source acting as a load.
« Last Edit: April 29, 2016, 10:37:36 pm by rfeecs »
 

Online Performa01

  • Super Contributor
  • ***
  • Posts: 1594
  • Country: at
Re: Operation of a longtailed pair with collector current mirror
« Reply #12 on: April 30, 2016, 06:14:52 am »
This circuit is a voltage amplifier.  Voltage in and voltage out.  Not current in or current out.

Aha. Can you explain how a high impedance output (whose voltage heavily depends on the load) can be called a voltage output? Maybe you're able to find some other Wiki article that explains that a voltage source is supposed to have a low source impedance, whereas current sources have a high one.

A long tailed pair with current mirror is a transconductance amplifier, voltage difference in, single ended current out.
Consequently, the (open loop) gain of such an amplifier is not expressed in V/V (voltage gain), but A/V (transconductance). The voltage gain can be calculated as soon as the load is defined. The load due to the transistor output impedances is just a parasitic effect, limiting the maximum achievable voltage gain.

And for the rest of your statement, I don't see anything that's not already covered in my reply #2.
 

Offline 3db

  • Frequent Contributor
  • **
  • Posts: 331
  • Country: gb
Re: Operation of a longtailed pair with collector current mirror
« Reply #13 on: April 30, 2016, 07:51:34 am »
This is the beginners section of the forum.
IMHO the reason for designing a circuit like that is so that we can get a HIGH GAIN.
It's obvious that a load would be connected to the circuit,why else would it be built.
Lots of this type circuitry would be inside op amps to achieve high gain.
So whats the point of ranting on about considering it as a current source just because it's Hi Z.
We want to use it's hi gain as a VOLTAGE amplifier.
We should try and keep things simple so beginners gain an understanding of why circuit
configurations have evolved to perform a function.

3DB |O

 
 

Offline LvW

  • Frequent Contributor
  • **
  • Posts: 282
  • Country: de
Re: Operation of a longtailed pair with collector current mirror
« Reply #14 on: April 30, 2016, 08:13:22 am »
This circuit is a voltage amplifier.  Voltage in and voltage out.  Not current in or current out.
The thing about converting to single ended is explained pretty well in the Wikipedia article:

There are more serious and reliable knowledge sources than wikipedia - and I agree to performa01s contribution.
 

Offline 3db

  • Frequent Contributor
  • **
  • Posts: 331
  • Country: gb
Re: Operation of a longtailed pair with collector current mirror
« Reply #15 on: April 30, 2016, 11:14:07 am »
Why is the current mirror there ?

3db
« Last Edit: May 01, 2016, 04:37:06 am by 3db »
 

Offline rfeecs

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
Re: Operation of a longtailed pair with collector current mirror
« Reply #16 on: April 30, 2016, 04:24:35 pm »
This is the beginners section of the forum.
IMHO the reason for designing a circuit like that is so that we can get a HIGH GAIN.
It's obvious that a load would be connected to the circuit,why else would it be built.
Lots of this type circuitry would be inside op amps to achieve high gain.
So whats the point of ranting on about considering it as a current source just because it's Hi Z.
We want to use it's hi gain as a VOLTAGE amplifier.
We should try and keep things simple so beginners gain an understanding of why circuit
configurations have evolved to perform a function.

3DB |O

Exactly.  People use an active load with a high impedance to achieve a high voltage gain.

This circuit has been used for fourty years.  It is a textbook circuit.  They describe how it has high voltage gain, not that it has a current source output.

It was in my college textbook 35 years ago (we studied the earlier edition):
http://www.amazon.com/Analysis-Design-Analog-Integrated-Circuits/dp/0470245999/ref=sr_1_3?s=books&ie=UTF8&qid=1462032712&sr=1-3&keywords=linear+integrated+circuit+design

Every real amplifier circuit has an input impedance and an output impedance, and the gain depends on source impedance and load impedance.

Looking at the circuit as a transconductance amplifier is equally valid.  I have no problem with Performa01s explanation apart from the wording that was a little confusing and he clarified that in his later post.  I thought it helpful to offer another explanation from Wikipedia.  It says basically the same thing, I agree. :clap:
« Last Edit: April 30, 2016, 05:07:40 pm by rfeecs »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf