Talk:Circuit Idea/How to Reverse Current Direction

Not a diode
Article says ''But it is very primitive and confusing to say "the input transistor Q1 is a diode". Actually, it is not a diode; it is exactly a transistor operating in the active mode. It would be a diode, if its collector was disconnected.'' I'd be inclined to notice that the CB junction of the transistor is short-circuited in Q1. The collector node is attached directly to the base node. Therefore the circuit "sees" only the BE junction, which is, of course, a diode. Sounds like a simple explanation to me. Brews ohare (talk) 00:37, 2 March 2008 (UTC)


 * Brews ohare, thank you for the response. It is wonderful that there are still (wiki)people showing willingness for discussing circuit phenomena! As you can see, I not only try to "humanize" Wikipedia articles by means of these discussions; I draw also inspiration from these conversations that I need to continue disclosing the mystery of electronic circuits on the pages of Circuit Idea. It will be so marvelous, if more key-wikipedians (e.g., Omegatron, Alfred Centauri, Light current, Heron, Rogerbrent, etc.) join this discussion! Of course, we have to be bold and fair enough for the purpose of such a frank discussion; we have not to hide behind "reputable" sources.


 * As far as I remember, I have also to reply to your comment in Wikipedia's Current mirror. Below, I will try to expose my viewpoint about this interesting topic.

What can be the current mirror building component?
I'm not familiar enough with microelectronics; I try to look at electronic devices from a macro- instead from a micro-level. From this viewpoint, the specific current mirror implementation is not so interesting for me; I try to "catch" the general idea behind it. Following this approach, let's see what an electronic device we can use as a building component to realize the two parts of the current mirror.

If you share my penetration, the so called transconductance (voltage-to-current conversion) is the main, general and inherent property of the device. It is not so important, if the relation between the output current and the input voltage is linear; but it is important for the device to keep up a steady output current when the load varies. Simply speaking, we need a voltage-controlled current-stable resistor that changes its present resistance contrary to the input voltage: if we raise the input voltage, the resistance decreases and v.v. This property allows us to build the input current-setting (programming, reference) part of the current mirror by applying a negative feedback. For this purpose, we just connect the device's output to its input, in order to reverse it (to transmute it from a voltage-to-current converter into the opposite current-to-voltage converter).

Almost all the electronic devices (tubes, bipolar-, MOS and junction field-effect transistors, the so-called transconductance amplifiers, etc.) are transconductors that control the current flowing through the load (if we want they to produce a voltage, we connect in series a resistor acting as the simplest current-to-voltage converter). But only two of the "conventional" active electronic components (the bipolar transistor and the E-MOS field-effect transistor) possesses the needed input-output relation.

MOSFET as a current mirror building component


It is exactly what we need since a MOSFET has a true voltage input behaving just as an open circuit. As a result, there are no any problems, if we connect the M1's drain to its gate (Fig. 1). There is no additional current to be extracted from the input source (there is no error because of the input currents); there are no any "diodes" connected in parallel to the device's output... There is only a regulating element (a voltage-controlled resistor) that changes its present resistance (between the drain and the source) so that to keep up an almost constant voltage. We have made it behave in this way - as a voltage-stable resistor, as a forward-biased diode! There are no any other elements in this arrangement...

An E-MOSFET has a threshold in its |transconductance curve but it affects only the output, not the input (although the threshold voltage is exceeded, the input resistance remains infinite). So, we can think of an E-MOSFET as a kind of "transdiode".



BJT as a current mirror building component


But BJT is a "nastier" and more confusing device... First, it is a double-faced component; we can think of a bipolar transistor simply as a current-to-current converter (a current amplifier) or, more precisely, as a voltage-to-current converter (according to Ebers-Moll model). Well, we might accept here the second viewpoint... But what do we do with the base-emitter diode? What is its role here? Is it important? Is the transistor T1 with a shorted CB junction a diode? What is it?

Yes, there is a diode in this arrangement... You are right, the base-emitter junction is really a true diode... But this is only the first diode. Because there is another "diode" - a regulating element (a voltage-controlled resistor) that changes its present resistance (now, between the T1's collector and the emitter) so that to keep up an almost constant voltage. Again, we have made it behave in this way - as a voltage-stable resistor, as another forward-biased diode!

So, in this arrangement, we have two diodes - the first (base-emitter junction) is genuine, the second (collector-emitter part) is artificial. And these two diodes are connected in parallel to each other. Only, the BE diode is low-power (driving) while the CE "diode" is powerful, "executive". The BE diode consumes only 1/(1 + β) portion of the whole current. The CE part serves as a shunting regulating element that diverts the great amount (β/(1 + β)) of the current. So, what the circuit "sees" is not only a diode, it "sees" a shunted, powerful diode... We can even say that since the CE shunting "diode" predominates over the BE true diode (it is β times more powerful than the BE diode) the circuit "observes" only the CE "diode"!

Well, Brews ohare, that's enough; I suggest correcting the original article text from

''...But it is very primitive and confusing to say "the input transistor Q1 is a diode". Actually, it is not a diode; it is exactly a transistor operating in the active mode. It would be a diode, if its collector was disconnected...''

to

''...But it is not sufficient to say "the input transistor Q1 is a diode" since it is not only a diode; it is a "buffered diode". It would be a bare diode, if its collector was disconnected and only the base-emitter junction was connected...''

or to something similar.

True diode as a current mirror input part
Brews ohare, I would like to ask you to take up a position on another "scenario" for building BJT current mirror. In this story, we may first use a true diode as a current-setting input part and, as usual, a BJT as an output part. Then, we will note some disadvantages of this arrangement that will make us use another BJT as an input part.

This idea dawned on me when I, looking for current mirror materials on the web, found the Tony Kuphald's Current mirrors page. He has used such a succession; only he has not given reasons for such a circuit evolution. I regard with respect his web materials; IMO they are the best written popular circuit stories. I know him and I have the intention of inviting him to join Circuit idea stories.

Only, I think there is something confusing and wrong in his material. Can you read thoroughly this material and then give your opinion on it? Do you see something wrong in this arrangement - to connect a bare diode in parallel to the T2's input? Can such an arrangement exist at all? Can it serve as a current mirror?

If we answer these questions, we can show in another credible way the reason of using an active diode. Then, we might invite Tony Kuphaldt to join this discussion, to edit this page or to write new one.

I have an opinion but I prefer first to see what you think about it.

Regards, Circuit-fantasist (talk) 18:35, 2 March 2008 (UTC)

Vout of Input Part is const
I'm confusing about this point: If this Vout of input part is a const and won't be changed with the input currrent, then the output current won't be changed also as this voltage is Vbe of output part and therefore will decide the output current. So the result is if we change the input current configuration resistor, the output current won't follow this change;

Would you please give me a more detail explanation? Thanks!


 * Hi, your question is interesting and important. Vout (or VF of an ordinary diode) is relatively constant. In some cases, e.g. in a case of a voltage stabilizer or if the second device has a large operating range, we assume it is constant because its variations are insignificant in comparison with the voltage range. In the case of a current mirror, we assume it depends logarithmically on the input current (see Fig. 8) in comparison with the transistor's input voltage range. Circuit-fantasist (talk) 16:03, 15 February 2009 (UTC)

The FB-based reverse is incomplete
It is amazing how one can decompose an initially miraculous circuit into principal blocks. Yet, thinking in terms of feedback causes trouble when trying to understand the metamorphosis turning the transistor into its inverse. This book helped me to understand that the feedback not only a way to keep the output at specific level but also a lookup of the input that produces given output. Normally, "internal input" i is adjusted by i = i+Δi, where Δi = ref - o is the difference between primary input ref and "internal output" o. In your example, the current (and collector voltage) is the reference input and base voltage is the internal input. But I see no comparator. Instead, I see that the internal output controls our golden reference, the input current! You see, instead of matching with the golden signal, feedback controls it. Next, the two drivers, the primary input and our transistor converter, must fight -- who is stronger (explained in Commutation cell). Analysis reveals that the mirror reference input always wins because, when the transistor is too closed, the current can go through the base (and open it). It will also close the transistor if the collector voltage is too low for the given current. The solution comes from understanding the transistor operation while the feedback view seems only to cause trouble. The book is otherwise great, yet, I feel like it lacks a last remark to explain the trick of actual FB implementation. Or, is it supposed that the audience is of competence to guess itself? --Javalenok (talk) 14:39, 17 October 2009 (UTC)


 * Javalenok, your thoughts are very interesting for me. I noted your insertion this evening; so, please let mi scrutinize them. If you want, you may join as well another interesting talk - Wikipedia discussion about Emitter-coupled logic where I appear as Circuit-dreamer and sometimes with my old WP user name Circuit-fantasist. Regards, Circuit-fantasist (talk) 20:10, 21 October 2009 (UTC)


 * Javalenok, I have found a very interesting source (page 12 and 16) that seconds this current mirror presentation. Circuit-fantasist (talk) 22:40, 24 October 2009 (UTC)