The constant current source

July 29, 2022
 by Paul McGowan

Some of the most important circuit technology in audio is also the simplest. Take for example the constant current source.

If we look at a simple single stage amplifier it is pretty straightforward: a single transistor (Q1) and a few resistors. Depending on the value of those resistors, we can get a certain amount of gain when we input an audio signal—small signal in and we get at the output a larger version of that signal. Easy peezy. This is the basis of how we started designing discrete amplifier circuits.

Add a second transistor (Q2) and another few resistors and that single-stage amplifier becomes a differential pair. Feed the output of that differential pair into a third transistor (Q3) and its associated resistors and voila! We have made a simple discrete op amp. Lastly, add a few more transistors (Q4 and Q5) at the output of this simple amp so that they provide more current (and protect the sensitive amplifying transistors before it), and now we have a great sounding discrete op amp: the original basis of all PS Audio phono and analog preamplifiers. (The two diodes D1 and D2 set the bias for the little output stage).

 

Good sounding, yes, but with a very simple addition, this circuit can sonically open up and sing like you would not believe. In fact, the simple addition of yet another transistor turned this wonderful little circuit into something rather extraordinary (at the time).

This was the 1970s at the very beginning of our journey and all this discrete analog stuff was heady indeed. We were pioneers foraging our way through the weeds of uncharted territory. A fellow engineer or audio nerd would whisper in our ear about some new discovery (like bypassing electrolytic caps with small film caps) and everyone in our tight little circle of nerds would pounce on it to see if a difference could be heard. If it mattered, it became audio gospel.

What we knew was that those first two transistors that make up the input diff pair were not ideal. Because the transistors were connected to the power supply through mere resistors, the rising and falling input/output signal forced those transistors into constantly fluctuating with every string pluck on a recording. What we would love is a way to have a constant current (like a class A bias) running through those two devices. If that were possible, we could fix the little circuit’s operating point for whatever sounded best and be assured it would never change with louder or softer music.

That was the holy grail.

Someone whispered the answer in our ears. “Shhhh….a constant current source will do it. A single transistor, a couple of diodes and a resistor or two and voila!”

No shit. Wow.

Look at the first diagram. See R2 (47K)? If we were to replace that resistor with a specially configured transistor, we could set the level of constant class A operation on the diff pair. Here’s something similar.

This is a bit simpler because of their use of an LED instead of two signal diodes to bias on the transistor, but the idea is the same. The LED as fed from the resistor RB turns on the transistor so it starts drawing current. The lower resistor, RE, determines just how much current is constantly flowing through this transistor. Simple.

Today, nearly 50 years later, this is just the way you do it. We would never consider using a simple resistor that lets the current flop around with the musical signal.

But,. back then, it was a miracle!

Subscribe to Paul's Posts

35 comments on “The constant current source”

    1. +1

      No more ‘Neighbours’.
      After 37 years, the very last episode finally aired here last night…thank god!
      Now we can all go & ‘get a life’ or start watching ‘The Bold & The Beautiful’.

  1. Paul, Too big of a leap between top circuit and bottom circuit. First, why do you need two diodes ( D1 and D2 )? They are identical and in series. Each only passes current one way. Second, how does an LED differ from a diode in the circuit? How does an LED preform eclectically?

    1. @tonyplachy,

      I know you asked @Paul, but I can help. You need either two rectifier diodes or one LED to get about the same (relatively) constant forward voltage drop. Rectifiers have a Vf around 0.7V and LEDs (depending on color) are around 1.8 to 2.2V. Transistor Vbe is similar to a rectifier (0.7V). So the additional voltage above ~0.7V provides the necessary voltage to turn on the transistor and set the current through Re. Current ~ 0.7V/Re for the rectifier case and a bit higher for the LED version.

    2. Yes, kcleveland has it right. The two diodes are needed to turn on the transistor. One diode is 0.6v which only matches the diode drop of the transistor’s own. So another is added so the base of the transistor is pulled far enough above the emitter to turn on the device allowing current to flow. The emitter resistor is what sets the amount of constant current.

      You can use an LED as well depending on the color. Each LED type is too a diode but instead of only 0.6V drop LEDs have more—enough more to rise above the base emitter junction of the transistor to turn it on.

      In actuality, today we use neither. Instead, we use a second NPN transistor. http://i.stack.imgur.com/PEm2I.png

  2. Your explanation reminds me of Bob Pease’s columns in EDN back in the day. I used to turn to Bob’s column first thing when EDN landed on my desk. I had a filing cabinet drawer full of his columns that i kept for reference.
    You both have a way of explaining things that cuts through a lot of the bull to get to the heart of things.

    1. Boy does that take me back! I also read just about every one of Bob Pease’s articles in Electronic Design. Lots of “knowledge nuggets” from Bob and was saddened to hear of his passing many years ago (hard to believe that was in 2011). There was also an especially bright and excellent teacher of difficult stuff named Jim Williams who used to work for Linear Tech. I once attended a seminar that he taught and was in awe at this ability to take very complex stuff and boil it down to its essence. Just noticed Jim died in 2011 too! And check out this crazy coincidence from Wikipedia “Pease was killed in the crash of his 1969 Volkswagen Beetle, on June 18, 2011.He was leaving a gathering in memory of Jim Williams.”

      Great memories for sure.

  3. Its been a while since I’ve seen hand drawn schematics. Nice!
    Simple schematic topology yes. What makes one different over another is little details like the transfer characteristics of the transistors which change as the device heats up. And then the other dirty little secret was to use a “high speed” low-impedance regulated linear power supply to keep the + / – rails stiff even under heavy current draw.
    The 1N4148 was pretty new to consumer gear in the 70s-80s-90s. It had slightly improved characteristics over the 1N914 and cost a little bit more. They pretty much are the same thing now.

      1. I think the 1N4153 is now obsolete. I have them in the clipping circuit in my Big Muff. I believe the 4153 has lower capacitance than the 4148…

  4. I feel like I’m back in engineering school. It’s nice to have a post like this once in a while.

    It will probably take me a while to digest this analog circuit that these two commenters and Paul have been discussing but I’m gonna have a lot of fun reviewing the schematic slowly.

    Thanks Paul

    1. I’m staring at that second circuit diagram, and I see an arrow representing current flowing in. Where is the output current arrow? See, I have no idea what’s going on.

        1. Thanks Tim. So the output arrow is above resister RB. In the circuit loop containing RE and LED1, are the electrons moving clockwise or counterclockwise?

          1. With regard to transistors, there are two types… NPN and PNP and there are times when you need to use one type or the other in your particular design. The three terminals are called the Emitter, Base and Collector. In this case the current flows from the Collector and is controlled by the Base bias allowing current to flow down through the Emitter to the Emitter Resistor which is the output of this circuit. That really doesn’t explain how the circuit operates or exactly what its function is but it may give you some idea of basic transistor terminology. Hope that I didn’t confuse you even more and when you asked your question.

            1. Thanks stimpy2. Knowing that RE is the output is a great help to me. It now makes sense. Maybe if the circuit had a picture of a speaker driver instead of a resister, this dummy would have gotten it. LOL

              1. Maybe Paul could create and sell an educational video like he did for the PowerPlant, in which he shrinks himself down and rides on electrons through transformers, resistors, capacitors, vacuum tubes, diodes, transistors, Jfets and other circuit components so dummies like me can understand how they work.

                1. I hope everyone here does realize that except for vacuum tubes electrons don’t flow anywhere. Current is NOT the flow of electrons but the flow of electrical energy. The energy is transferred from electron to electron in the metal wire, but electrons do NOT flow through the wire. So think in terms of current flow, energy flow, but NOT electron flow!

                  1. Tony, thanks for the reminder. Yes, it’s the electrical charge that moves through the circuit, not the electrons. The electrons only move a tiny distance as they carry the charge and transfer it to neighboring electrons. In the video Paul could be a charge (packet of energy) riding electrons for short distances and then jumping from electron to electron. Yes, I know electrons aren’t exactly particles. They are also wavelike, so maybe Paul can surf the electromagnetic wave. LOL

                    1. Clarification to my statement that electrons only move a tiny distance: In a DC circuit free electrons do move (drift), albeit very slowly, through the entire length of the wire as long as voltage is applied. But the transfer of electrical charge from electron to electron happens at near light speed.

                    2. Flow was the term used by instructors to help visualize and trace the flow of the circuit. Current going in one direction and electrons (electron flow) the other. Math comes out the same. Things like Thevenin’s theorem, right hand rule, etc, helps explain things. Truth is we really don’t know why it does what it does. We know how to work with “electricity”, the math works, and theories are used to help explain what is going on.

                    3. I agree, but rather than perpetuating instructors’ incorrect statement that electric current is the flow of electrons, I would rather say electric current is the flow of energy.

                      https://www.youtube.com/watch?v=bHIhgxav9LY

                      Even the above video, talking about Maxwell Equations and Right Hand Rule, doesn’t really explain what happens. A bunch of formulas are just predictors of behavior. Just like Newton’s gravitational formulas don’t explain gravity. They just predict how masses interact with each other.

                    4. Joseph, What a great video! It took me right back to junior year E&M. The energy flow explanation was the most compete I have ever heard. Maxwell, Poynting and Heavyside, what a trio. All of them FRS. They were so brilliant. I had to use a Heavyside function in my thesis work to calculate the self does to a sphere with uniform radiation inside it. I sure brings back memories.

                    5. Tony, I knew this video would resonate with you. Be sure to watch the guy’s sequel in which additional explanations and clarifications are offered.

                      I think ALL audiophiles should watch these videos because they demonstrate how the geometric arrangement and configuration of circuitry can affect the sound, because the electromagnetic field of every wire and every component is interacting with each other, as well as with the electromagnetic fields of other electrical devices in proximity. It explains why component designers need to finesse their circuit board configurations for the most optimal electromagnetic field interactions. How the circuit components are laid out and shielded makes a big difference.It is too complex to figure out mathmatically, so trial and error combined with experience is required for the best sounding designs.

                      Also, for those who think cable design doesn’t make a difference, they should realize how energy is really transmitted. It’s not electrons moving like train cars through a tunnel. It is much more complex than that. So fascinating.

  5. A few years back, I was collecting vintage Fender amps. I came across a box of schematics for many amps from the early 60s and forward. I still have them.

    All tubes in those days. I had studied tube circuit design, but lost interest later as transistors kicked in.

    Cool stuff.

  6. An author must know his or her – or – in this anti “binary” world – it’s – audience. Otherwise, if you present your readers with rational, useful, facts and information, you get the following result:

    “A big Yawn”
    “Just give me the remote!”
    “No more ‘Neighbours’”
    “I feel like I’m back in engineering school”

    Audiophiles – and the pseudo audiophiles collected here – understand, and avidly consume, audiophile tales of lore. Magic and fairies and speaker terminals that’re called “tube connectors” by its seller. The break in time for an amplifier, or a DAC – or for a newly installed electrical service to your house.

    You know – audiophile talk.

    Pontificating about the mystical properties of audiophile bling – not the incomprehensibly factual techniques of circuit design – is what gets audio attention – and discussion.

    You must inject mystical fairy dust into that dry circuit design and theory. Such as, applying the magical green marker used to mystically "improve" the sound of CD's to color the transistors and diodes; or using special audiophile cable to connect all of those incomprehesible PNP and NPN transistors and diodes and resistors. There we go. No we're in audiophile land.

    Rock on.

  7. Paul,

    While in Tech school early 70’s, doing the math on a single gain stage was not my favorite pastime, if you know what I mean. Along came the first HP calculator with reverse polish logic. Using this calculator brought the time spent on math down to 45 minutes. Now look at us, wow. haha..

    1. I still have my HP 15C. Use it all the time, though not nearly as much as I did years ago. Used to have it programmed for impedance calculations, don’t even remember how to do that any more. (Don’t need to, either.)

      1. haha, yeah you and me both, I now have to get out the old ohm’s law pie chart. Had an HP-25 that gave up about 6 years ago. It was my first foray into learning programming in a way. (if I remember right it was purchased at Dillards for $240)

  8. @Paul – you say you would use another transistor rather than an LED for the constant bias voltage now – is this a better config?
    Seems a shame not to use LEDs wherever possible, they are after all. like lasers, never not cool 🙂

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301
1-800-PSAUDIO

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram