Not so hard to own one, if like Alan you've been a TEK application engineer for MANY years!

The 576 was long discontinued by the time I started for Tektronix. I was lucky enough to get this unit from a friend a few years ago that was cleaning out a house and found it.

@@w2aew, wow that was an awesome find, and it is one of the newer ones.

I so love the analog slides! The presence of QR-code (and its content) makes me think how organised Alan (probably) is!

Thanks for that - always good to know.

Good point. In recognition that it's his name, we should use a capital E.

Thera are no magnets in the Hall.

The misconception is close though. Makes more sense than if his name were Late! Your comment is appreciated though, I was wondering why it was called Early.

I was amazed to find that Jim Early described this effect in BJT's in 1952! J.M. Early, "Effects of space-charge layer widening in junction transistors", Proc. IRE, vol. 40, pp. 1401-1406 (1952).

In the start commercial the 👍is pressed, because you know it is interesting!

The bottom line is that the transistors that determine the current source value (Q1 and Q2) do not see the voltage at the load. The "cascode" transistor Q3 isolates them from the load change, so there's no Early effect in the actual current source devices.

That's right!

Tek no longer makes handheld DMMs, but I do have a few of the older ones. I like the Fluke handhelds better.

w2aew I know Tek also just put their name on some Asian designed test equipment vs what they designed in house. But you can’t go wrong with Fluke DMMs. EEVblog intentionally abused one beyond belief and it just kept working. And they also retain their accuracy remarkably better than just about anything else even when they’re 20+ years old never having been recalibrated. They’re arguably in a class unto themselves.

Please allow me to refine your statement. Gifted, highly trained, eloquent people like AEW using fantastic quality vintage test gear deserve statues. I have a 576 too, but I certainly do not deserve any recognition at all!

It is!

We STILL use these circuits in the discrete world.

There are a lot of subtle things to be careful of when doing that. One is the output impedance of the source as described here. Another is that many capacitor types will change capacitance with bias voltage. Best to use highly stable film caps for this. Not ceramic or electrolytic.

​@@w2aew You are right I will try it, thanks a lot again.

LM334

@@robegatt I've used them. They're expensive compared to GP transistors, when you want to control dozens of LEDs, with each LED having it's own CC source.

@@piconano the good old dilemma: integrated or discrete with all the "salad" around them?

Q1 and Q2 should be as closely matched as possible, for the sake of current accuracy. Q3 can be anything. You might even use a larger transistor for Q3, because it will have higher voltage across it, and will tend to dissipate more heat than Q1 and Q2, which will never have anymore than 1 volt across them at the same current. Keeping Q1 and Q2 the same temperature will help, to keep the difference in current between the reference and the output from changing very much. The output transistor can be whatever temperature it wants to be, since it's effectively being driven in cascode.

@@vincentrobinette1507 Thanks! I guess Q3 still does need enough hfe to provide the current? Or in other words, if you want to mirror something like 100mA, you don't want to drive like 1mA or so? Darlington would probably work in that case.

@@vincentrobinette1507 Also I think this could be a (MOS)FET as well maybe?

@@p_mouse8676 You're exactly right, you will lose the beta of Q3. If you don't want to lose that, a Darlington pair will reduce that to the beta of the drive transistor multiplied by the beta of the power transistor. All that means, is that 1.4 volts appears across Q1. If you use a Sziklai pair, (PNP transistor driving an NPN transistor) you can keep the voltage across Q1 the normal ~.7 volts. It doesn't matter much, the important thing is to keep the voltage across Q1 as consistent as possible, to keep the output current as constant as possible. I would imagine, you could even use a MOSFET, which would allow ~3.5 volts across Q1, with no beta loss whatsoever. the problem with that is the gate threshold voltage varies with temperature, which would create current drift caused by varying the voltage across the reference resistor. (R1)

@@vincentrobinette1507 Thank you so much, makes a lot of sense now! :)

The Early effect is taking place in Q3 for sure, but is masked by the fact that the current available to it via Q2 is fixed. The feedback around Q1 causes the Q3 base to be adjusted accordingly.

Difficult to comment without seeing your circuit

Is beta/2 * output impédance of third transistor. Jean-Louis

Putting a current mirror as the load on a differential amplifier is a very common thing to do inside of op amp themselves. Gives a tremendous amount of voltage gain due to the extremely high output impedance of the current source.

@@w2aew Thank you...

They're called MePads, and are available from QRPme.com. www.qrpme.com/?p=product&id=MeSl

@@w2aew Thanks a lot! It's a great source for me. I did not know it. 73s.

The two transistor version wouldn't be as stable with temperature changes.

@@w2aew ah interesting, ill go play with that a little.

The explanation starts at about 6:10. Bottom line - the transistors that "set" the current do not experience any C-E voltage change with the load applied, thus no change in current due to the Early effect. The "output" transistor that is providing the current to the load is simply passing the constant current through it - it isn't part of *setting* the current value.

It is from www.remarkable.com

Current sources have a lot of uses: often used to bias differential amplifiers (emitter or source-coupled pairs), as well as an active load inside of operational amplifiers as well as other apps within ICs. Also used to bias diodes such as LEDs, and used to create linear voltage ramps when used in conjunction with capacitors, etc. Lots of uses.

It is not an oscilloscope, it is a Curve Tracer. en.wikipedia.org/wiki/Semiconductor_curve_tracer

Infinite output impedance simple means that the value of the current doesn't change when the voltage at it's output changes.

@@w2aew Thx! I guess i thought of 'impedance' not simply as a relationship between deltas. My mistake :)