EEVblog 1688 - Constant Current Sources EXPLAINED + DEMO

[EEVblog]

00:00 - Constant Current Sources
11:08 - Practical uses of constant current sources
14:00 - Circuit examples
19:13 - TL431 Example
20:25 - LM317 CC circuit
21:42 - Low Side Source vs High Side Current Sink
24:19 - Bench examples
26:24 - Your PSU is a CC generator!
28:04 - Keithley 225 Constant Current Source
31:09 - LED strip example of Compliance Voltage

[vcvs]

How would the top circuit (high-side current source) be realized in practice? In particular I’m thinking about how to make the three terminal device which tracks the reference voltage difference. A differential amplifier followed by an opamp? Suggestions for resources about this would be appreciated 

[thm_w]

How would the top circuit (high-side current source) be realized in practice? In particular I’m thinking about how to make the three terminal device which tracks the reference voltage difference. A differential amplifier followed by an opamp? Suggestions for resources about this would be appreciated 

You can use premade part like LT3080: https://electronics.stackexchange.com/questions/473288/how-to-choose-the-series-pass-transistor

Maybe relevant:
https://electronics.stackexchange.com/questions/381525/how-to-calculate-this-current-limit-circuit https://electronics.stackexchange.com/questions/692166/how-to-understand-current-limiter-circuit https://ecstudiosystems.com/discover/textbooks/selected-circuits/ac-to-dc-power-supplies/current-limiting-circuit/

Not sure what you are asking though, how to make a three terminal linear regulator?

[EEVblog]

How would the top circuit (high-side current source) be realized in practice? In particular I’m thinking about how to make the three terminal device which tracks the reference voltage difference. A differential amplifier followed by an opamp? Suggestions for resources about this would be appreciated 

Literally any PSU constant current limiting circuit implementation.
As thm_w said, the LT3080 is literally this, and was used in my early uSupply design video series.
Then the latest uSupply design did it more discretely, and put the voltage regular after the current regulator.



And there is the P DC PSU handbook:
https://www.changpuak.ch/electronics/PowerSupply/HP_AN90B.pdf

But there are probably a hundred ways to implement current limiting, which is why I just explained it generically as basically measuring the voltage across the shunt, compare to reference, drive pass transistor with opamp.

[JoeyG]

Voltage to Current Source
Varying the current on Dave's current sources often require a variable resistor to  adjust the current.

I'm making a VI curve tracer and I wanted to vary the current from a mcu dac.  IE Voltage to current source.
I found some app notes from ADI however in the end  I found an article in Wireless World that uses some transistors
to make a voltage to current source     (Wireless world magazine articles here https://www.worldradiohistory.com/Wireless_World_Magazine.htm

You can see my LTspice modeling of the Wireless World article here.
The  Q4_c (the PNP collector)  drives the load (R2_Load)

[vcvs]

Cool, thanks for all the links!

My confusion was about that the amplifier in the middle of the generic figure both measures the voltage drop and tracks the reference voltage I was thinking about how to implement that part discretely with op amps.

[jh15]

Wish I could remember my learning in transistor circuit approximations. The teacher would draw a simple xistor current source, then apply it to circuit examples. One could pick any part of time in a dynamic waveform, and go from there. Just as i got the hang of it, digital circuits dominated.
     I recently got an old book on transistor circuit approximations, and it has become Greek to me. Wish i had my notes (was done on something called paper back then).
     For EE stuff, that would bring back the original attraction of EEVblog. I have had no problem with the recent years, but seeing EE stuff when you started was exciting..
     

[JoeyG]

Just learn LTspice and create some simulations

[Kleinstein]

The linked video from Wireless world has another related circuit element: a current mirror. That is a current controlled current source. It is a common element inside ICs, but can be a bit tricky with discrete transistors where the matching and thermal coupling is not that good.

[thm_w]

My confusion was about that the amplifier in the middle of the generic figure both measures the voltage drop and tracks the reference voltage I was thinking about how to implement that part discretely with op amps.

You can have the current read by a first opamp fed into a second opamp, which compares against the reference voltage.
https://www.circuitlab.com/circuit/6vnmv5/opamp-current-limiter/

[tppc]

Hi Dave,

Thanks for this video, very instructive for young players.
Correct me if I am wrong, but I think you should reverse the OP amp inputs on your High side regulated current source model.

BTW : No calculators that change model under the white board this time 
However markers change their color and position. 

[EEVblog]

Correct me if I am wrong, but I think you should reverse the OP amp inputs on your High side regulated current source model.

That was deliberate. As I mentioned, this is not an exact practical circuit example, and I explained how I was effectively showing a differential amplifier across the shunt resistor, the left side of which would be positive.

[tppc]

OK, sorry for this.
Thanks for your patience. 

[rvalente]

I've got curious on this cute CC Dummy load, could not find any reference on google.

[temperance]

I found some app notes from ADI however in the end  I found an article in Wireless World that uses some transistors

Depending on the required accuracy you need thermal coupling between Q3 and Q4. This will be difficult to achieve because they don't dissipate the same power.

There are better options if you need better accuracy:
https://www.ti.com/lit/an/slyt768/slyt768.pdf?ts=1749420278451&ref_url=https%253A%252F%252Fwww.ti.com%252Fsolution%252Fanalog-output-module

Page 5 for example.

[thm_w]

I've got curious on this cute CC Dummy load, could not find any reference on google.

It was from mailbag: https://www.youtube.com/watch?v=Xfs0dglIVOM

[jirij]

One thing I don't think was mentioned in the video (and could go into one of the "common beginner mistakes" compilations) is that many (most?) CC PSU implementations do the limiting before any output capacitors. This is also the standard for any current-measuring PMICs that typically have a few small MLCCs near the chip (presumably for fast transient response on the feedback pin), then a current sensing resistor, and only then the big electrolytics.

In practice, if you want to test a ~3V LED at 20mA, you short the terminals, dial in 20mA, open the circuit, then ramp up the voltage to "good enough" 5V because CC will save it, and then you connect the LED, you get a big spark and a dead LED.

...
OTOH you can drive LEDs from voltage-controlled PSUs because, as you increase the DC voltage, any parasitic effects (like ripple) become less relevant. I've built a custom 17-LED lamp based on careful V/I measurements of the specific LEDs I wanted to use (~2.74V @ 110mA) and am running it on a CV 48V Meanwell PSU with the trimmer pot adjusted down to a voltage low enough where the LED heatsink stays below ~50C.
Yes, a LED Vf drops with higher temperatures (and they get brighter / draw more current as they approach 50C), but it's still surprising how little effect +- 1V has. Even if the PSU was wildly inaccurate, the LEDs would not burn up.

Is it advisable? No. But I can have a switch near the lamp head and the lamp doesn't self-destruct when I turn it on.

(Yes, the best solution is a CC PSU with voltage set to just a few volts above, and the LEDs don't mind an inrush current that's 2-3x higher than the operating one. Most "LED driver" PSUs have TL431 or zeners clamping the output voltage and changing them is easy, as long as the PSU is not in a brick of silicone. Oh well, in-line resistors to the rescue.)

[Kleinstein]

If a LEDs likes voltage drive depends on the LED. Blue and white LEDs are usually OK as they have quite some internal series resistance. Especially older red LEDs tend to show thermal run-away and will likely get too much current.

The note about the output capacitor is good. Many of the lab supplies have a surprisingly large capacitor (like 1000 µF range) at the output. In addition (often the lesser problem than the capacitor) that can be a delay in the cross over from the CV to CC mode in the controls. This kind of acts like a simulated extra capacitor.

[jh15]

The output capacitor in the HP supplies i use, you could disconnect the output capacitor that was effectively on the output terminals. Set the voltage and current limit, maybe shorting the output terminals, and think you would be limiting the overall current at start. If the capacitor was involved, you could get a large current spike into the load.

[Kleinstein]

Many supplies need at least some capacitance at the output to be stable (not oscillate) at all. More capacitance may be needed to reduce the overshoot in the CC to CV transition or with load steps.

[jh15]

That is why the default configuration is that way. The option for the engineer is made available easily.

[notadave]

After watching EEVblog #1688 I have tried to systematize how to built a current source:

True regulation
  Two Op-Amps + Voltage regulator
  Series current mirror (Darlington BJT) + OpAmp + U-Ref
Some regulation that requires initial trimming
  JFET with bias resistor aka constant current diode
  NPN+PNP+Trimmer resistor+Voltage reference (assumes that the two BJT can be trimmed such that they act as a voltage mirror)

Two types of current mirrors may be used
  Series with darlington transistor
  Parallel with parallel monolithic BJT (Only good to get more currents from one, not exact.)

The only true current source I know is a voltage gap reference and a dual CMOS-OpAmp circuit that regulates the voltage over a resistor that shares it's current with the load. This method is very precise for a wide range of currents. It may contain additional transistors to boost power.
The simplest method I know is the JFET with bias resistor aka constant current diode.
There are many ways in between. One of the most clever ways I first saw here in this video: The use of a darlington transistor as a current mirror at emitter and collector. With the voltage regulated resistor at the emitter.
To use a linear voltage regulator is conceptually identical to the use of a U-reference + OpAmp + Transistor, it just conveniently comes in one package. It is the only method I have ever needed but it has the potential weakness that the current at the Adj-terminal might be large in comparison to the needed current.
Should a current against the positive rail be regulated with a positive 3-terminal voltage reference (because that is available) there is a way.
Should more identical currents be needed the current mirror as used in bipolar ICs may be used with matched transistors, but I would not trust it to be exact.

More video blog entries on the topic are:
#567
#577
#579
#1397