13
\$\begingroup\$

Please don't state the obvious that nothing is a perfect constant current source. And please do not give examples of how to create them from components.

I looked at this SE Q/A which stated the obvious - that you can not have a perfect current source.

So when I say constant current source, I am not referring to the mathematical concept used in circuits, but "natural" components that provide a close approximation to this. Key word is approximation.

A bit of research pulled up one interesting example which was a solar cell. Is this a valid example?

A google search pulls up a similar Quora question here. Which gave the example of a solar cell.

This is a more technical site and I was looking for a more engineering answer not a mathematical or laymen's answer.

Thanks.

\$\endgroup\$
5
  • \$\begingroup\$ For what anount of time do you define constant? The solar energy is not constant in time on the same spot on the earth... \$\endgroup\$
    – Huisman
    Apr 20, 2020 at 16:29
  • 2
    \$\begingroup\$ While they are not a constant current source, bicycle dynamos are an interesting third case: constant power source. They provide approximately constant power at different load currents, see e.g. this graph. \$\endgroup\$
    – jpa
    Apr 21, 2020 at 4:57
  • \$\begingroup\$ @jpa they can also approximate a constant current source, at least over a range of loads similar to their rated output. It's not clear whether that graph is taken from the AC (but not sinusoidal) output or after rectification (with diode IV characteristics). They're interesting (and not in a good way); I'm planning some tinkering if real work dries up while I'm locked out of the lab \$\endgroup\$
    – Chris H
    Apr 21, 2020 at 10:01
  • \$\begingroup\$ @jpa Errr... fluctuating between 3 and 5.5W is not constant, even with a whole salt shaker. \$\endgroup\$ Apr 22, 2020 at 15:27
  • 1
    \$\begingroup\$ @Peter-ReinstateMonica If you are particularly strict, no unregulated source is constant. Batteries are often approximated as constant voltage sources, even though their voltage will drop quite a bit with load. Over the optimal range in that chart, the variation is just from 5W to 5.5W - at the ends of the chart you can consider the dynamo as operating outside its designed range. \$\endgroup\$
    – jpa
    Apr 22, 2020 at 15:47

14 Answers 14

20
\$\begingroup\$

Photovoltaic cells and betavoltaic cells are examples of devices that produce more-or-less constant current up to some maximum open-circuit voltage, just as a battery produces more-or-less constant voltage up to some maximum short-circuit current.

\$\endgroup\$
4
  • 4
    \$\begingroup\$ thanks so much for not blabbing on about infinite when I am asking a practical question. \$\endgroup\$
    – jennifer
    Apr 20, 2020 at 17:45
  • \$\begingroup\$ To be exact, their constant-current behaviour starts to decline quite soon as the voltage reaches a mere fraction of the open-circuit voltage (as a diode current follows a nearly logarithmic function of voltage). Practically, this is some 0.5 V for a silicon cell. \$\endgroup\$
    – dominecf
    Apr 21, 2020 at 22:42
  • \$\begingroup\$ This confuses me. When I read the question I understood it as "which power source makes the same current run through a resistor of n Ohm and, say, 2n Ohm and 1/2n Ohm." Never is voltage mentioned. (That the voltage is higher with the larger resistor is implicit, but it is not the "input parameter".) \$\endgroup\$ Apr 22, 2020 at 15:31
  • \$\begingroup\$ @Peter-ReinstateMonica Isn't that pretty much the complement of a battery where the voltage stays relatively constant regardless of the load? The current may vary over 7 or 8 orders of magnitude with little change in voltage. \$\endgroup\$ Apr 22, 2020 at 15:34
7
\$\begingroup\$

I would say that a solar cell, or more precisely a series of solar cells are a good example of a current source, because the more of them you add, the higher the voltage you can reach while the current remains the same, which is approaching the second characteristic of a current source:

  • the first characteristic being a constant current whether the current source is shorted or whether it is being impeded by some load resistance,
  • the second characteristic being a very high source impedance coupled with a very high source voltage which is needed in order to keep pushing the same current through whatever load resistance is trying to impede its current.
\$\endgroup\$
6
\$\begingroup\$

Batteries can be an interesting example of both a current and a voltage source. A single cell behaves as a voltage source with an internal series resistance, which can also be thought of as a current source with an internal parallel resistance.

If you wire many cells in parallel, you get closer to an ideal voltage source, but if you wire many cells in series, you get close to an ideal current source, which will deliver something close to short circuit current of a single cell over an arbitrarily wide voltage range.

Of course, this works with many other things, including solar panels. You can build a better approximation of a current source by wiring many bad approximations of current sources in series. If that's not an engineering answer, I don't know what is ;-)

\$\endgroup\$
3
\$\begingroup\$

Photodiodes are treated as one under the right conditions.

Voltage sources with very high series output impedances (I think piezos might be one) can also be treated as current sources if the load impedance is relatively low (i.e. the load impedance can vary but as long as it remains low relative to the source impedance). The reason is that the extremely high output impedance dominates the circuit and therefore determines the current. The relatively low load impedance affects the current little in this case so long as it remains low relative to the output impedance.

\$\endgroup\$
3
\$\begingroup\$

Various manufacturers (including IXYS, Microsemi, On Semi) make two terminal current regulator ICs. Some examples would be...

NSI50010YT1G 50V, 10mA, SOD-123 package
https://www.onsemi.com/pub/Collateral/NSI50010Y-D.PDF

IXCY10M90S, 900V, 100mA, TO-252-3 package
https://ixapps.ixys.com/DataSheet/DS98729A(IXCP-CY10M90S).pdf

Another example would be current regulator diodes. Some examples include...

Semitec E-501 , 100V, 0.5mA, through hole
https://www.mouser.com/datasheet/2/362/P22-23-CRD-1729293.pdf

https://www.mouser.com/Semiconductors/Discrete-Semiconductors/Diodes-Rectifiers/Current-Regulator-Diodes/_/N-ax1ml

A J-FET with its gate and source shorted forms a current regulator. One example would be...

InterFet J556-7, 50V, 3mA
https://www.mouser.com/datasheet/2/676/jfet-j556-j557-interfet.r00-1649142.pdf

Similar to a J-FET, a depletion mode MOSFET with its source and gate pins shorted becomes a constant current source. One example would be

Infineon BSS139. The current would be roughly Vgs/ Rds = 112mA nominal
https://www.infineon.com/dgdl/Infineon-BSS139-DS-v01_08-en.pdf?fileId=db3a304333b8a7ca0133eed3136c61f2

Any reverse biased P-N junction that is exposed to light (including photo-diodes, as well as regular through hole rectifier diodes in a clear glass packages) will pass a constant current that is proportional to the amount of light.

\$\endgroup\$
2
  • \$\begingroup\$ Your example of the gate-source-shorted FET is pretty much how most of those two terminal regulator ICs work, I imagine. Possibly with a resistor in it to tweak the current. \$\endgroup\$
    – Hearth
    Apr 22, 2020 at 2:03
  • \$\begingroup\$ @Hearth I suspect as much, but they don't always give the details in the datasheets so I listed them as a separate category. \$\endgroup\$
    – user4574
    Apr 22, 2020 at 14:42
3
\$\begingroup\$

One view of an ideal current source is that it "enforces" a current, and there is no limit to the voltage that the current source will apply to make that current flow. There actually is a device that approximates this behavior very well, even if only for very short times: the humble inductor.

As a reminder, the formula linking voltage and current of an inductor is V = L * dI/dt, or using words: the change in inductor current is proportional to the timespan and the voltage across the inductor and inversely proportional to the inductance. If our inductor is "very big" and the timespan we look at is "very small", then the current is practically constant, even for large voltages.

Contrary to other current sources, the magnetic field inside an inductor will make sure that the current does not change, and it is able to generate extreme voltages to ensure that current flowing. For example, if you "charge" an inductor with a low voltage (e.g. 12V) and suddenly break the circuit, the voltage across the inductor can easily increase by several orders of magnitude (e.g. 5kV) - just to force that current flowing. In reality, you always have some parasitic capacitance which is able to absorb this current - or you have a spark, which is what happens when the inductor is so adamant about making sure its current flows that it forces the air to become a conductor.

This principle is used in every step-up converter: the higher output voltage is caused by an inductor whose current path (to ground) was interrupted; it then forces the current to flow into the output capacitor, even if that capacitor is at a higher voltage than the input.

To summarize: for very short periods of time, an inductor behaves very much like an ideal current source.

\$\endgroup\$
1
\$\begingroup\$

A practical example of a current source may have an ideal current source as an essential part of its equivalent circuit with some other components or characteristics built into it.

The equivalent circuit of a solar cell has a current source that is based on an ideal current source with a resistor and diode in parallel and a resistor in series with the output.

A 4-20 mA control device provides an output current that is proportional to its input. For a given input the output current is constant over a range of load resistance. The load resistance range for constant current is usually something like 100 to 1200 ohms.

There are motor controllers that provide an output current that is proportional to the motor torque reference. The motor torque reference and come from a closed loop speed controller that has a speed error signal that is used as a torque reference. In that case, a part of the overall system is essentially a current source.

\$\endgroup\$
1
  • \$\begingroup\$ To add a point to your answer: We use 4-20mA for field devices because the current is always the same at every point in the circuit (KCL). If we had, say, a pressure transmitter on a pipeline wired to a PLC in a cabinet 100 yards away, we don't have to do anything to account for the wire resistance. That makes for easier design, installation, and maintenance. \$\endgroup\$ Apr 21, 2020 at 20:45
1
\$\begingroup\$

Need more examples?

Car alternator (without the usual voltage regulator attached)

Welding equipment for manual or TIG welding

CR2032 battery (e.g. when used to power a single led)

A bipolar transistor away from its saturation region

A field transistor IN its saturation region or a termionic valve in the equivalent region

A power LED driver

A HID driver

A photomultiplier

An electrochemical cell in diffusion mode of operation (well, the CR2032 example above is a particular case here)

None of these is an "ideal" current source, but in the general case is used, designed, regarded and engineered as a current source.

\$\endgroup\$
1
\$\begingroup\$

Most of the answers here list circuits that have some (nearly) horizontal portion of their I-V curve. The current-source approximation then holds, but it is usually easy to get outside of it when the load impedance changes a bit.

You can, however, get very close to an ideal current source with a Van de Graaf generator, provided a separate HV supply charges the belt at the bottom (i.e. the belt is not charged by mere electrical induction, which would make output current proportional to the output voltage).

Then the amount of charge transported by one belt revolution is virtually independent of the output voltage (no matter it is −1 MV, shorted or +1 MV), up to the point a long multi-megavolt discharge in air shorts this current source.

\$\endgroup\$
1
\$\begingroup\$

There's an obscure device, the constant-current transformer, which provides constant current AC.

https://www.globalspec.com/reference/59613/203279/4-11-The-Constant-Current-Transformer

There are a few variations on that theme. Some have a moving slug, some are saturable reactors. Historically, they were used for series street lighting, where all the lights are in series. Each one has, instead of a fuse, a paper disc with contacts on both sides. If the bulb burns out, the high voltage that drives the entire loop is across the disc, the disc burns through, the contacts come together, and the lamp is bypassed. It's sort of a mirror world of parallel distribution.

This trick is still used today for airport runway lighting. You have hundreds of lamps spaced out over a mile or two of wire. Providing them with constant voltage power would require distribution transformers at multiple locations. With constant current, you can power the whole loop from one point. High voltage, though, a kilovolt or more.

\$\endgroup\$
1
\$\begingroup\$

The parcical way to make a constant current source in a lab is to have a high-voltage battery connected to a large-value resistor, say 1000V with a 1M Ohm resistor. As long as the resistance of the circuit is much less than the large value resistor the current resulting current of one mA will be unaffected by the what happens in the circuit.

\$\endgroup\$
1
\$\begingroup\$

We often use a 2-transistor constant-current source to control LED brightness. All our products are battery powered - If we just used a simple resistor in series with the LED, then it's brightness would vary depending on battery state. LED's are current-controlled devices. Their output is proportional and mostly linear to the amount of current passing through.

\$\endgroup\$
0
\$\begingroup\$

If the illumination is constant, each pixel of the TESS cameras is an almost perfect current source, with current proportional to optical intensity. This is what makes it possible to detect the slight dimming of a star when an exoplanet passes between it and us.

\$\endgroup\$
2
  • \$\begingroup\$ So that would be true for any CCD camera then, no? \$\endgroup\$ Apr 22, 2020 at 12:51
  • \$\begingroup\$ @leftaroundabout Yes. True of any photodiode within the limit set by the height of the potential barrier that prevents photocurrent from flowing the "wrong way" within the device. \$\endgroup\$
    – John Doty
    Apr 22, 2020 at 13:41
0
\$\begingroup\$

A permanently excited AC generator is a nice example of a current source. A typical application is the bicycle dynamo. While it is labelled "6V 3W", the best label for it would be "500mA current source".

The frequency of the AC voltage generated by the generator is obviously proportional to the rotational speed. As the induced voltage in the coils is proportional to rotational speed as well, the open-circuit voltage of a bicycle generator is proportional to your driving speed. The main idea of the bicycle generator as current source is that the winding behaves as inductor, so it has an impedance proportional to frequency. As soon as the frequency is high enough, the impedance of the winding greatly exceeds the 12 Ohm load of the light bulbs, so the bulbs can be simplified to a short circuit. What remains is a voltage source that is proportional to speed in series with an impedance that is also proportional to speed, so the current (which is voltage divided by impedance) is constant.

The answer of fraxinus already hints (amongst many other things) at a "car alternator without regulator", which actually works the same way. The regulator in modern three-phase alternators actually controls the excitation (i.e. the magnetic field) so that the current source matches the power consumption of the car, although the regulation usually is indirect by controlling the output voltage. The battery is the element that provides a stable voltage. If excessive current is generated, it charges the battery and the system voltage is higher than the "pure battery voltage" due to the internal resistance of the battery. If the car consumes more current than the alternator supplies, the system voltage is supported by the battery, but lower than the "pure battery voltage" again due to the internal resistance of the battery. So controlling the system voltage indirectly ensures current equilibrium at a specific target battery voltage.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.