We have retrofitted dimmable LED panels in our office spaces. These use a 0-10 Vdc control voltage. The control leads have an internal pull-up to about 10 Volts so that the panel runs at full intensity if there are no connections to the control leads.

The open-circuit voltage is about 10 Vdc and the short circuit current is about 1.3 mA. We connect the control leads from all the panels in a given space together. Thus - the current sums. The most number of panels in any given space is 8 - most are two or three or four panels.

I have been trying to come up with a 2-wire dimmer circuit but without success. My best result is using a LM10 op-amp but the (about) 0.27mA quiescent current limits the maximum voltage and does not allow the panel (s) to reach full brightness.

I do not wish to pull external power supply lines to the dimmer location in each space.

It occurred to me that perhaps I don't need continuously variable brightness. So: an eight position rotary switch with old-fashioned RED LLEDs in series might be the way to go. These LEDs have an about 1.7 Vdc drop which is fairly constant as the current is varied. Low-voltage zener diodes just do not work well - the knee is extremely soft.

I am hoping that some of the extremely bright people who inhabit this space might have alternative suggestions.


  1. This is an adjustable two terminal shunt regulator.

  2. Desired voltage range is zero to 10 Vdc. Minimum intensity is with a short across the control leads. Maximum brightness is with the leads open (about 10 Vdc across the leads).

  3. From comment: "How would you power the op-amp". The LM10 can operate as a shunt device. That is: it can take the current it needs from the load two terminals. Downsides are the minimum voltage (1.1 Vdc) and the quiescent current (270 uA). This is shown in the data sheet.

We have used many thousands of LM10 devices as two-terminal thermocouple transmitters with great success over the past couple of decades. But I have much more current (and voltage) to work with in that system.

After doing a bunch more testing, it appears that a PWM solution will not work/ There is a nominal 3.3uF capacitor across the control leads. This completely swamps out any attempt at PWM. The total capacitance sums with the number of panels in parallel.

The Vbe multiplier suggested by Simon Fitch works well, except at the low end. There is a very noticeable brightness change between the lowest brightness setting on the pot versus a short circuit. I will ask the users of the space if the lowest brightness setting is acceptable or if they require it to go down to the lowest possible value with a short across the control leads.

[question]: I see a down-vote on my question. Can I ask why? What edit (s) can I make to eliminate the problem? Guidance appreciated!

  • \$\begingroup\$ What do you mean by "2 terminal" shunt regulator. Most regulators have at least 3 terminals, Vin, Vout and "ground" or "adjust" or something similar. For example the TL431 has three terminals. A zener diode by itself is not really a shunt regulator. You really need a resistor and a zener (i.e. 3 terminals) to make a zener shunt regulator. So what do you mean by 2 terminal? \$\endgroup\$ Feb 28 at 1:05
  • \$\begingroup\$ What is the minimum controlled voltage you need? i.e what is the voltage required for the lowest brightness level? \$\endgroup\$ Feb 28 at 1:56
  • \$\begingroup\$ "I do not wish to pull external power supply lines to the dimmer location in each space." Then how would you power the op-amp? \$\endgroup\$
    – RussellH
    Feb 28 at 2:24
  • \$\begingroup\$ You can tie all the control leads together. You only need to draw power from one location. Your phantom powered approach isn't going to work without considerable compromises to performance - a highly abbreviated dimming range, for example. \$\endgroup\$
    – MOSFET
    Feb 28 at 4:03
  • \$\begingroup\$ Why can't you simply use a potentiometer? Say 100k with audio taper? \$\endgroup\$
    – MOSFET
    Feb 28 at 4:30

5 Answers 5


Vbe Multiplier

I was going to suggest a Vbe multiplier like Simon, because that's the simplest adjustable shunt regulator, but it has an issue: it can't bring voltage to zero, so you can't turn off the LEDs. You'd need a pot with a switch on it, like on the old pocket radios. But you'd need the switch to open when turning the pot right, and they usually close. So maybe a separate switch in parallel with the pot, so you have brightness and on/off controls separate, which also offers the feature of remembering the brightness level from yesterday when you turn the thing on!

You could use a TLV431 (1.2V, 80µA) but that also limits minimum voltage to 1.2V.

From a LED driver datasheet

enter image description here

Rotary Switch

Your rotary switch idea fixes this problem, as the lowest setting can short the wires and really output 0V. You could also use a switch with more positions and silicon diodes instead of LEDs.

If you use LEDs in series as your voltage references, you can make them visible to give visual feedback, could be useful if the knob is not in the same room as the lights. Also, it would look damn cool.


Most 0-10V LED drivers also allow PWM dimming. The signal is filtered internally so you won't get flickering, but in this case you need a PWM generator that is powered by the output signal. To get power when the LEDs are off, some cheating is required,. Any PWM value below a few %, say 1%, means "off" so we can get away with 1-10µA of "free" supply current.

enter image description here

So we have good old CD4093 schmitt trigger NAND gate (1µA), it oscillates around 1kHz with a 10Meg potentiometer, so with a 220pF capacitor that's i=C.V.f=1µA, a few more µA to drive the MOSFET (pick the lowest possible gate charge).

It should be possible to get rid of the MOSFET, and connect the CD4093 output directly... but the output resistance of these gates is quite wimpy, so it probably won't go down to 0V.

At 2µA this circuit can also be powered by a CR2032 coin cell for about 10 years...

  • \$\begingroup\$ I like this a LOT! I will breadboard it today and see how it goes. Many thanks! \$\endgroup\$ Feb 28 at 18:29
  • \$\begingroup\$ You're welcome! You'll need a MOSFET with a really tiny gate charge because every time it switches that uses some current and you don't have much... Something like 2N7000 or any low voltage FET with really low input capacitance... \$\endgroup\$
    – bobflux
    Feb 28 at 18:35

In the blue box is a "rubber diode", which develops a voltage across it \$V_{AB}\$ roughly proportional to R1. To a large extent, behaviour is consistent regardless of how many dimmer inputs you tie together (in other words, regardless of total source current).


simulate this circuit – Schematic created using CircuitLab

R3, R4 and R5 represent multiple current source dimmer inputs, all joined. The only part you build is in the blue box.

Using the components shown, when I sweep R1 from 0Ω to 500kΩ, \$V_{OUT} = V_{AB}\$ varies between +0.7V and +9.9V:

enter image description here

It might also be a good idea to make R2 adjustable, to obtain maximum range for the particular number of dimmers you have tied together.

  • \$\begingroup\$ I notice your circuit has a Vin (10 V) a Vout and a ground, i.e. 3 terminals, as I would expect for a regulator. The OP requested a "two terminal shunt regulator" and I do not know what that means, but perhaps it means something other than what I imagine it to means. Do you understand the term "two terminal shunt regulator"? \$\endgroup\$ Feb 28 at 5:53
  • 1
    \$\begingroup\$ @MathKeepsMeBusy All OP builds is in the blue box. It has two just terminals. Everything else is the dimmers themselves. \$\endgroup\$ Feb 28 at 5:55
  • \$\begingroup\$ @MathKeepsMeBusy I think I know what "two-terminal shunt regulator" means. Do you think that my two-terminal shunt regulator in the blue box is not a two-terminal shunt regulator? \$\endgroup\$ Feb 28 at 6:01
  • \$\begingroup\$ Well, I wouldn't exactly call a Zener diode or rubber diode a "regulator" in itself, because they don't limit current in themselves (other than overheating). Your circuit has 3 resistors which limit current, and I guess these correspond to the "dimmers". This is just language that I am unfamiliar with. \$\endgroup\$ Feb 28 at 6:05
  • 2
    \$\begingroup\$ @MathKeepsMeBusy I see. This rubber diode is a voltage regulator. I'm not trying to regulate current at all, since that will be decided by the dimmers, and will vary depending on the number of dimmers connected. \$\endgroup\$ Feb 28 at 6:06

The open-circuit voltage is about 10 Vdc and the short circuit current is about 1.3 mA. We connect the control leads from all the panels in a given space together. Thus - the current sums. The most number of panels in any given space is 8 - most are two or three or four panels.

These panels were designed that way for a reason. All you need to adjust brightness is a potentiometer. That's why they went to all the trouble. You'll also note that the current vs. voltage curve is designed so that it takes a linear taper potentiometer and lets it control the brightness in the usual pseudo-logarithmic fashion.

In other words: all you need to do is to figure out the value for a potentiometer, and if you'll be using an 8-position switch, you'll just make a basic stepped rheostat. Easy-peasy lemon squeezy :)

These use a 0-10 Vdc control voltage.

How do you think they meant for people to control them? The only sensible option is a rheostat. There is nothing else you can do that won't require an adjustable shunt voltage reference that will have to go to zero volts. Those are hard to come by as you've noted.

As soon as you stop fighting their approach, it'll suddenly turn easy :)

  • \$\begingroup\$ The manufacturer sells a packaged module that requires 12 Vdc supply. As noted above: each panel sources about 1.3mA into a short circuit. It does not appear to be a constant current source; rather, just a pullup resistor to a nominal 10Vdc internal power supply. \$\endgroup\$ Feb 29 at 3:44
  • \$\begingroup\$ Using a potentiometer (rheostat) is easy so long at you match the resistance to the number of panels connected in parallel. I don't want to have to do that - it is not a universal solution. \$\endgroup\$ Feb 29 at 3:45

The following schematic does what I need it to do. I am using a Lorlin 1P 12T shorting switch from Mouser. This gives 12 discrete brightness levels from minimum (short on the control leads) to maximum (control leads are open).

I thank Simon Fitch for his suggestion of using a compound-connected transistor pair.

I may modify this answer with a table of measured LUX values and DC voltages vs resistor values.

Sorry the schematic is so large - I have not mastered the art of cropping PNG images.

I hope this helps someone else facing a similar problem.

enter image description here


These dimmers certainly exist.

While product recommendations are off-topic, I've used the Thorn LC-POT before and can confirm it only has two terminals, and no need to adjust based on number of lights.

It shunts to a very low voltage (perhaps 0.5V?) at minimum brightness - we ended up putting a diode in series with each dimmer to lift the minimum voltage so that it couldn't turn the lights completely off.

Searching for a 1-10V dimmer reveals a variety of other similar dimmers that appear to be two-terminal and have no adjustment method.

It may be worth acquiring one (I don't think I have one here) and reverse engineering it.

  • \$\begingroup\$ These panels reach minimum brightness when the control leads are shorted. The voltage drop caused by Simon's Vbe multiplier (about 0.7 Vdc) causes a noticeable change in brightness between that voltage and a short. A shunt regulator with a minimum voltage of 1 Vdc is easy. One that goes very close to 0 Vdc is significantly more difficult. \$\endgroup\$ Mar 7 at 18:22
  • \$\begingroup\$ @DwayneReid Some drivers are designed to turn fully off when provided with 0V, and they typically turn off below about 0.7-1V. Others cannot be turned off using the dimming wires, and on those I would expect that the brightness could be linear all the way from 0V. \$\endgroup\$ Mar 8 at 22:30
  • \$\begingroup\$ @DwayneReid The drivers I was using on that project did turn off if the dimming leads were shorted, and my client did not want that. \$\endgroup\$ Mar 8 at 22:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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