Bi-directional constant current device

[betwixt]

Here is an interesting challenge, I'm looking or ideas.

I've been asked to design a very low cost lighting device that uses voltage reversal to select which set of LED lights are illuminated when they are wired in two parallel chains of alternately reversed parallel LEDs. The chains will be driven with a 2KHz square wave with a short dead time at transitions and each chain will have about 500 LEDs, 250 wired one way and the other 250 wired in opposite polarity across the two wires. The maximum voltage allowed is +4V or -4V +/- 5% on one wire relative to the other. The chains might be shorter or longer depending on the end users chosen configuration but the current required has to be fixed at 300mA (600mA for both chains) from source. So essentially it is a voltage limited constant current but reversing polarity generator.

My first thought is a single constant current source supplying a H-Bridge with the four switches driven by an MCU. Given the low voltage it can be done without bootstrap supply. I wonder if any readers have alternative ideas. Cost is paramount. A single 12V unregulated supply is available.

Brian.

 

[BigBoss]

An NMOS/PMOS pair charge pump might be a good starting point, though.
Non-Inverted and Inverted signals will commutate the transistors.

 

[crutschow]

My first thought is a single constant current source supplying a H-Bridge with the four switches driven by an MCU.

Can't think of a better way, as that's how I would do it.
A simple bridge such as below, such work.

 

[dick_freebird]

I think I'd go with a dual MOSFET driver and run the channels complementary, "poor boy H bridge" minus any phase control / BBM. Headroom and resistors for constant(-ish) current. Probably need back-diodes as many LEDs rate lower reverse than forward voltage and need protecting.

 

[Easy peasy]

H bridge is the standard way to approach this

 

[betwixt]

Can't do it that way Dick_freebird, there are only two wires and the LEDs self protect by being across each other so Vf is all they will ever have across them, Resistors to limit the current are not an option either as the number of LEDs can be different in each installation.

I'm thinking of a linear regulator, something like a 7805 wired for constant current then a H bridge with complimentary NPN & PNP so top and bottom of the arms can be driven B-E directly from a micro with 4 IO pins. If the bridge can run at the same voltage as the MCU it should be possible to drive the transistors by inverting the signal to the two top transistors, I'm not sure if that's feasible given the constant current requirement though as VCC would be variable. The prototype I was shown used direct drive from two MCU pins but obviously the current available was much less than needed in a production unit.

Brian.

 

[wayne3]

Can't think of a better way, as that's how I would do it.
A simple bridge such as below, such work.

View attachment 196434

Hello, since I do not have your contact information, I am sorry to ask you questions here, please forgive me if it may cause any disturbance.

I would like to inquire about the simulation of a ramp generator circuit with infinite hold time, I tried the same circuit but couldn't get the same result, can you tell me why? I would love to hear from you.

 

[dick_freebird]

I have seen LED data sheets where rated Vr is lower than expected Vf. Seems kooky but that's what they say. Back-back "freebie" protection might not suffice.

 

[danadakk]

LM7805 as Current Reg -

 

[crutschow]

I would like to inquire about the simulation of a ramp generator circuit with infinite hold time, I tried the same circuit but couldn't get the same result, can you tell me why?

I perhaps could if you post your LTspice .asc simulation file.

 

[wayne3]

I perhaps could if you post your LTspice .asc simulation file.

Thanks for your reply, here is my .asc simulation file.

 

[crutschow]

Thanks for your reply, here is my .asc simulation file.

You have a high frequency going into one circuit and a very low frequency in the other with the two outputs going to a differential amp.
I don't understand what you expect to see from that odd connection and large timing difference(?).

 

[wayne3]

You have a high frequency going into one circuit and a very low frequency in the other with the two outputs going to a differential amp.
I don't understand what you expect to see from that odd connection and large timing difference(?).

Sorry, I didn't describe my problem clearly and carefully check my attachments. This is a new add-on that only contains the ramp generator circuit with infinite hold time that you mentioned earlier, but in my simulation results, both the uphill and downhill ramps have some hysteresis, rather than aligning with the high and low levels of the square wave.

 

[wayne3]

You have a high frequency going into one circuit and a very low frequency in the other with the two outputs going to a differential amp.
I don't understand what you expect to see from that odd connection and large timing difference(?).

The image below is the ramp generator circuit with an infinite hold time you mentioned earlier.

 

[danadakk]

@wayne3, when you say infinite hold how long do you want the hold
to handle and what is droop spec ?

 

[wayne3]

@wayne3, when you say infinite hold how long do you want the hold
to handle and what is droop spec ?

In fact, the purpose of getting an infinite hold time is to get a defined range of capacitor charge and discharge voltages. I wanted the capacitor charge and discharge voltage to change linearly over time, which meant that a stable and consistent current needed to be supplied, but in the actual simulation attempt, assuming that a constant current source is generated by the current mirror circuit and the capacitor is charged and discharged, we found that the current fluctuates when the capacitor is added.

 

[wayne3]

@wayne3, when you say infinite hold how long do you want the hold
to handle and what is droop spec ?

As shown in the simulation image example, when no capacitor is attached, the circuit is unloaded and can generate a steady current.
However, when a capacitor is added, a stable current cannot be generated.

 

[danadakk]

Certainly some asymmetries between current sink versus source :

Larger cap :

 

[crutschow]

in my simulation results, both the uphill and downhill ramps have some hysteresis, rather than aligning with the high and low levels of the square wave.

Your file still shows the same odd combination of the two ramp circuits, which I still don't understand.

--- Updated Jan 28, 2025 ---

In fact, the purpose of getting an infinite hold time is to get a defined range of capacitor charge and discharge voltages. I wanted the capacitor charge and discharge voltage to change linearly over time,

Okay, here's the circuit modified to provide a linear ramp and hold, up or down, using two separate control inputs to stop and hold the capacitor voltage at any desired value:
The V2 /Chg input goes low to charge the capacitor, and the V3 Dis input goes high to discharge the capacitor.
If V2 is high and V3 is low than the capacitor voltage (yellow trace) is held with no charge current in or out (45s to 50s, and 70s to 80s in the sim).

Is that what you want?

 

[danadakk]

Alternatively the digital approach. Timing and V's accurate -

Test Bed DDS and WaveDAC burst tool

Occasionally I need to generate custom wave forms and burst a known number of cycles into a DUT. This is a single chip solution, board I typically I use is $ 10. You just drag and drop components onto schematic, wire internally with a wire wizard tool and out to pins, right click and config...

www.edaboard.com

Arbitrary waveform, single chip, control of # cycles burst, V's, inter burst timing, flrexible
trigger capability, real infinite hold......