Hello, I came across crutschow's design in this thread http://forum.allaboutcircuits.com/showthread.php?t=86252 and am wondering if it could be adapted for my problem. It seems most single-colour holiday light strings are wired with every other LED reversed, allowing various special effects but not what I want, which is just continuously on with adjustable brightness. The controller evidently uses PWM and reverses alternate pulses on its full-on mode. The supply is 24v DC.
I'm wondering if something similar, but fast enough to avoid it looking like a strobe, could do the job.
Any ideas from the experts please?

 

A 555 sources and sinks current, which allows you to conduct current per your need. Two 555's in this case to get independent PWM. The 555 would need to be buffered with N gated silcon for one LED string, and P gated silicon for the other string (a kinda flip-flop ckt with two 555's). Then just up the frequency of the PWM on 555 to something like 200Hz and adjust % PWM on both 555's to your dimming needs. Duly noted, 80% PWM on the N silicon switch is 20% on the P silicon switch. Why is this? Because PWM is based on 100%, on+off time per cycle is always the same, tied to frequency, etc.

You following the idea?

 

Here's my take on a circuit to do what I believe you want (LTspice sim below):
It uses two CD4093 Schmitt-trigger NAND gate packages, and one CD4013 flip-flop to drive the H-Bridge.

The U1 Schmitt-trigger NAND gate is configured as an astable to clock the U3 flip-flop (yellow trace) and reverse the direction of the PWM signal voltage (from the NAND gates) across the LEDs (blue trace) every 10ms to avoid flicker.
U8, configured as an adjustable PWM oscillator, generates the higher frequency PWM signal (green trace).
The 50kΩ Pot U9 adjusts the PWM duty-cycle to control the LED brightness (shown at the 50% brightness point).
Regulator U7 (7812) generates 12V for the control circuits, since they can't operate directly on 24V.

The MOSFETs can be just about any N-type (bottom) and P-type (top), that can carry the LED current with low voltage drop (low on-resistance).

 

@crutschow
V(m1,m2) is the drive current? It still drives only half the LED's at any given time.

 

@crutschow
V(m1,m2) is the drive current? It still drives only half the LED's at any given time.

It's the drive voltage.
It alternates to drive the (as I understand) inverse connected LEDs on two wires (below from the TS's post).

It seems most single-colour holiday light strings are wired with every other LED reversed,

 

It's the drive voltage.
It alternates to drive the (as I understand) inverse connected LEDs on two wires (below from the TS's post).

But when one set is on the other is off, which is a "flicker". I proposed a on-on scenario to avoid the "flicker".

With one set on and one set off, the PWM frequency needs to be about 6x that of the main flip-flop frequency, so if main is 120Hz the PWM needs to be around 720Hz, if using your ckt that shows 6 cycles in each PWM segment. It perhaps will suffice for just a visual eye thing, but could pose issue in other cases (light sensors and such, etc). If the rope of LED's cannot be modified then I think your ckt is the easiest way.

 

Awesome! Can't thank you enough for this design and the discussion is interesting too. I won't be able to build it for a while but I will let you know how it goes. For context we bought a pre-lit tree with 750 tungsten bulbs in 2003, in recent years it's come out of storage with several strings dead (despite ballasting them with an old electric fire to reduce the surge) so I'm really looking forward to this update. Not to mention the power saving.
Best wishes