It's amazing how much using flux helps. I used to think I was the worst solderer ever, but then someone told me to use flux and now I'm … not the worst solderer ever.

That said, flux fumes make me ill. Either have some really good ventilation or hold your breath while it burns off.

Ahh.. I think that's it. What I thought (from the photo) were gaps, are just the black silk-screened markers for placing the resistors. So those need ground.

Which means you'll need to repair things. I recommend STRONGLY that you avoid soldering anything directly to the resistor pads -- the resistors will likely detach.. ugh.

So, three more cuts (where we should have put them in the first place), and then repairs to the original cuts on the ground:

They do connect to what used to be ground. Also connected to the same resistor on the gate end but on the flip-side of the board are those ceramic resistors which then attach to lines leading to the ICs.

I almost DID cut the board originally in the NEW locations. Dang.

Note that you can hook up one of them to try first (the top one in the photo, perhaps), and then do the others only after verification that things are working.

To repair the cuts, just scrape some copper bare on each side of the cut (Dremel tool?), and lay a small piece of bare wire across the gap as a bridge. Then solder it in place.

Cheers

Still doesn't work. Gate voltages are different now. Decreased by about .2 to .3 volts. All three still read differently.

Continuity is broken and connected where it should be as per the newest diagram.

I can still light the LEDs by shorting drain to either gate or source on any of the FETs.

Well, it is possible that Gate-to-Source voltage "matters", then. I don't know enough about that stuff.

Probably best to set it aside for a day and read/think about it some more.

It does matter. And it's the only possibility left anyway, so..

One could conceivably correct for that with new resistor values, but I think your best bet now might be to just restore the original circuit, and get LED strips with the correct polarity.

Or get a controller that works with your existing LED strips, if they exist (I didn't find any).

Oh well. We tried.

Actually, if you want to pursue this further, we should be able to calculate the correct resistance values, and it might work out that only the larger three (component side of board) resistors need changing.

To do this, we need to know what resistance values the three tiny ones (copper side of board) have, and similarly for the three larger (component side) resistors. And measure the voltage input into the larger resistors from the chip that controls them.

It might work out nicely, or not.

Edit: Background: each pair of resistors for each transistor, form a simple voltage-divider circuit, scaling the output from the control chip into a voltage appropriate for controlling the transistor. The voltage we need is whatever is appropriate to achieve the correct Gate-Source differential for the transistor. We don't know that value (no datasheet), but we can figure it out from the measurements above and those already posted earlier.

I don't have any spare resistors around here, so unless I can extract some suitable ones from some other piece of junk kit I might have lying around, it means I have to go out and get them.

But, for shits and giggles, let's see if this makes any sense....

The SMT resistors say "103" on them. The ceramic ones are 390 ohms according to an online calculator. 4 bands, orange, white, brown and gold.

Okay, so this multiplies the control signal by:

10000 / (10000 + 390) = 0.96

Mmm.. that hardly seems worthwhile, but, okay
(the resistors also provide some protection to the controller).

And what is the (full brightness) control-voltage you see, on the input to the larger resistor from the control circuitry?

And what is the voltage (with respect to ground) that you see on the rewired Source when the LEDs are switched completely "off" by the controller?

The idea is that we then modify the first resistor (if possible) to achieve a difference of (0.96 x control-voltage) between Gate and Source.

Cheers

I don't think you will be able to get it to work without some more modifications.

The 10k resistor is a pull-down resistor and the 390ohm resistor is just protection for the uC.

MOSFETs are voltage-controlled, while regular bi-junction transistors are current-controlled.

The MOSFET is connected between the cathode-side of the diode and ground, so you will have a voltage 0-5v between source and gate depending on intensisty of the LED. When it's connected to the anode-side of the LED, as the modification is doing, you would still need 0-5V difference between source and gate, but now you have the LEDs raising the reference for the source and you will never get more than 5V out from the uC, making it not work.

Unless I'm mistaken, what you need to do to get it to work, is to replace the MOSFETs with their opposite equivalent connected between anode and Vcc, and a transistor/MOSFET between uC and gate (with neccessary resistors).

We come across this every so often, we supply equipment that drives the LED signs you see by the side of the road.

From what we see customers doing/using, the most "common" way that these are constructed is with a common voltage and you switch ground in and out via a PWM, occasionally we've come across signs that are configured the other way, where the grounds are are common.

To be able to drive the common ground configuration, we have to supply an adaptor board which like StigOE points out, has the "opposite" FET configuration. (i.e P & N channel FETs swapped)

The alternative method, is to swap the LED's so that the A & K are the other way round, but that's not possible with SMD RGB LEDS because there's no rotational symmetry.

The guy I bought the controller from says all his LEDs are common voltage - haven't heard back from the other person yet.

I think the easiest thing to do at this point is to rework the board back to original function before it gets too complicated to do so, and pick up second set of LEDs. I can hopefully pass the current set along to a friend who doesn't need to use the alternate controller.

Everyone, especially Mark, thank you so much for the assistance. Even though the rework didn't work, it was fun to do and good practice. Even though I really need a new soldering iron, or at least a new tip on this crappy one, I think I did a half-decent job. The anticipation of success by performing the mods was also a treat on what would have otherwise been a relatively dull day - weather was crap which pre-empted a few projects.

The easiest way to change the function would be to use 3 external PFETs; currently it's a low-side switch, with the gate voltage going above ground to turn the FET on and let current flow from D to S.

To change this to a high-side driver (enabling the current flow at the supply voltage side), you'd take a PFET and wire it like this:

- Source of PFET to LED power supply (body diode cathode points to power supply)
- Gate of PFET connected to PFET Source with a resistor - say 10k or 100k
- Drain of PFET connected to R/G/B LED string power line

...then, connect the gate of the PFET to the appropriate R/G/B of your power control board.

The PFET will turn on when the gate drops below the source by >Vgs. When the lighting controller NFET is off, the PFET gate is pulled up to the source by your resistor - ie, no current will flow.

When the NFET in the LED controller turns on, the gate of the PFET is pulled low, meaning Vgs goes negative (Vgate < Vsource, as Vgate will be near ground and Vsource will be your supply voltage). Current will then flow and the LED will light.

Ok, so you'll need to find yourself 3 PFETs (at least 20V rated; getting 60V ones would be best for robustness), but you should be able to do this all "in the wiring" as it's just a single FET and a single resistor per channel. You can also try just one channel to see how it works.

Ok, so call me cynical but this has all the hallmarks of "pre-spam".