constant current control for RGB LED with DC-DC step down module

[waspinator]

I'm using a DC-DC step down module with constant current control, but I'm having an issue if I'm not using the negative terminal on its output. The negative terminals seem to be connected to each other and to ground, but I can only get constant current control to work when connected directly to the module's negative terminal. I've tried this with a few different models, and they all work the same way. Is there some feedback happening when using the negative terminal that doesn't work when using system ground?

I'm trying to use three of these modules to control the three channels of an RGB LED, and I was hoping to only have to use 4 wires (RGB + ground), but I guess I would need a separate negative for each of the colors?

[paulca]

What happens if you combine the two diagrams.

Add a ground link on the top one, right hand side.

Although first I would check there little to no voltage between IN- and OUT-

[chris_leyson]

Is there some feedback happening when using the negative terminal that doesn't work when using system ground?

The current sense for the LEDs is internally connected between OUT- and IN-. And yes unfortunately you have to use separate ground return wires for R G and B. Google LM3404 for a popular LED driver chip that uses this sort of current sense mechanism, it's explained on the first page of the data sheet "Typical Application Diagram"

[Whales]

The DC-DC module may be doing its current measurement on the 'low side'.  In other words: inside the module a small shunt resistor for measuring current may exist between the negative output terminal and ground.

EDIT: When you say that the negative output terminal is connected to ground, how did you measure this?  Did you follow the traces or use a DMM?   If the latter then there might still be a small shunt resistor there (less than a few ohms) OR an entire low-side switching mosfet that was 'on' at the time of measurement.

 It would be cheaper/simpler to low-side switch, even though it has some disadvantages, so they may have done that.

[waspinator]

My multimeter reads 0.000V between the two negative terminals, but when I hook up the output negative to the input negative constant current stops working.

[Whales]

My multimeter reads 0.000V between the two negative terminals, but when I hook up the output negative to the input negative constant current stops working.

That means no current is flowing through the negative terminal and into ground.  Even if there was 5 ohms between the neg terminal and ground, ohm's law says you will see no voltage unless there is current flow.

Try hooking up the module in the way you know works properly and then take this same voltage measurement whilst it's running some leds/loads.

[waspinator]

That means no current is flowing through the negative terminal and into ground.  Even if there was 5 ohms between the neg terminal and ground, ohm's law says you will see no voltage unless there is current flow.

Try hooking up the module in the way you know works properly and then take this same voltage measurement whilst it's running some leds/loads.

Yes, when I measure the voltage between the two negative terminals with a load I read 0.02V. I guess I'll need those separate negatives. Thanks

[Whales]

Yes, when I measure the voltage between the two negative terminals with a load I read 0.02V. I guess I'll need those separate negatives. Thanks

I might still be wrong, however.  0.02V is a very small number.  That could just be PCB track resistance.

Your best bet to understand what is going wrong is to look at the PCB and try to draw a schematic of what it does on the low-side.  Reverse-engineer it (assuming schematics are not freely available).

[waspinator]

I think it works similar to the module explained here. There's a output side current sense resistor on the negative terminal that works to control the current.

https://youtu.be/NyTze5HU4nI?t=3m18s

[paulca]

I bought a current meter on ebay that had a sense voltage of 0.075V per 10A!  So 0.020V isn't that small for current sense. 

What was the load to produced 0.020V?

It is likely that the current sense is between OUT+ and OUT- and using IN- as the ground return means it bypasses the sense resistor and constant current doesn't work.  That would suggest it went to max current with no limiter.

[waspinator]

yes, that's probably it. The load was 20W (10R, 20V, 1A).

[chris_leyson]

I think it works similar to the module explained here.

I think you're right and that's probably how your module works. I must admit I've always had to use a single chip solution when designing LED drivers purely because of space constraints, LM3404 for fixed current and TPS92512 for variable current. Post a picture if you can.

Chris

[waspinator]

I think it works similar to the module explained here.

I think you're right and that's probably how your module works. I must admit I've always had to use a single chip solution when designing LED drivers purely because of space constraints, LM3404 for fixed current and TPS92512 for variable current. Post a picture if you can.

Chris

Do you have some favorite reference designs for either the LM3404 or the TPS92512?

[Mr.B]

There is the current sense resistor in the -ve return...