An overall opinion of a led dimmer scenario

Thread Starter

jeffkrol

Joined Dec 8, 2015
42
This isn't going to be very specific to a circuit but there is a design that is used in some aquarium lights that I would say goes outside the box in function.
Specifically some lights have constant current drivers on their circuit board (one in particular a pt4115)
Instead of using the dim pin they dim it err "globally" on the power line to the board.
Like one would with the numerous constant voltage ribbons of leds.
One light in particular uses a TC-420 for this purpose.

The tc-420 PWM dims the main 24v line to the chip (and of course diodes) using a MOSFET output.
Point is instead of using the "on board" dim circuit it constantly (at about 500Hz duty) fires the driver.
Seems it could be problematic.
The light I modeled above someone had problems with bricking a number of TC-420's

There is a Japanese made light that people modify in the same manner.
CC drivers on board, just pulse voltage input to dim.

Soo any feelings on constantly "flicking" the voltage to a constant current driver?

The theory, tc-420's are cheap and driving leds at constant current is better over constant voltage is sort of sound.
Mentally, having the constant current chip constantly starting up at 500Hz duty cycles seems err crude.

As I wrote this it came to mind that the light may just stress the mosfets in the TC-420 and has nothing to do with the "driver design" exactly.
TC420 should be good to 96Watts per tc 4a spec and 24v ps per channel.
The specific light also had diodes burn out at a more than usual rate but that could just be the quality (est. current was 450mA) or poor heat sinking.


Question still applies on the overall topology.

Edit for specifics on light
composed of 3 channel with 4 6 diode led boards. 4 boards need to be wired in parallel and have an effective current draw of 4 x 450 or 43W total
Well within the 4A limit of a channel on the tc-420.
Thing is a manual dimmer works fine as to not dying. Most of those are rated at 6A though
 
Last edited:

LowQCab

Joined Nov 6, 2012
4,025
I see that ~40 people have looked at this thread and not commented,
I'm guessing that the consensus is that You should replace everything with
a system that is more coordinated and pre-planned-out.
.
.
.
 

Thread Starter

jeffkrol

Joined Dec 8, 2015
42
I see that ~40 people have looked at this thread and not commented,
I'm guessing that the consensus is that You should replace everything with
a system that is more coordinated and pre-planned-out..
Actual it is how a commercial product "thought it out".
It's weird enough to peak my curiosity.
Certainly not how I build my lights.


But thanks anyways..

Maybe I'm not explaining it well
It would be "sort of" like this schematically I think
Pretty obvious that I have the tinker toy knowledge of electronics btw.. ;)

With the mosfet/555 as the dimmer coupled to a buck constant current driver (which in
the real case actually could be dimmed without it and a different controller but I digress)
I know the diagram isn't accurate but it is the "concept" as I understand it.
The tc-420 uses NPN mosfets and + as the common ground to the led strings.. pulsing
the low side (again as I understand it)

Pulse the supply to the buck driver...

buildled.JPG
Help me learn here.. Apologies for my ignorance.
Actual parts, a mixture of a 24v switching power supply, constant voltage program/controller, and constant current drivers on board.
buildled2.JPG
 
Last edited:

LowQCab

Joined Nov 6, 2012
4,025
Very interesting ........
But way too much additional complexity just so You can keep the original Circuitry.

I would start from scratch and use a Buck-Converter as a "Switching-Current-Regulator".
I've worked-out how to do it,
complete with a simple PWM-Input directly into the Buck-Converter-Chip using the EN Enable-Pin.
No-Synced-Blanking,
No Low-Voltage-Power-Supply,
No programming a Controller,
Extremely low Parts-Count,
Up to ~5-Amps total Load.

Unfortunately, if the LED String requires more than 60-Volts,
an external, higher-Voltage-rated, FET-Switch is required.
But, quite often, it easy to re-configure the LED Strings for a lower Voltage requirement.

505-LT1074CT#PBF-ND
.
.
.
 

ErnieM

Joined Apr 24, 2011
8,377
Inside the controller, bottom image right side, you see 4 largish transistors marked R G B W and P. Well, that's for Red Green Blue White and "P" for ???

These typically act as low side switches for PWM control of brightness by switching the LEDs on and off quick enough so your eyes perceive the average brightness. Something like below which assumes the LED strings can control their own current.


Untitled.jpg

(Looks like your got the version where they left out the ESP8266 Wi-Fi controller.)
 

Thread Starter

jeffkrol

Joined Dec 8, 2015
42
Inside the controller, bottom image right side, you see 4 largish transistors marked R G B W and P. Well, that's for Red Green Blue White and "P" for ???
'P" is for positive.. All these use + as "common"...
RGBW are the negative side..
 

Thread Starter

jeffkrol

Joined Dec 8, 2015
42
Thanks for the replies but nobody seems to be actually getting my question.
Simply stated say you have a buck constant current driver but to dim you PWM the voltage input to the driver NOT using (or it doesn't have) it's "internal" dim circuit..
Not a "how to do it" which I have a basic understanding of.

(Hypothetical)
Say I'm using a Meanwell LDD-HW.
Instead of using the 5V pwm signal wire and having the driver circuitry doing the dimming I'm just "pulsing " the power supply input TO the driver using say a mosfet switch.
Basically killing the power to the LDD on a periodic basis to dim the Leds.


I have no NEED to do it this way (nor would I) I was just wondering about any ramifications of doing it this way.

A long time ago I (I believe) started the tc-420 modification to use as a controller with Meanwell LDD's using a tap off the mosfet gate to use as a +5V PWM dim signal. I probably wasn't the only one but felt like it at the time. ;)
https://www.plantedtank.net/threads/how-to-use-tc-420-to-control-ldd-drivers.980633/


IF there are no drawbacks I could just hook the LDD's directly to the tc-420 without the gate wire.
This would have limitations of course like no strings of diodes needing more than 24V cumulative V(f) at the chosen current. Well 21v since you "lose" 3-4V in the ldd.
The TC works with ps 12-24 and I found as low as 9v at least, lower limit may be dependent on the properties of the voltage converter to the logic circuit.
Which adds another versatility to the tc-420.
Using a 10v ps one generates a 10V PWM "box" to dim drivers that can use that type of circuit.
And with a 10v ps and some tweaking possibly a 10v analog output using a r/c filter to smooth it. I assume more complicated (adj power supply to guarantee 10v and getting 10v analog is a bit err ..sloppy.


As of now I think it would work (for others) though my gut tells me in the deep down scheme of things there must be a catch to this.

There is little practical advantage though besides some cost.
Only reason "I" went down that road (modified tc-420) was to dim Meanwell-ldd's
All the "native" 5V PWM controllers were at least 2X the cost of a TC-420.
Also though the software was Chinglis and a bit confusing it had a better multi-point program than most other "commercial" dimmers.
And of course to see if it works. ;)

Of course 100% DIY using Aduino, Pi or the esp32 boards makes this all a bit old school or crude.
BUT you don't need to program.. that's a plus.

Sorry for the tangent..
 
Last edited:

Thread Starter

jeffkrol

Joined Dec 8, 2015
42
Tinker toy version..
What could go wrong?
buildled4.JPG

Minor side note TC + is bridged internally. Wouldn't matter what screws you put it on.
 

Thread Starter

jeffkrol

Joined Dec 8, 2015
42
That is possible, but there is also a power transistor for the "P" channel. Why switch the power when you control all the grounds?
See I'm not asking "why" since it is already done commercially. I wonder about the "theory" behind it ..and any ramifications.

For reference how to do it right..
TC420(1).JPG

Borrowed and corrected from here:
https://www-mikrocontroller-net.tra...uto&_x_tr_tl=en&_x_tr_hl=en-US&_x_tr_pto=wapp


"Industry standard" for strip and other constant voltage strips uses a logic level NPN Mosfet to switch the low side.

This worked to "reverse" the mosfet output to pnp to "hack" a light that ran negative ground and had 3 "plus" channels.
Using the orig pulsed negative output to run the gate of the PNP Mosfet.


Light was a Finnex non-dimmable planted plus.
tc420finnex.JPG

Like I said. tinker toys..

HUH., just noticed the board had RGBWP(?) for channels. So that is what you must have been referring to?
Maybe it's "pink"
Currenty it would ususally be RGB WW CW for warm and cool white.:)
tc420channels.JPG
 

Attachments

ThePanMan

Joined Mar 13, 2020
773
What I'm seeing:

@ 3:46 in the video he pulls out the LED's from the housing. Other than the LED's there's no electronics internally. So all you need is a dimmable power supply or variable CC regulator of sufficient current. Why all the complexity with PWM, MOSFET's, inductors and capacitors? If you're modifying anything then simply ditch the electronics package for your own?
 

Thread Starter

jeffkrol

Joined Dec 8, 2015
42
What I'm seeing:

@ 3:46 in the video he pulls out the LED's from the housing. Other than the LED's there's no electronics internally. So all you need is a dimmable power supply or variable CC regulator of sufficient current. Why all the complexity with PWM, MOSFET's, inductors and capacitors? If you're modifying anything then simply ditch the electronics package for your own?
"sigh"
Probably my fault, a lot of extraneous data here.

ONLY thing that is important is this:


Can anyone see any issues with this?
This company makes them like this.. and they work.
https://makemyled.com.au/
Only difference from the above is the constant current drivers are "on board" not a module..


Thing is it's "outside" the box thinking.
Continuously "dimming" the step down driver..
 
Last edited:
Top