5W RGBW CC driver
xanadu
Posts: 3,347
I need to run two 5W RGBW LEDs in parallel. 700mA max per LED, so around 2A cont. per channel (LEDs in parallel) would suffice.
Is there a such thing as a CC driver in a TO-220 style through hole package with that kind of rating?
Is there a such thing as a CC driver in a TO-220 style through hole package with that kind of rating?
Comments
Do you mean linear, or switch-mode ? Usually you current regulate per-led, so that's 8 x 700mA drivers
Yes, with an oversized heatsink and 20mm 12V fan per PCB.
SMPS.
Is that top wire actually connected because it doesn't as secure as the other two.
Do you expect a module, with inductors included.
With many LEDs to drive, alternatives are to use a low cost solution per LED, or have one SMPS block, Voltage out, and then linear current regulation from there, with minimal voltage loss.
eg compact (SOT25) feedback regulators (200mV ref) like AL5815/AL5816 can drive external TO220 (etc) power devices, (MOSFET or BIPOLAR) and PWM DIM
Here's a diagram of the existing setup. I'm replacing everything, VIN, uC, LEDs, wiring, etc.
There's an electrical connection. I wouldn't be surprised if the heat shrink was the only thing holding it in place.
The VIN can be 5V to 15V. I can do SMPS before the regulators. They appear to be controlled via PWM, over the full range of brightness. As for modules and there is very limited space available. I've seen a few that fit the requirements but wouldn't be able to squeeze 8 of them onboard.
I ordered some new LEDs so I have some numbers to work with. They are dimmer but I can get the same brightness by modifying the diffusers they're using.
Power: 4 Watt (Max)
Emitting Color: RGBW (Red Green Blue White)
Recommended Current: 300mA / Color
Forward Voltage: R:2.0-2.4V G:3.0-3.4V B:3.0-3.4V W:3.0-3.4V
On Sunday I'll test the brightness and then if it goes well source parts, or a custom PCB. I need 6 sets of controllers, 8 channels each, CC, 6-7V VIN, PWM controlled. It would be nice to have a low part count.
If anyone is interested in fabricating the boards, PM me and we can talk on the phone.
Thanks.
eg simple switching parts like AL8861, will prefer short connections (same PCB) and need access to both ends of each LED.
The AL179x series looks useful, but may be tight on your minimum VIN.
That has a simple PG signal for LED dropout, and can drive up to 4 channels of common anode LEDs, PWM dim.
What is the PWM frequency & control range ?
I'm not sure what the freq or range is on the existing equipment. I'm replacing it with a Propeller and my own code, so I'm flexible.
I'd suggest the KISS but efficient path, and use (eg) 4 x AL5816, plus any suitable decent NPN regulator transistor (not a FET) plus a LM339 quad comparator connected B-C on every LED's NPN.
That LM339 is used to tell the Prop "VLED-too-low", and it ramps a shared SMPS UP - Join 2 for 2 channels.
In operation the Prop+SMPS sits bouncing against that VLED lower limit, and that keeps the linear drop across the NPN to no less than ~0.7V
SMPS-VLED is nominally 3.4V+0.2V sense drop + 0.7V Vce = 4.3V at 2.4A - any SMPS will do here.
With a Prop, one alternative approach to VCE sense, would be to use a series R divider to the Logic threshold (~ 1.65V) from each LED cathode. (8 pins)
An OR test on those shows if any are getting close to dropout.
LM339 gives (0.7)*300m = 0.21W per linear transistor
Simpler 3v3 Prop Threshold sense, gives (1.65-0.2)*300m = 0.435W per linear transistor. (+ more on RED, unless you add a series diode)
Addit: a pair of dual BAT54A can analog 'or' the lowest-of-4 LED-K voltage, & may be simplest.
No MCU needed, and that can feedback to the SMPS directly to set the Linear headroom.
Low Sat NPN is used, good Hfe, perhaps NJT4031NT1G (SOT223), or ZXT690BK (DPAK) or NSS1C301ET4G or 2STD1665 etc - or if you want small, maybe PBSS4560PA,115 ? 3-HUSON (2x2) or Dual ? PHPT610030NKX
I went back and looked at the setup. They're using IRFZ44N to drive the LEDs.
Put Red+Red in series at 9-12V, same with the other colors, as a 8pin led have separate cathode and anode I take it? (always include part number when asking in forum)
use 4 of these: https://www.mouser.com/datasheet/2/115/AL8860-1019719.pdf
Or a single 4channel led driver (but some only do global dimming):
https://www.mouser.com/Semiconductors/Driver-ICs/LED-Lighting-Drivers/_/N-7zhqfZ1yzvvqx?P=1yzvta6Z1z0wbyxZ1z0vzanZ1z0wd8oZ1z0wbz3Z1z0wbyqZ1z0wd73Z1z0wadvZ1z0wc12&Ns=Pricing|0
fancy: http://www.ti.com/lit/ds/symlink/lp8860-q1.pdf
it only do 150mA/channel so use 5 weaker led in series, as this IC boost voltage.
String-LED's share the current but each one do a voltage drop.
or a 8 channel if not putting LED in series:
https://www.mouser.com/Semiconductors/Driver-ICs/LED-Lighting-Drivers/_/N-7zhqf?P=1yzxao1Z1yx7o2fZ1yzt4toZ1z0w877Z1z0wd8oZ1z0w2kx&Ns=Pricing|0
That's a 17.5 mOhm MOSFET, tho the 17.5 mOhm matters little as it will be linear-mode operated.
You just need a low enough Rds on to allow a low overhead, even 100mOhms, will be 30mV at 300mA
With the analog-or option I added above, you do not need the NPN base voltage, so NPN stipulation drops & you can choose any package/part.
A broad search helps these days, as often 'higher rated' parts are cheaper !
eg if I filter 10mOhm to 100mOhm, these compact dual N-MOSFETs look good, if size matters to you. I choose a package that looks easy to cool, and tolerable to handle and solder...
TSM200N03DPQ33 RGG 2 x 30mOhms at 4.5V, 3x3 package with flat-tabs, for good copper connect.
AON7804 2 x 26mOhms at 4.5V - both ~ 18c/1k
These would have a common SMPS pre-regulator, either one for every 4, if you expect future expansion, or one for every 8.
These days, a SMPS that does not need an external schottky can be found.
eg the ST ST1CC40 looks appealing - Sync rectifier and up to 3A, and can be used as either 100mV iSense LED SMPS, or you can divide ~4V DC to 100mV FB, for DC-pre-regulator use.
Thanks for all of the suggestions.
I didn't think it would be bright enough, but I found another layer of diffusion that could be removed. The person that made it was driving it too hard, then reducing the brightness with diffusion. I have the parts on order for the CC driver so I'm still going to make at least one. Maybe more, I have a pile of 4W RGBW LED.
Using a dozen smaller one gives you pre-defused light.