I’ve now received the v3 revision of the FidoLight PCB boards, and initial testing is promising. As you can see in the photo below, I’m now using all SMT components, including the pin headers used to mount the Teensy 4 MCU. That will allow me to mount the board directly to the metal enclosure (i.e. heat sink) that I described in the previous update.
Some improved component choices resulted in slightly lower power dissipation (loss) in the LED driver circuit, but I think a little more work is required to achieve the ~80% efficiency promised in the LM3405 data sheet.
Due to really severe shortages of switching-type voltage regulator chips, I used a linear 12v-to-5v regulator (lower left in photo). Since the expected current on the 5v supply is expected to be about 120mA, the resulting power dissipation in the regulator is 0.8W. That’s not exactly helping my thermal management situation, but more on that below.
Distributed Lighting Controller (DLC) Concept
Some of your comments have made me feel a bit guilty about using such a powerful/expensive MCU to control a single light fixture. So I decided to make two significant changes to the FidoLight concept:
- The 4-channel constant-current driver circuit will be implemented on a small PCB to be mounted inside the LED housing (see Update #1), using the housing as a heat sink. The board will receive 12v/Gnd and 4 PWM signals via a weatherproof Cat5 cable.
- The MCU and related circuits will be implemented on a separate Distributed Lighting Controller (DLC) board. Each DLC will be able to control two FidoLights, as well as an array of other LED fixture types. The low power consumption will allow the DLC to be housed in an inexpensive plastic, waterproof enclosure.
The DLC board can be configured to drive any of the following LED types, in many different combinations:
- One or two FidoLights, as described above.
- One or two of these 48W RGBW flood lights. These operate on 24VDC and incorporate constant-current drivers, but require PWM signals that can sink the full LED current, up to 500mA per channel.
- One or two non-addressable (analog) RGB/RGBW LED strips operating at 12v or 24v. The total power per strip can be as high as 100W. The control methods is essentially the same as for the 48W fixtures above.
- Up to 8 addressable LED strips operating at 12v (e.g. WS2815). The total power is only limited by cabling considerations.
- Multiple addressable LED flood lights operating at 12v or 24v. Very few of these exist in a form that can be used outdoors, but I recently stumbled upon this 10W fixture, which might be a reasonable substitute for the FidoLight (except that its RGB only).
The DLC will use the same 8-conductor 24AWG Cat5 cable to receive power and the RS-422 serial bus, and will still have provisions to daisy-chain multiple DLC on the same cable. The on-board voltage regulator has been redesigned to accept either 12v or 24v, with the restriction that all DLCs on the same bus must use the same voltage. Although 6 of the cable conductors are used for power/ground distribution, there may be cases where additional current carrying capacity is needed. To address this, the DLC will have provisions for an auxiliary 2-conductor power cable using up to 16AWG wire.
Voltage Regulators (Argh!)
It seems like the global supply of buck-type switching regulators has almost completely dried up. The one that are available are very expensive, and require a large number of external components that must be very carefully selected to achieve optimal performance. For the foreseeable future, I’m switching to an approach that I used when I first started designing my own boards: buying complete DC-DC converter modules.
The module shown at right accepts an input of up to 28v, and has a sustained output of 500mA at 5v. At $3.00 USD, it’s actually less expensive that a discrete design, and takes about the same amount of board space. SMD versions are also available, but are less common.
I’ve completed the redesign of the FidoLight board (now on v4) and am ready to send it out for fabrication, pending completion of my v3 testing and optimization. The DLC circuit design is done and I’m now doing the PCB layout. I’ll probably wait a couple of weeks to gather some comments before I commit to anything. Thanks for your input!