Important update - read this!
Thanks to all the changes in components, board layout, etc over the last few pages, I can imagine it would be hard to piece this design together. I appreciate all the input I've had. This morning I tried to clean things up and make a single post collecting all the changes, updates, etc. I also updated the BOM, got all the part numbers from digikey, and did some other housekeeping. If you've been following along in the thread, or are new to this thread and are considering trying these drivers, please feel free to
ignore the previous information and just work from THIS post. Besides some tweaks and improvements, there are some outright errors earlier in the thread. Read on. . .
First, a recap of the goals:
1) A DIY driver similar to buckpucks, but more efficient, cheaper, and with capacity for a higher number of LEDs.
2) The option to do all through-hole parts if the builder desires.
3) The option to customize for different drive currents.
4) PWM dimming, as from an Arduino or other source.
I've had about half a dozen prototypes working over the last 4 or 5 months. The prototypes I've built to date were run at 350mA or 500mA, with a 24v power supply, running 8 LEDs each.
The prototypes were built on protoboard as individual drivers (i.e. one driver per circuit). The new approach, outlined below, puts TWO drivers on ONE circuit, so you can drive 16 LEDs from each PCB, with "two channels" of LEDs. This makes for a smaller footprint, slightly less wiring, and slightly lower cost. This concept of doubling up is untested but we'll have prototypes soon.
Next, another rev of the board, including the SMT parts, and re-numbered. I'm going to (hopefully!) consider this the final version of the board. Besides the changes discussed so far, I increased the pad spacing for two of the diodes because when I double-checked the package specs, it would have been tight.
Also, a revised BOM. This is with the SMT resistors, but includes through-hole resistors if you want them. Also, this includes the "better" caps discussed a page or two back.
Also, an
important note. Earlier in the thread I was calling the PWM resistor a 1k resistor. That is incorrect. It should be 10k, according to the datasheet and the prototypes I have on protoboard. My mistake.
Another important note, if you have not been following along. The four "tricky" resistors (R1, R4, R5, R8) have multiple options. You can get an SMT resistor, or one or two through-hole resistors. This should let you parallel two through-hole parts to get the correct value if you need to. The pads you don't use are simply left unused. In other words, there will likely be unused pads on your driver, don't worry that you missed something - it's supposed to be that way.
The above BOM is for a 500mA driver. If you want other values, change:
1) Rsense (R4 and R8). Common values would be: .68ohm for 350mA, .33ohm for ~700mA.
2) The inductor (L1 and L2.) If you want 700mA, I would probably try for a bigger inductor. It should have at least 1A current rating, preferably more, and you might want to look for a value around 120 - 150uH.
Download the design spreadsheet for the NCP3065 from OnSemi's website if you want to play with the values and see the math.
I will be ordering the above BOM and building some prototypes of the double-driver within the next week or two, then doing a batch order of the PCBs. If people want to try this instead of waiting to see me do it, PM me and I'll email you the Eagle files.
