Afterthought question regarding the quantity of LED chosen.
Context: I was looking at optimizing power to use medium size transistors in a higher current design, and was looking at the quantities of LEDs for each color and the voltage gap (calling the gap Vsupply - (Vf * Qty LEDs). In a linear circuit, that which does not become light would become heat.
The optimal design would set the quantity of LEDs be as many as one can fit budgeting for the transistor Collector-Emmiter operation. In looking at the quantities above, 10x3.3Vf for Blue would make 33V = 3V gap. 12x2Vf for Reds (rounding) make 24V = 12V gap. And 6x3.3Vf for Green make 19.8V = 16.2V gap.
With the femtobuck (switching regulator), that should be no issue.
With the current mirror however, holding the driver transistor with a larger gap (Vce) will require a larger power transistor. Instead, one can manage the gap by increasing the number of LEDs and keeping the Vce to the operating minimum.
Is there a reason to limit the quantity of green LEDs to 6, or red to 10?
A gap of 16V @ 350mA would require that transistor to dissipate 5.6W. If instead of 6 one uses 10 green LEDs the gap would be reduced to 3V @ 350mA = 1W.
The current mirror may not be a good fit for a large voltage power supply, unless precise control of a very large quantity of LEDs is required.
Nevertheless I was looking at how to adjust the design for the larger LED current (350mA) and thought I would run the question on the specific quantities of LEDs.
Yes - light output. The green LEDs are about twice as bright as the reds, and the blues are also very bright. So the goal was to try to get an even amount of light from each channel, thus the uneven quantities of each LED on the board. We specced them all at 350mA because thatās close to the output of the femtobuck (330mA), so weād be running them a little under full power, hopefully extending their lifespan).
Great!. If that is the reason, then there would not be a problem -other than the cost and the space in the board- using 10 green LEDs running at a lesser current, which would achieve the same light output of 6 green LEDs running at maximum current.
As I said earlier, this design may not be the best fit for that level of power, but is an interesting challenge to contemplate.
Space is a bit of a concern as well, but squeezing in a few more LEDs would probably be possible with a few modifications to the board. Weāre trying to keep them close together because the entry port of our integrating sphere isnāt especially wide and we canāt really make it any bigger.
Iām not convinced we need as many IR LEDs as we have (but we wonāt know until we test that, at some point), so potentially the space we use for those could be occupied by other colors.
and itās possible we may end up needing to use your design after all, since the DACs weāre using arenāt available again until February, and weāre down to our last three. Iāve had two that were bad, out of the box, which seems unusual. Canāt find a supplier anywhere. If these last three I have now donāt work, we may be redesigning in a hurry!
One of the questions of this design is how to run higher current LEDs. Keep in mind the design above is already running at 100mA per string.
Two ways to go about it:
If multiple strings are needed, larger transistor drivers would be required.
If one string is enough, an alternative is to use multiple transistors on the mirror.
As a simple test, I left the current resistors as above (so Q4 is running at about 90 mA), and combined used Q5 and Q6 and Q7 and Q8 collectors to drive a single string of high current LEDs. These are 8 White LEDs with a Vf of 3.2V which makes the total 19.3V and are running off a 24V power. About 6.8W total.
Here the result, multimeter measuring current through the LEDs.
Needless to say, the onboard regulator now needs a good heatsink.
