Cap, I have no certain memory just a big fuzzy thought that large series-parallel arrays of LEDs with only the one resistor feeding them were frowned upon in this type of design? Wrong fuzzy memory, or is there something to that? No drawback?
Thanks for asking a question I can actually answer
Up until Paul wrote up his excellent explanation of the elegant regulator circuit he designed for the FrankenBebi, I was a little unclear on how the circuit functioned as well.
Anyway on to your question... Generally speaking you would be correct for a couple of reasons.
First, a single (or even a group) of resistors for current control is not advisable in a situation where the input voltage may vary. Onboard, we could have from less than 12Vdc (batteries at less than 25% SOC) to over 18Vdc (battery equalization cycle). With a fixed resistance acting as current control, LEDs would dim at low voltages, and at higher than "nominal" voltage excess current would be dissipated as heat.
Second issue is parallel/series arrays are a problem in and of themselves. If we loose one LED in a series "leg" the current (say 100mA) that was divided among 5 legs is now supplying only 4 legs with a resultant 25% increase in forward current. If this exceeds the max forward current of the installed LEDs the magic smoke will escape, and we have a dead FrankenBebi
The latest design deals with these issues in the following manner.
Second issue first... The regulator circuit as designed is only driving the Cree C530D LEDs at about 60-70 percent of their rated max forward current. If we should lose one leg of the array, we are still below the rated max forward current for the balance of the array, hence no magic smoke..
Of course this situation would entitle you to a warranty replacement, but the FrankenBebi would continue to function until it arrives. The earlier change in board design
staggers the LEDs in any given leg around the radius preventing an entire 45 degree sector from going dark, as no LEDs in the same Leg adjoin each other.
As to the first issue (current control)... While the circuit contains a number of resistors, none of them actually provide current control as such. The current control circuit (which is on the low-side of the array) Uses a combination of 2 resistors (R3, R4) an N-channel MOSFET (Q3) and a general-purpose NPN transistor (Q2) Paul's explanation is as clear as it gets so here is a quote from his document
The regulator uses the base-emitter junction of Q2 to sense the voltage across R4. When this voltage exceeds the 0.68V forward bias voltage of this junction, Q2 begins to conduct current through R3, reducing the gate voltage on Q3 and reducing the current through the LEDs. If the current drops slightly, the base-current in Q2 is reduced, which causes the voltage at the gate of Q3 to rise, increasing the LED current.
This current regulator is insensitive to battery-voltage variation, maintaining a fairly constant LED current for battery voltages between 11.25Vdc and over 20Vdc.
Here is a local copy of the schematic for reference.
Hope that answers the question... Nice exercise for me to prove to myself I understand the new circuit. My previous background is in the mechanical end of things (ME not EE) so this whole project has been a real learning curve on the electronics end. I will be posting the latest (and last?) revision of Paul's circuit discussion paper (doc revision 1.2) which includes another section on testing, expected vs real test values, and troubleshooting. It will be accessible via the same links in the above post
in an hour or so...
Good opportunity for me to thank you for your earlier suggestion about the "icecube" case. We are well on our way to solving the issues on that one, and it should be "ready for prime time" about the time our new batch of PCB's arrive in a few weeks.
(aka Cap' Couillon)