Cell phone car chargers are plentiful, inexpensive, and rarely compatible with a new phone. An upcoming car trip created a need for a small variable supply I could use for the duration. These factors, and a mention of hacking phone chargers on one of the sites I regularly read inspired me to crack open an old charger.
First, to the internet with the IC. A search revealed that the On Semi MC33063 is a multipurpose 1.5A switching regulator. The device is not fixed voltage, special purpose, or exotic in its requirements. Further, the reference design on the datasheet indicates that output voltage is set with a resistor divider. So far, this is looking promising. It is worthwhile to proceed to the next step, which is reverse-engineering the device.
Reverse engineeringThere are multiple methods to reverse engineer existing boards. This particular board is a single-sided PCB with through-hole components, so nothing complex is called for. My method used was to generate a netlist by hand, using the IC as starting reference. Looking at the solder side of the PCB:
Start by choosing a set of connection points as your node 1 - since we are using the IC as starting reference, U1-1 (U1, pin 1) will define node 1. Now make a list of all parts and their pins that are connected to that node (connected to U1-1). For this circuit, they are one pin of resistor R1 (arbitrarily called R1-1), U1-7, U1-8, and two non-populated holes. Moving to the next pin of U1, we repeat this, until we get a complete list of nodes that connect to U1. Finally, we go back and get a list of connections to nodes that don't connect directly to U1. You will get a netlist like the one below:
n1 U1-1, R1-1, U1-7, U1-8, non-populated hole, non-populated hole
n2 U1-2, L1-1, D1-1
n3 U1-3, C2-1
n4 U1-4, C2-2, D1-2, Vout-, C3-2, D2-1, R2-1, Vin-, C1-2
n5 U1-5, R2-2, R3-1
n6 U1-6, R1-2, C1-1, F1-2
n7 F1-1, Vin+
n8 D2-2, R4-1
n9 Vout+, L1-2, R4-2, R3-2, C3-1
Note that many of my naming conventions for parts and nodes are arbitrary - they must just be consistent through this process. In this case, the cathode of diodes and the positive side of polarized capacitors were pin 1 of the device, and non-polar devices were simply a matter of order encountered. This is a nice board to work with - the parts are all labeled with designators. If they are not, make a sketch or take a picture and assign your own labels to parts to help generate the list. As a reality check, go through your connections list and make sure that all pins land in a node.
From this list, we can start to construct a schematic. Again, we will reference this to the primary (only) IC. Note that node 4 is the ground node, node 6 is the fuse-protected input voltage node, and node 9 are is the output voltage node. These reference can help make drawing the schematic less convoluted. Start drawing the parts from the IC, and get them all on the page. Fiddle and re-arrange until it looks logical. In this case, it helps greatly that a) the device is simple, and b) it seems to follow the reference schematic given in the On Semi datasheet.
Repurposing (aka Hacking it)Now the fun begins. You have a schematic and a datasheet that is very helpful in designing devices based on the the IC. For a hack, we're primarily concerned with changing the voltage - I want to deliver 3-10V out, adjustable with a precision resistor. The datasheet provides the design equations for setting the output voltage (translated to our schematic above):
|Vout| = 1.25 ( 1+ (R3/R2))
R3 = R2*((Vout/1.25) - 1)
If we set R2=1 for now, the solutions for R3 with Vout@(3,10) are (1.4,7). This tells us that R3 has to have an offset factor of 1.4 and a range of 5.6. Scaling R2 up to something reasonable, 1k, R3 is now 1.4k - 7k. As I had a small 5k multiturn pot available, this worked out well. A fixed 1.2k resistor in series with the 5k multiturn provided the offset needed. A bit of bending, futzing, and fiddling later and the multiturn + 1.2k fixed replaced the 6.98k. The 1k fit the space vacated by the 2.61k neatly.
The first picture is before modification, the second is after modification. If you look very carefully, you can see the fixed resistor pressed up against the upper left corner of the multiturn trimmer. Testing this before reassembly, output ranged from ~2.8V to ~8.8V. This is below specification, but acceptable for my immediate application.
Going further / final notesIf I redesigned with parts ordered instead of parts on hand, I would probably use a 1k + 10k multiturn for R3, which would allow a lower output voltage as well as one almost to the input voltage. While unnecessary for my purpose, an increase in current capacity is possible with this topology - there are some simple reference design changes to allow much larger maximum currents. This design does not have current-limiting either - something that is useful both for mobile projects and development projects.
As an aside, this hack (as has been noted in several forum entries) is nothing novel or breathtaking. It is, however, a worthwhile guide to how to go about reverse-engineering and modifying a product.
Attached: pdf of blank sketch sheet created for schematic drawing.