Soft start and overvoltage circuit for charge pumps

[phasetrans]

Hey all,

I'm working on a design that uses the Microchip TC7660H charge pump (120kHz nominal switching frequency). This part has a 10.5V absolute max voltage rating. I decided to design a soft start and over-voltage protection circuit for it, which is shown in the image below. The parts in LTspice are the same parts I am actually using.

-C1 and C3 are the V+ input capacitances from the design. C3 is X7R mlcc. R_load is the rest of the pedal
-Peak soft start current is <600mA
-Overvoltage will trigger between 10V - 10.7V input (i.e. before D2 drop). This depends on Vgs(th) variations of the pmos
-In event of over voltage, the circuit doesn't start
-D2 is reverse polarity protection. I'm not worried about the extra voltage drop vs. a schottky diode

-C2 is presumed a 20% X7R MLCC at the minimum tolerance (i.e. 800n)
-R5 is to limit current on the internal protection zeners of M2. Vgss of this pmos is +/-8V
-R5 was chosen so that the R5/R2 and R1 voltage divider is below M1 minimum Vgs(th) with 18V input
-D1 must be >200mW rated to insure its safety with accidental 18V input
-D2 should be >500mA surge current rated for 1 second

Hopefully someone else finds this useful for their designs, for a run of qty 10 PCBs, it adds less than $0.50 in per-board cost from Mouser.

[phasetrans]

Quick note that I am going to rework this with a current sense instead. The RC time constant isn't reliable enough if there's ringing in the DC supply at higher frequencies.

[phasetrans]

I think this is where I'm going stop for this little idea after revisiting last night. No sense resistance after all:

It uses a pair of PNPs. 2N2907 or similar. When you plug it in, Q1 saturates and limits the current flow into C1 and C3. Because of hfe variations, the saturation current will vary some, but I just want the inrush to be "less" so this variation is fine. R1 chosen from the existing resistors of the overall circuit based on max hfe.

The base of Q2 starts conducting when it goes above 9.1V, which is at about 10.5V total input (D1 diode drop + Q2 E-B drop). Q1 base is brought up to (almost) Q1 emitter, and the transistor doesn't conduct. This protects everything after Q1. R2 could be a little smaller, but the value here is common in the first circuit I am going to build with this.

[vigilante397]

I never saw this when it was initially posted, so this is my first time seeing it. I definitely get the overvoltage protection when using a 7660 charge pump, but what made you want to add the soft start? I can't imagine there are a lot of pedal circuits that would be sensitive enough to need or benefit from a soft start.

Looks great by the way

[phasetrans]

I never saw this when it was initially posted, so this is my first time seeing it. I definitely get the overvoltage protection when using a 7660 charge pump, but what made you want to add the soft start? I can't imagine there are a lot of pedal circuits that would be sensitive enough to need or benefit from a soft start.

Looks great by the way

The thought was to reduce the current to be easier on the input caps and (unknown) upstream power supply. This bit of the circuit comes to steady state well before the charge pump does. (100u / 1u at all 3 7660 cap locations).

You pay (in switching ripple) for any series resistance before the charge pump, and the PNP gives you the saturation behavior for free, so why not use it.

I figure it's pennies of parts to provide care and feeding of the single biggest BOM line item, the 7660H.