Simple Boost Converter for duck_arse -- Shared w/ All

[Transmogrifox]

duck_arse thought I should start a thread on this.  This was sort of a spin-off from a thread about 5V USB to 9V power converter.

It's also a good option for those wanting to derive 24V or 30V rails from 9-12V wall-warts.  I originally developed the design to supply 320V DC plate voltage to a tube stompbox from a 9V wall wart supply (built and actually working).  So it is very flexible in the range of things you can do with it.  Main advantage is it is really simple.

Here is the schematic for a 5V to 9V converter aimed at powering a single FX unit from a USB connection.  Of course if you wanted to do multiple pedals, then the power goes up and the calculations need to be re-visited.
http://www.cackleberrypines.net/transmogrifox/BoostConverter/5V_USB_to_9V_FX_Converter.PNG


Here is an example calculation for the schematic above.  Please note efficiency is pretty bad at light loads like this.  For heavier loads I have seen efficiencies up to 85% -- I think I saw 90% in certain loading conditions on my 12V to 25V converter.
http://www.cackleberrypines.net/transmogrifox/BoostConverter/SMPS_Boost_Equations.pdf

For those who don't want to follow the link, here's the basic run-down.

You need some known variables defined:
Vi == Nominal supply voltage input
Vo == Output Voltage + Rectifier Diode Drop
Fsw == Desired Switching Frequency (Aim for something like 200 kHz)
T == 1/Fsw <==Switching Period
n == estimated efficiency.  This is a guessing game without going into more detailed analysis
P == Output power, including power needed to drive the feedback node
Vbe == Base-Emitter turn on voltage for current sense transistor

It's a 2-step process. First find value of Lboost:

L= Vo^2*{T/[1+Vo/(n*Vi]}/(2*P)

Next find the necessary value for the current sense resistor:

Rsns = L*Vbe/{Vi*T-Vi*[T/(1 + Vo/(n*Vi))]}

The actual switching frequency depends strongly upon the value of R23 in the schematic linked.  To get close to the calculated switching frequency to have to make it as large as you can make it and still start up.  For reliability, since you don't want to teater on the edge of wondering whether the thing will start up, give it some healthy margin, and then compensate by increasing switching frequency, or accept the fact that it will be slower than calculated.

The main thing with an SMPS used for audio circuits is to keep it well away from the audio band (>35 kHz).  It may seem like calculating for 200 kHz at max load you are in "high cotton" but keep in mind with this particular topology that switching frequency drops with light loads.

Even though this was designed to be a USB power converter you could also use it in a stompbox to milk every last bit of juice out of your battery.  For example, when battery discharges down to 7 Volts, this circuit can give you 9V out for as long as there is enough energy to be had.

It won't start up until batt voltage bleeds down low enough to turn it on.

[duck_arse]

that guy, he's always causing trouble.

M2 looks new, what does it do?

[Transmogrifox]

M2 is a crude undervoltage lock-out.  It also works as a bit of a soft-start, which in this case, is not the intended function.

Either way it's there to make sure the oscillator has started up before connecting in the load.  This converter can't start up into a capacitor that is larger than Chold if there is a resistor in series with it (like the 47 ohm).

Since most FX often have a diode or some R in series with the input as well as filter caps much much larger than 1uF the oscillator won't start up if those loads are connected at power-on time.

If this is to be a generic circuit to be connected to any FX (like if it was USB to 9V spider) then you need a huge Chold that is guaranteed to be larger than any load capacitance, or you use an output current limiter so it doesn't begin to charge output capacitance until after the oscillator has started up.

Once it has started up then it continues to run plugging out its max power until it gets the load charged up to the regulation point.

If you're designing this into a specific circuit then you can size Chold appropriately and do away with the UVLO.

This particular circuit is an example application tailored to powering a single effect with current <10mA or so.   To make a generic USB to 9V spider the power output would need to be increased -- and then you probably want a more beefy FET for primary switch.