Single supply voltage conversion help

[smugtronix]

Hey all:
I'm working on adapting a simple FET boost from one of my favorite guitar amplifiers into stompbox form. While I have no problems with the actual FET booster, I'm at a loss about designing an elegant power supply.  The original circuit ran on +/- 15VDC from a linear transformer. Most guitar players use a standard 9VDC plugpack. A linear transformer is obviously out of the question, as I am trying to keep this box as small as possible. While I know how to create +/- 15V from a single power supply, the only ways I can think of doing it are pretty ugly and expensive, requiring at least two ICs.
So far, I've come up with 2 pretty stupid ideas.
Option 1:
18v plugpack-> 78L15-> LT1054 to do the inversion:
This is the simplest way I can think of doing it, but it's messy. The LT1054 is a pretty expensive chip (compared to the ICL7660 I normally use), and you would lose the ability to use a common daisy chain power supply, which I would prefer.

Option 2:
9V plugpack-> 7805-> Murata 5v to +/- 15v DC-DC converter.
This is not only an obnoxiously expensive solution, it's extremely inefficient (estimates say the converter alone will draw 100mA, which is definitely overkill for a single FET circuit!)

I'm at a bit of a loss. Are there any single-chip solutions that can solve my problem? If not, what would be some simple ways to get the +/- 15V? I don't need a ton of current, or insane efficiency, just something practical and simple. Thanks!

[Psi]

When you say option 2 is "obnoxiously expensive" how much is that exactly?

PCBmount power modules are usually in the range of $5-20 depending on specs and i wouldn't call that obnoxiously expensive

A quick look at digikey found this one
http://www.digikey.com/product-detail/en/RS3-1215D/945-1565-5-ND/2321411

9-18V input
+/- 15V output @ 100mA

[smugtronix]

I was looking to spend in the neighborhood of $5-10 USD. I actually have to do some measurements on the actual amplifier to find out how much current the thing is actually drawing. It will be significantly less than 100mA.

[SeanB]

9 VAC plugpack giving Ac into the unit, and 2 half wave rectifiers and 2 4700 uF 25V capacitors inside, then 2 7808 regulators to give 8V rails. Otherwise use the unregulated rails with a LC or RC filter to reduce noise further, this will give a 12V rail that will do around 50mA with little ripple.

[duskglow]

The solution that comes to mind for me, which could be entirely wrong, is a 9V AC input, a charge pump, an inverting charge pump, a 7815 and a 7915 regulator.   Dave had some recent videos on how to do that.  Not going to get much current but it sounds like you don't need much current.

[smugtronix]

After thinking about it, I may just use an ICL7660S in a doubler/inverter configuration to get 27 volts total swing. I have a feeling that'll be close enough, and if it isn't, I could bump the input supply up to 12v, and get 36v total swing. I'm just not sure how important the +/- 15 supply actually is. I'm already using trimpots for both Rd and Rs (to compensate for JFET inconsistencies), so it shouldn't be too much extra work to get the thing exactly where I want it. 

[Psi]

What about something like this for US$3.76
http://www.hobbyking.com/hobbyking/store/__26910__arduino_lm2577_dc_dc_adjustable_step_up_power_converter_module.html

Set it to 30V output from your 9V input and then use a virtual ground to get +/- 15V (either resistor divider or an actual chip)

That would be quite cheap

[Paul Price]

Chances are the +15 -15 power supplies are not required to get the circuit to work the way you want.

Post a schematic.

 It might just be that your circuit would do just fine with + and - 4.5V.

[leafi]

If you are going to use a switching supply I would love to hear if the ripple becomes audible when the signal is amplified. The Fet will only have so much of a ripple rejection to it. I would love to hear what you solution is.