What's the best way to generate reference voltages?

[bentomo]

So I'm trying to generate some reference voltages for some TFT LCD scan drivers. I'm having some trouble figuring out what the best approach is.

I need:
 +15v 240uA output
 -10v 260uA output

Driven by 3.3-5v input.

Charge pumps don't seem to generally have the range I need, at least looking at what digikey has in stock.

So my next option would be to use some boost converters. But what do I do about keeping them out of discontinuous mode? Should I just add a load resistor? This seems like it will waste a lot of energy since a lot of bucks operate at around 50 mA minimum or more. My application is going to be battery powered as well so efficiency is important.

I feel like I'm missing something since +-15v and +-10v were a common requirement for opamps in my EE1 and EE2 classes.

Is there a type of power component out there that's designed for this that I'm missing?

Thanks!

[bktemp]

You could use ICs designed for that purpose like TPS65150. It generates all voltages required for a typical medium sized TFT LCD.
Another solution is using a cheap boost converter for generating +15V and adding a charge pump for a negativ voltage (about -14V). Then use a linear voltage regulator or zener diode for generating the reguired negative voltage level.
There is nothing wrong with operating the boost converter in discontinous or even PFM mode at such low current levels.

[bentomo]

Wow that thing is awesome! I ignored those all in one IC's because I thought they were generally more expensive and for larger monitors. Mine is just a 5.6" lcd but it looks like that chip will do the business and is still pretty cheap. It looks like I'll still need a dedicated constant current source for the backlight but I've got a decent one of those picked out already.

Thanks!

[bentomo]

So a little update on this, I got everything working up to a certain point.

I've got the LED power working using the AP7525 and the backlight is coming on, so that's good.

However, the TPS65150 doesn't seem to be fully functional yet. The 5v boost regulator comes on, but the VGL charge pump only goes down to -5v when it should be -10, which prevents the VGH and Vcom from coming on since it has to wait for VGL to be within its target voltage first before the power on sequence can continue.

I confirmed it was a fault issue by disabling the fault timer and the voltages even out at 4.998 on the 5v boost and -4.67 on the negative charge pump, and the rest settle at 0v. With the fault timer ENABLED all the voltages shut off at the 100ms programmed time. So somehow the IC knows that it's not at the correct voltage yet, but doesn't increase the voltage. That's with and without a load applied.

My board soldering looks good with the meter and under a microscope, I'm fairly certain I can rule that out. I've looked over my schematic at least a dozen times.

There was 1 mishap at the beginning where I used the pin numbers for the sot23-6 footprint on the QFN6 footprint.  A bonehead rookie mistake and I got some nice magic smoke out of it. Fortunately, you can just barely squeeze a sot23-6 onto a 2x2mm QFN6.

So I'm thinking it's possible that I fried the TPS65150, (because I decided to place everything instead of testing the circuits individually  ) but I'm still doubtful of that since most of the current would have been going through the led boost converter and wouldn't affect the TPS65150. I couldn't get much out of the AP7525 datasheet's block diagram to tell what happened, best I can tell is I wired Vcc and GND backwards, so I figure I was shorting out through the body diodes of the internal mosfets. Still, I ordered a couple backups and will know for sure next week.

So I'm not sure what else to check. It seems odd to me that the chip is "half working" getting to -5v but not -10v, but still functional enough to fault out. So maybe I fried some of the control logic but the fault logic is separate. 

Thoughts?

Another little strange side note. The lcd data sheet says that Ivgl should be 250uA. So from a -10v source I figured about 38kohms, I hooked up a resistor instead of the LCD so I won't risk frying that. I measured VGL to GND on the LCD and I don't get so much as a bump on the meter. It's as if there's no connection anywhere so I'm not sure how it would draw any current let alone 250uA. Maybe it's just specifying leakage during operation since it's just a reference voltage?