negative bias generator

[ealex]

Hello

I need to generate ~ -28V at 5mA for a LCD bias voltage, from a 5V supply with some components that i have at hand ( i can't place any order right now + it's a one-off )

I've thought of a simple royer circuit using some ring cores i had available:
http://www.kaschke.de/fileadmin/user_upload/documents/datenblaetter/Ringe/R30-5.3.pdf
http://www.kaschke.de/fileadmin/user_upload/documents/datenblaetter/Materialien/MnZn-Ferrit/K2006.pdf

the inductor values are not optimized - i can reduce L1 and L2 to 5 turns without saturating the core.
Also L6 - i'll pick up something that i have available ( 470uH for example )
Q1 and Q2 form the royer oscillator. C1 sets the oscillator frequency - i'll tune it so it oscillates at ~ 100kHz
Q3 is used as an enable signal - there must be a delay between 5V and -28V otherwise the LCD drivers go crazy.

If i'm careful i can fit almost all components inside the core ( hot-glue everything in place once it's finished and forget about it )

C3 acts as a inrush current limit so the circuit will not overload the 5V rail on start-up and will limit the current if the LCD drivers decide to go crazy.


I don't think a charge pump / voltage multiplier will work with this voltage ratio ( almost 6:1 )

Are there any other solutions ? Maybe a 1 transistor oscillator ?
Or a 1 transistor fly-back - will it work with a simple ring core ? i have some E32 and E42 cores available but they are too big.

In the image:
green - output voltage
red - enable signal
yellow - input current

I've attached the LTspice circuit / simulation.

[Marco]

Why wouldn't a invering multiplier work? Just use a higher multiplication and use a (makeshift) regulator to get it to the required voltage.

[Howardlong]

LT1617?

http://cds.linear.com/docs/en/datasheet/16171f.pdf

[dannyf]

If you have a pin or two to spare, a mcu can do that.

Otherwise, try a 555.

[paulie]

It's true. A PWM pin with transistor, inductor, diodes, and cap. 25 cents tops. I managed to generate 400v that way for a geiger counter tube.

(Dannyf is sooooo smart. I wanna be like him when I grow up.)

[ealex]

I've thought of the PWM / inductor, etc but the LCD will be driven by an fpga - the pwm control might not be that easy.

the simple boot configuration will not work, i need a negative voltage in respect to system ground.
in the spice simulation i've used a positive output voltage just for convenience - the output is isolated

i've seen the polarity inverting buck configuration but i don't have a P-FET available and i don't think i can source one quickly ( i can't order anything right now and i can't find much more than basic stuff around here )

i've simulated a voltage multiplier - see the attached image and ltspice files. it needs 14 diodes and capacitors.
the input signal is a 100kHz 5V square wave - i can make this from the fpga + a transistor driver.
the first image shows the output voltage with no load
the second with a 2.8mA load - the voltage drops to 22V with 1V ripple  => that's why i did not want to use a multiplier.

[dannyf]

It is very usable as a proof of concept and/or quick hacks, but it is not suitable for every situation either.

Change the diodes' polarity, or ground the other end of the bridge.

[ealex]

hello
update: managed to dig up a MC34063 from the scrap box
i've built one of the basic applications from the data sheet - it works better than a simple unregulated royer. It uses ~ 35mA with no load, 5V in and -28V out.

i've also managed to find a logic level n-mos ripped from a mother board -> made a simple ON/OFF swithc using it. With 3V on the gate it conducts pretty well.