Small MOSFET power buffer

[Sniper1]

So here is some context:
I need some negative voltage in order to use a 5V LCD with a 3v3 MCU (STM32 ) and after a bit of research i found out that charge pumps are a easy way to do this and for the LCD i would need 1-0.5mA at -1V or a bit less , this should be ok/doable with a GPIO pin. BUT i also have some op amp circuitry and filters and it would be nice to be able to get 0V there too but that circuit needs like 13mA according to LTSPICE.
So i tried to make a CMOS buffer in spice and failed hard , multiple op amps in parallel seem to work , so does a CMOS 555.
repo: https://github.com/immortal-sniper1/Progrmable_Active_Loads/tree/master/Small_load_V1/simulations

So can anyone give me a hint of what i am doing wrong?
I was thinking of using some TO-92 like:
https://eu.mouser.com/ProductDetail/Microchip-Technology-Atmel/VP2106N3-G?qs=NZF3WWV0eEGNFUSyHi4JQQ%3D%3D

https://eu.mouser.com/ProductDetail/onsemi-Fairchild/2N7000BU?qs=k2x4EL1%2FKj6oeXMHAfSm5A%3D%3D

As for the current i saw in SPICE the peaks are bout 280mA but the RMS is about 60mA and this makes me wander if op amps in series will realistically work.
NOTE: since the PWM will come from a STM32 i can vary the duty and f but from what i read 50% is best and if f is great it is better as in lower cap values and since i can go as fast as 1MHz why not?
Also i have a 12V boost so if it is the only way i can also borrow from that switching node and get -12V but then ill need some negtive regulation back to -1 -2 V...
Am i missing something?
PS is there any other simple SMPS inverting design that i can DIY with 1-2 GPIO in order to get a 15-20mA negative rail?

[Shay]

So here is some context:
I need some negative voltage in order to use a 5V LCD with a 3v3 MCU (STM32 ) and after a bit of research i found out that charge pumps are a easy way to do this and for the LCD i would need 1-0.5mA at -1V or a bit less , this should be ok/doable with a GPIO pin. BUT i also have some op amp circuitry and filters and it would be nice to be able to get 0V there too but that circuit needs like 13mA according to LTSPICE.
So i tried to make a CMOS buffer in spice and failed hard , multiple op amps in parallel seem to work , so does a CMOS 555.
repo: https://github.com/immortal-sniper1/Progrmable_Active_Loads/tree/master/Small_load_V1/simulations

So can anyone give me a hint of what i am doing wrong?
I was thinking of using some TO-92 like:
https://eu.mouser.com/ProductDetail/Microchip-Technology-Atmel/VP2106N3-G?qs=NZF3WWV0eEGNFUSyHi4JQQ%3D%3D

https://eu.mouser.com/ProductDetail/onsemi-Fairchild/2N7000BU?qs=k2x4EL1%2FKj6oeXMHAfSm5A%3D%3D

As for the current i saw in SPICE the peaks are bout 280mA but the RMS is about 60mA and this makes me wander if op amps in series will realistically work.
NOTE: since the PWM will come from a STM32 i can vary the duty and f but from what i read 50% is best and if f is great it is better as in lower cap values and since i can go as fast as 1MHz why not?
Also i have a 12V boost so if it is the only way i can also borrow from that switching node and get -12V but then ill need some negtive regulation back to -1 -2 V...
Am i missing something?
PS is there any other simple SMPS inverting design that i can DIY with 1-2 GPIO in order to get a 15-20mA negative rail?

There are plenty of ways to generate negative rail, using somthing like ICL7660 or MAX1044, or isolated dc-dc converter:
https://he.aliexpress.com/item/32840613546.html?spm=a2g0o.productlist.main.1.1a3eNWhgNWhgUf&algo_pvid=8b6aaad7-833b-40b3-950c-c43d84eb6b78&algo_exp_id=8b6aaad7-833b-40b3-950c-c43d84eb6b78-0&pdp_ext_f=%7B%22sku_id%22%3A%2265123770336%22%7D&pdp_npi=2%40dis%21ILS%212.62%212.1%21%21%21%21%21%402100b88516695944540848124d0749%2165123770336%21sea&curPageLogUid=26lT8rWjUOWN

As long as you have a uController that can generate a square wave, you can make a charge pump:


the frequency of the square wave shall be as fast as possible in order to reduce ripple at the output. I recommend playing with the capacitors value.
The average current is about 50mA. each GPIO pin should be able to generate ~40mA max. so either utilize 2 charge pumps and connect the outputs together, or parallel the GPIO pins.

[Sniper1]

A cording to this:
https://eu.mouser.com/datasheet/2/389/dm00440565-1799229.pdf 
It is 16mA per pin then again i am not sure if it is constant or peak. Not sure how i can turn 2-3 GPIO on at the same time on a stm32
but also if i reduce the caps current and negative voltage will fall and since i need only -1V that may be a safer route.
BTW i also have 1-2 free op amp would puting them in parallel end up with better overall?
https://www.ti.com/lit/ds/symlink/lm2904lv.pdf?HQS=dis-mous-null-mousermode-dsf-pf-null-wwe&ts=1669635009383&ref_url=https%253A%252F%252Fro.mouser.com%252F
it seems to have 40mA per channel so maybe use 2 like a buffer ?
Ill simulate a bit more anyway .
THX for the advice

[mawyatt]

The average current is about 50mA. each GPIO pin should be able to generate ~40mA max. so either utilize 2 charge pumps and connect the outputs together, or parallel the GPIO pins.

Or one could just use a hex inverter like 74ACT04 (74HC04) as a buffer by paralleling all the inverters.

Best

[DavidAlfa]

Use a 74hc14, use first gate as free-running oscillator, rest of the pins as current buffers.
This will make about 17V, adjust the doubling stages to the required voltage. Each stage adds ~4.3V.

[Sniper1]

THX
voltage is not that big og a concearn if it is under -0.7V but current is that is why i am useing only 1 stage but yea 74 series chips are a good idea + i can make all 6 be buffers and adjust the PWM f and duty with my STM32