Simple Voltage Regulator without ICs

Thread Starter

D_racon

Joined Oct 4, 2021
11
I wanted to make a voltage regulator for making a higher voltage 5v, and found a old design that worked using a schotky diode 2 resistors and to transistor. I found that such a cirquit despite not seeming to efficient would be fine to generate a reference voltage. but I was thinking about things like potential stress on the diode or fluctuations in the voltage under higher current(with stronger transistors). so I taught about making a voltage regulator that uses that as reference voltage and then uses that to control it's voltage. originally I was thinking about using something like a schmit trigger or a 555 timer in a feedback loop comparing the voltage to the reference voltage with a slight offset so it would act like a PWM signal and either be on or off.
however it is late in the evening and I knew I shouldn't go to deep into something like this since there are many good cheap IC's for regulating voltage, also using high frequency switching, and I knew that if I started to read the datasheets I would go further into it and eventually go way to deep in things and spend more effort on it than it would be to go to a closby store and get a regulator IC there. especially since I will most likely eventually still use a off the shelf IC for it since the parts are quite sensitive to over or undervoltage.
and I eventually came with the idea that just greatly amplifying the signals of the feedback loop might have a similar effect of keeping the transistor/mosfet responsible for driving the output in a high efficiency state.
VrefbasedVoltageRegulator.png
on the right bottom is the old reference voltage circuit. the one on the top left is the first design I came with, however that one was mostly meant to use a schmit trigger or such a 555 contraption instead of a simple transistor there on the top left but I didn't draw it like that because perhaps normal transistors also could keep it efficient enough and a schmit trigger might have to much voltage fluctuation and is a lot of work to build by yourself.
the V2 on the bottom left is the second design I came with it uses a PNP transistor, a NPN transistor and a PNP mosfet. when the output voltage is to low the PNP transistor enables because the trigger is pulled low, then the NPN transistor is also triggered because the trigger of that one is put high. the NPN transistor makes a NOT gate with the transistor above it, so when it is on the voltage above it gets to low, so the PNP transistor enables and the V OUTPUT rises., it is capacitor stabilized. and it should keep the voltage at VOUT at the voltage the 5v reference generates.
all would have a shared ground.
But would it make any sense to use a cirquit like this, would it have a benefit like stability or keeping the mosfet in stable states. or would it really make no sense to that and would it be better to just replace the first transistor of the old REF circuit with a more powerful transistor or mosfet/

and would something like a 555 or schmit make V1 or V2 more efficient, this is especially the case if it isn't yet more efficient than the old REF circuit.
 

Papabravo

Joined Feb 24, 2006
21,159
I'm not entirely clear on your ultimate intention. You said:
I wanted to make a voltage regulator for making a higher voltage 5v

I take this to mean that you want a voltage regulator that will take 5V to a higher voltage. Is that correct?
 

Thread Starter

D_racon

Joined Oct 4, 2021
11
I'm not entirely clear on your ultimate intention. You said:
I wanted to make a voltage regulator for making a higher voltage 5v

I take this to mean that you want a voltage regulator that will take 5V to a higher voltage. Is that correct?
no sorry if I was unclear, I mean making a higher voltage 5v as in making a higher voltage into 5v, so the input is a voltage above 5v, lithium cells 2s for simplicity(so 7.4V or 8.4vmax to 6vmin,) the output target is 5v.
so the input voltage is high(6V to 8.4V) and the output voltage is low(5V).
even thaugh the current case is mostly just like that for in a micro RC car. however it doesn't need to stay speciffic to those voltages since to me the principle and main idea is the most important. kind of just to get more insight into it rather than focussing on the speciffic schenario which gave me the question.
the idea was to make a BEC, and that is super simple, but I didn't just want to know how to make one only by buying a part that regulates it for you and then seeing as reason that it works because it works, so I wanted a more low level version, even thaugh the real project might likely use one of such ready parts/ICs since they just are smaller than what I most likely can make at home and in a rc car as long as my finger size is important so it was mostly due to the question that came into my mind which means the voltage range in answers don't need to be speciffic to the speciffc case I have.

the most important parts are the parts of regulating a high voltage to a lower voltage with minimal parts and preffered without specialized IC's since those tend to already do everything for you.
the other main important thing was to see if those designs I had would even make any sense to use or if I could just better strengthen up the old cirquit I wanted to use as a refference voltage.
 

BobTPH

Joined Jun 5, 2013
8,813
Well, to start with ALL linear regulators have the same efficiency which in Vout / Vin, unless they use significant current for their internal operation, so why are you obsessing about efficiency?

I redrew the circuit to make it more readable, and, lo and behold, it is a regulator circuit IF you substitute a Zener diode for the Schottky diode. It regulates reasonably well at Vz + 0.7V. I used a 4.3V Zener to get about 5V out. Here is a simulation. It shows it operating over inputs of 5 to 20V and with a 100mA load.

1668546095300.png
 
Last edited:

Thread Starter

D_racon

Joined Oct 4, 2021
11
Well, to start with ALL linear regulators have the same efficiency which in Vout / Vin, unless they use significant current for their internal operation, so why are you obsessing about efficiency?

I redrew the circuit to make it more readable, and, lo and behold, it is a regulator circuit, if you substitute a Zener diode for the Schottky diode. It regulates reasonably well at Vz + 0.7V. I used a 4.3V Zener to get about 5V out. Here is a simulation. It shows it operating over inputs of 5 to 20V and with a 100mA load.

View attachment 280786
the main idea behind increasing efficiency was to make it handle more power and generate less heat. the idea was that by adding a second loop which uses the previous as reference voltage that amplifying the feedback from the feedback loop it would behave more like a digital/switching signal, and optionally it would allow to replace a part with a part that actually changes it into switching.
but I guess that also when amplifying it it would still somehow balance itself causing it to be just as inefficient as the others indeed.
btw. what software do you use for those drawings and simulations? what I currently use doesn't allow simulation and it quite old and buggy. as well as that it doesn't support rotating elements in all ways.
 

ronsimpson

Joined Oct 7, 2019
2,989
Many of use use LTSPICE. it is free.
lithium cells 2s for simplicity(so 7.4V or 8.4vmax to 6vmin,) the output target is 5v.
Many linear regulators do not work well if the input and output voltage are too close together. Look for "LDO" linear regulator.
Getting from 6 - 8.4V to 5V is pretty simple. Even at 6in 5out you probably should use a Low Drop Out regulator.
 

Ian0

Joined Aug 7, 2020
9,671
the main idea behind increasing efficiency was to make it handle more power and generate less heat. the idea was that by adding a second loop which uses the previous as reference voltage that amplifying the feedback from the feedback loop it would behave more like a digital/switching signal, and optionally it would allow to replace a part with a part that actually changes it into switching.
but I guess that also when amplifying it it would still somehow balance itself causing it to be just as inefficient as the others indeed.
btw. what software do you use for those drawings and simulations? what I currently use doesn't allow simulation and it quite old and buggy. as well as that it doesn't support rotating elements in all ways.
It's not going to become a switching regulator unless you have an inductor.
 

Papabravo

Joined Feb 24, 2006
21,159
the main idea behind increasing efficiency was to make it handle more power and generate less heat. the idea was that by adding a second loop which uses the previous as reference voltage that amplifying the feedback from the feedback loop it would behave more like a digital/switching signal, and optionally it would allow to replace a part with a part that actually changes it into switching.
but I guess that also when amplifying it it would still somehow balance itself causing it to be just as inefficient as the others indeed.
btw. what software do you use for those drawings and simulations? what I currently use doesn't allow simulation and it quite old and buggy. as well as that it doesn't support rotating elements in all ways.
The problem with your analysis is that it ignores the immutable rule of DC to DC conversion schemes. The rule is that the power out will always be less than the power in. Sometimes it will be a great deal less. In all non-switching applications, the difference between the power out and the power in will be consumed in the regulator itself thus severely limiting the ultimate efficiency.

Switching regulators achieve their gain in efficiency primarily by consuming minimal power in the regulator itself. The LDO (Low dropout) regulator is likely to be the best you can ever hope to achieve with a circuit made from discrete parts.
 

Thread Starter

D_racon

Joined Oct 4, 2021
11
" a 3 terminal regulator " true, that would be much simpler, but that doesn't directly mean it is more fun.
and for the stuf about needing a oscilator or inductor, my idea originally was to add a small offset voltage/hysteresis to that first transistor which compares the voltage to the refference voltage, by replacing it using something like a schmit trigger, so that it would work kind of similar to a schmit trigger oscillator
 

Ian0

Joined Aug 7, 2020
9,671
and for the stuf about needing a oscilator or inductor, my idea originally was to add a small offset voltage/hysteresis to that first transistor which compares the voltage to the refference voltage, by replacing it using something like a schmit trigger, so that it would work kind of similar to a schmit trigger oscillator
That works.
But when exploring a very simple switcher design, you must consider what happens to the operating frequency at low or high power. You don’t want to find out that when the regulator has no load it oscillates at a mad frequency.
The 78S40 and NCP3064 work much the same way, but they have a fixed frequency clock, and a Set-reset latch. The clock pulse resets The latch, switching off the pass transistor. The output voltage is connected to a comparator. When the output voltage drops, the comparator sets the latch and switches on the pass transistor.
The pass transistor switches power through the inductor. (You can’t avoid the inductor).
 

BobTPH

Joined Jun 5, 2013
8,813
" a 3 terminal regulator " true, that would be much simpler, but that doesn't directly mean it is more fun.
and for the stuf about needing a oscilator or inductor, my idea originally was to add a small offset voltage/hysteresis to that first transistor which compares the voltage to the refference voltage, by replacing it using something like a schmit trigger, so that it would work kind of similar to a schmit trigger oscillator
Look up “bang bang buck converter.” It implements the oscillator with a comparator with hysteresis. I used this method to make a constant current led driver to use as a book light with, if my memory serves me, just 6 parts. It was designed ti run off the semi depleted 9V batteries removed from smoke detectors each year.
 

Papabravo

Joined Feb 24, 2006
21,159
Look up “bang bang buck converter.” It implements the oscillator with a comparator with hysteresis. I used this method to make a constant current led driver to use as a book light with, if my memory serves me, just 6 parts. It was designed ti run off the semi depleted 9V batteries removed from smoke detectors each year.
It is true that we have seen hysteretic converters before, but the comparator is definitely an integrated circuit, which seems to obviate it as a contender given the title of this thread.
1668605228269.png
 

rsjsouza

Joined Apr 21, 2014
383
and for the stuf about needing a oscilator or inductor, my idea originally was to add a small offset voltage/hysteresis to that first transistor which compares the voltage to the refference voltage, by replacing it using something like a schmit trigger, so that it would work kind of similar to a schmit trigger oscillator
I agree it is much more fun to learn, design and assemble your own circuits with discrete parts - you are on the right track!

Although the schmitt trigger oscillator could potentially work (a great reference is here, you need a way to store energy while the bypass/switching transistor is in cutoff.

In principle, one of the ways that switching regulators increase efficiency is by reducing the time the main transistor has current flowing through it (the main heat source on a purely linear regulator). That is why both an oscillator and a storage element such as a capacitor or inductor is required to "hold" the output voltage long enough until the next cycle of the oscillator recharges them through the transistor and Vin. Tweak the timing and the amount of energy stored in these elements and you can create a multitude of voltages and current outputs.
 
Top