Dead simple switched regulator with constant current 0-10A

[harvie]

Hello,
i am trying to design constant current step-down SMPS with following parameters

+ 0-10A (maybe 20A) set by pot
+ very cheap
+ minimal number of components (eg: LM358, N-MOSFET, zener diode, few passives)
+ very inprecise (i don't care about +-0.5A ripple)
+ voltages up to 18V (or determined by used opamp)
+ should survive short circuit (= should limit current regardless of load)
+ should be cheap4ss circuit for electrolysis experiments and similar, controlling resistive wire heater, eventually charging old automotive batteries, nothing serious
+ think of this as of joule thief kind of circuit, JUST NOTHING SERIOUS. really.

I am playing with this simplified circuit i've created in falstad simulator, however i don't know how to properly calculate values:



How i suppose it to work:

On the upper-right part of schematic you can see traditional step-down topology with N-channel MOSFET and terminals for attaching load probably with some ripple rejection capacitors which are missing at image.

Step-down circuit is in series with shunt resistor with parallel capacitor which should smooth out the voltage readings on shunt and keep whole thing running at frequency reasonable for opamps and MOSFET. Shunt is selected to have reasonably low resistance, so there will not be much of loss.

Shunt voltage is then compared using comparator with reference voltage made using zener diode with proper current limiting resistor or 7805, etc... then there is resistive divider which lowers the reference voltage to something more similar to what we can sense on shunt.

Comparator is followed by schmitt trigger to ensure proper saturation of MOSFET which follows. and reduce unwanted oscillations.

So when cicuit is powered on, the shunt voltage rises over reference voltage (as shunt capacitor charges), which makes it to close MOSFET, therefore shunt voltage drops (as shunt capacitor discharges), which makes MOSFET open and everything repeats again. I hope this will reach some kind of balanced state which will hopefully provide +- constant current with possibility of seting it using potentiometer and ampermeter. Current doesn't even have to be predictable, but it should have +- stable RMS as long as you don't turn the pot.

Do you think that this can work? How should i choose proper inductor? Can you help me with some basic mathematics behind this circuit?

PS: i know that this can be done with specialized ICs like LM2596 using external error amplifiers used to fool CV feedback inputs. i know that there are cheap4ss linear solutions using LM317. and this is NOT my homework for school. i am just trying to rethink that concept. i want to find some poor man's high current regulator design. I saw simple 20A current regulators on the web, however they are not suitable for short circuits as there is no choke. especially "HHO fringe scientists" are using these kind of circuits.

 

[BradtheRad]

It so happens that I too have experimented with Falstad's simulator, including switched-coil converters. I fashioned a hysteresis-driven buck converter which works similar to the way you describe.

First, however, since you ask about component values, go to this thread and see my posts. They contain links to my simple tutorial circuits of buck, boost, and buck-boost converters. Press a switch to cycle the converter. Watch the scope traces.

https://www.edaboard.com/threads/268178/

 

[schmitt trigger]

A 20 amp, 0.1H inductor will be huge, steel core weighting like 10 to 20 kilograms.
I've a 5 amp 0.1H inductor at home, and weighs 5.5 kilos.
Also it will oscillate at only a few hundred hertz and the inductor whine will drive you crazy.


In an SMPS converter, the inductance must be at least 1000 times smaller.

 

[harvie]

I know... I've found that in my favorite shops with electronic parts they have only much smaller toroid SMPS chokes... In fact they have only 3 kinds of 10A chokes:

+ 0,15mH 10A 27mOhm iron dust core
+ 100uH 10A 44mOhm iron dust core
+ 68uH 10A 42mOhm iron dust core

so i want to use inductors of +- these values. let's take this as constant. now i don't know how to select proper frequency of step-down and how to tune my circuit to stay at least at such frequency, so inductor core will not be fully saturated (in resistive mode) as i guess that would mean short-circuit of power supply.

Also that's why i am asking for help... the inductance shown in schematic is not what i want to use in real thing. I want to figure out how to set frequency of this self-oscillating circuit and choose some small, but proper choke. I don't even know if i want ferrite or iron dust one. But iron dust is probably cheaper and easy to salvage from dead ATX PSUs, so i would preffer to stick with that.

 

[BradtheRad]

For high current needs, you can interleave two or more converters. Share current between them. Control them with staggered clock pulses.

 

[harvie]

Do you think that this is good idea for "just" 10A? BTW can't i just put few identical inductors in parallel using single switch? Probably not... But that's not exactly issue yet, as 10A inductors are available...

First i have to find a way to tune the circuit to proper frequency... Maybe i can start with 5A or 2A version and then upgrade circuit to higher currents in next design iterations. Maybe even 5A would be quite satisfactory for most of my planned use cases... Primary goal is to keep everything as simple and cheap as possible.

 

[BradtheRad]

Here is an example of interleaved buck converters.
The output is 20A. 10V.

It uses 100 uH inductors.



The .2 ohm resistor signifies 'invisble' parasitic resistance. The amount is just a guess. A high value will reduce available amperage.

The clock signals are staggered. It will require some effort to make their duty cycle variable, with one adjustment.

 

[harvie]

Thanks, but actually i am not interested in such complex circuits... I will more likely sacrifice some amperage if it can't be done with single switch and single 10A inductor that i've suggested. I am looking for SIMPLE(st) possible solution to buck current limiting. Stating that i am not looking for real PWM circuit. These are well documented and it's easy to find design.

I just want to know how to tune my comparator "oscillator" to frequency suitable for the chokes i've listed. And how to calculate this frequency.

 

[BradtheRad]

You were on the right track with your design. I think you would have gotten it before much more time went by.

Here is my buck converter driven by hysteresis, using an op amp.

Single coil, 100 uH.

Coil current is 10A or less. You can increase or decrease this, based on how you adjust the hysteresis resistor.



The op amp drives an analog switch. At the moment it is closed.

The sense resistor is strategically placed. It partially senses load voltage, and partially current through the coil.

Notice the amount of current going back and forth through the capacitor.

 

[harvie]

Well this looks pretty much as what i've wanted. However i still have few questions:

1.) Is it OK to use low side switching as in my schematic? I think it's more clear to me to drive N-MOSFET in low side. also low-side shunt is more easy to understand for me...

2.) Am i right that shunt resistor should be at least 5W?

3.) What determines frequency of your circuit? is there some kind of RLC equation involved? This is one thing i still could not understand...

4.) Do i need to use resistor network for hysteresis? Maybe i can use "comparator" ic instead of "op-amp". These should have some hysteresis network already built in... am i wrong?

5.) you say that sense resistor partially sense current and partially senses voltage. however it seems to me to be more CC than CV regulator. so what is point of this?

 

[BradtheRad]

Well this looks pretty much as what i've wanted. However i still have few questions:

1.) Is it OK to use low side switching as in my schematic? I think it's more clear to me to drive N-MOSFET in low side. also low-side shunt is more easy to understand for me...

Yes.

2.) Am i right that shunt resistor should be at least 5W?

Yes, for .05 ohm carrying 10A.

The lower the ohm value, the less power is dissipated as heat. I chose just a large enough value to generate changes of a few tenths of a volt. This makes it easy to read on the simulator, and is easily detected by an op amp.

3.) What determines frequency of your circuit? is there some kind of RLC equation involved? This is one thing i still could not understand...

It's related to the L/R time constant. A small Henry value has quicker action. The ampere level changes faster. Hence the waveform rises and drops faster.

To get greater amps, you must leave the switching device On for a longer time. This slows the frequency.

4.) Do i need to use resistor network for hysteresis? Maybe i can use "comparator" ic instead of "op-amp". These should have some hysteresis network already built in... am i wrong?

A schmitt trigger is okay.

However an op amp (or comparator) allows you to change the hysteresis range. You can explore how to operate the coil in continuous conduction mode, discontinuous mode, etc.

I have not seen a schmitt trigger available in Falstad's. It must be an update.

5.) you say that sense resistor partially sense current and partially senses voltage. however it seems to me to be more CC than CV regulator. so what is point of this?

Just a change from the usual. I played with different positions for the sense resistor. I was looking for a way to combine both (a) the switching function and (b) voltage regulation. My aim was to economize, same as you.

By placing the sense resistor after the coil, I obtained voltage regulation. It made things easy to adjust. I could see relationships between sense voltage, and hysteresis range, operating frequency, ripple on the load, etc.

By the way, my simple design is untested. It looks as though it should work with real components, although there's no guarantee.

 

[harvie]

The lower the ohm value, the less power is dissipated as heat. I chose just a large enough value to generate changes of a few tenths of a volt. This makes it easy to read on the simulator, and is easily detected by an op amp.

So when i'll use .005 ohm or less for maximum efficiency will the real-life (non simulated) comparator be still able to compare such low voltages with enough precision? I am not sure about resolution of typical comparators, however op-amps are told to have "infinite gain", so it mmight work hypotheticaly.

It's related to the L/R time constant. A small Henry value has quicker action. The ampere level changes faster. Hence the waveform rises and drops faster.

According to what you say, the frequency should double when you will half the inductor. So that means this circuit will automatically adapt itself for smaller inductances without going to saturation?

A schmitt trigger is okay.

However an op amp (or comparator) allows you to change the hysteresis range. You can explore how to operate the coil in continuous conduction mode, discontinuous mode, etc.

So the hysteresis range affects frequency or output voltage/current?

Just a change from the usual. I played with different positions for the sense resistor. I was looking for a way to combine both (a) the switching function and (b) voltage regulation. My aim was to economize, same as you.

However it seems to me that this way one can't be really sure if it does CC or CV regulation... I can imagine that in your chematic you can adjust CC using pot and CV using hysteresis range, but i am quite doubtful about it...

- - - Updated - - -

BTW... what about using inductor as shunt? It has some declared resistance in saturation mode which means when it's voltage drop goes down to some small value at 10A we can tell it's saturated and we should simply close the switch. However i don't know what will happen at smaller currents.

 

[BradtheRad]

So when i'll use .005 ohm or less for maximum efficiency will the real-life (non simulated) comparator be still able to compare such low voltages with enough precision?

Yes. I have used a few inches of wire to act as a sense resistor. Just a few milli-ohms, and it generated a few tens of mV. An op amp was able to detect it.

According to what you say, the frequency should double when you will half the inductor. So that means this circuit will automatically adapt itself for smaller inductances without going to saturation?

Correct, except the part about avoiding saturation. A real circuit may need current-limiting safeguards. To sense the flux field intensity will require a different kind of detector.

So the hysteresis range affects frequency or output voltage/current?

I've seen how several adjustments all influence each other. It gets tricky, especially when you start wanting versatility, varying supply voltage, varying loads, etc.

In some cases the duty cycle tries to rise until oscillations stop altogether. The switching device is left on, exposing the load to full supply voltage. This disaster must be prevented, which requires more work and effort.

However it seems to me that this way one can't be really sure if it does CC or CV regulation... I can imagine that in your chematic you can adjust CC using pot and CV using hysteresis range, but i am quite doubtful about it...

I wish I could say my schematic was tested rigorously, well-researched, proven in service, etc. Instead it's just an experiment. Successful only in simulation. Nevertheless Falstad's is ideal for this sort of thing, as you know. It is a great advantage to watch electrons (or current bundles) moving through wires. And to see immediate results after you change anything.

BTW... what about using inductor as shunt? It has some declared resistance in saturation mode which means when it's voltage drop goes down to some small value at 10A we can tell it's saturated and we should simply close the switch. However i don't know what will happen at smaller currents.

This is something you can try, of course. Improvements are possible. Let your ingenuity be your guide.

 

[harvie]

Correct, except the part about avoiding saturation. A real circuit may need current-limiting safeguards. To sense the flux field intensity will require a different kind of detector.

I am not talking about magical saturation prevention. I've meant it like when i manualy tune the circuit for 100uH 10A inductor to not go tu saturation and then switch to 50uH 10A inductor it will remain tuned. doesn't it?

Anyway... I plan using 10A polyfuse as "safeguard"...

I wish I could say my schematic was tested rigorously, well-researched, proven in service, etc.

I am going to give it intense test as soon as i will understand how to desing proper inductor and keep it away from saturation...

How do i calculate the proper frequency for given inductor and current to stay out of saturation?
Does the duty cycle matter unless it's 100%?
What else i need to know to prevent saturation?

 

[BradtheRad]

I am not talking about magical saturation prevention. I've meant it like when i manualy tune the circuit for 100uH 10A inductor to not go tu saturation and then switch to 50uH 10A inductor it will remain tuned. doesn't it?

Anyway... I plan using 10A polyfuse as "safeguard"...
...
...
I am going to give it intense test as soon as i will understand how to desing proper inductor and keep it away from saturation...

How do i calculate the proper frequency for given inductor and current to stay out of saturation?
Does the duty cycle matter unless it's 100%?
What else i need to know to prevent saturation?

The range of hysteresis travel is not dependent on the inductor. You can use any inductor value. The frequency will adjust automatically.

Here is my experimental buck converter where the op amp drives everything.



Falstad's simulator can export a link of a circuit.
Click the link below, and it will:

(1) Open the falstad.com/circuit website,
(2) Load my schematic (above),
(3) Run it on your computer.

https://tinyurl.com/pa5c9sy

Three 'tap wires' are available for voltage regulation.

Another load can be switched in.

Hysteresis feedback is a potentiometer adjustment.