Constant current circuit for buck converter

[ElectronsFlow]

Hi and thanks for viewing!

In the past I've used a few buck converters off ebay. So far they've done me well, but I'd prefer to have the ability to build my own. That way I'm in control of the quality of components used. I was planning on using the LM2678T-AJT as the heart of my buck converter. Seems to have decent specs with 1.2~37VDC range, 3 amps constant & 5 amps peak. And with the addition of a pot/trimmer it makes a great adjustable buck converter.

I found a simple circuit online I like. However one feature I need is missing, and that's constant current control. I've done my own research but haven't yet found an answer. I believe most CC circuits use a mosfet for current control? naturally would be ideal to have the current adjustable via a timmer/pot.

And if anyone wants to build on the schematic here is the eagle sch file:
LM2678T-ADJ Buck converter.rar

 

[crutschow]

Typically the current is controlled by monitoring the voltage across a resistor in series with the ground lead.
The LM2678 has a minimum control voltage of 1.2V so you would need that much voltage drop across the series resistor at the desired current for control of the current.
If that's too much voltage drop for efficiency considerations then you could add an op amp non-inverting gain circuit to reduce the resistor voltage to say a 100mV or so.

 

[ronsimpson]

and that's constant current control.

Why CC?
Are you driving LEDs that need 2A?
OR
Do you want current limit like in a bench power supply? Can set for 100mA through 2A via a know on the front panel?

 

[ElectronsFlow]

Why CC?
Are you driving LEDs that need 2A?
OR
Do you want current limit like in a bench power supply? Can set for 100mA through 2A via a know on the front panel?

You're exactly right. I need CC for driving leds. And limiting current during testing of new circuits. I do have plans for making a small lab PSU with this buck converter assuming it works decently.

 

[ronsimpson]

This will output a voltage that causes 1A in the LEDs. It will not care what the voltage is.
I usually look for a part with lower FB voltage. But it will work like this.

This does not help your "bench power supply" project.

 

[ElectronsFlow]

This will output a voltage that causes 1A in the LEDs. It will not care what the voltage is.
I usually look for a part with lower FB voltage. But it will work like this.

Thanks for the info, I will in the future find that design useful for driving LEDs. I imagine changing R1 value will change the amount of current flowing through the LEDs? Very useful, but as you already said not useful for a bench psu.

 

[spec]

Hi EF,

Below is an outline circuit for you to experiment with.

The circuit generates a constant current adjustable between 0A and 5A (or up to the maximum current that the SMPS will provide).

The output voltage must be 2.2V or over.

spec

 

[ronsimpson]

There is a series of current monitors. The IAN168 is an example. Here I added a 0.12 ohm resistor on the top side. Note the voltage feedback will remove the voltage drop across the resistor. This new IC looks at the voltage across R4, amplifies by (in this case 10) and puts that across R5. (R5 sets gain) T1 amplifies the current so it can pull up "FB". If the current is too high, the supply changes from voltage regulation to current regulation.

Problem: INA168 dose not work if the output is too low. I think less than 2V. So I so not like this but It is a start. (will not work if you short out the output of the power supply) Just like Mr Spec's circuit.
I would use a LMP8278. It does work down to 0V. (actually -2V) There are several parts that work to 0V or below.
Watch out the power supply for the LMP8278 is +5V and you can use a diode in place of T1.
You might look more for other parts that have an input range from 0 to 40V.

 

[spec]

Hello Sir Ron,

We have been thinking along the same lines again. I too have been considering a high side current monitor chip for other power supplies on ETO and this one.

With a constant voltage/constant current power supply there is always the awkward area when the power supply hands over from constant voltage control to constant current control (frequency stability, voltage accuracy, and current accuracy) .

I suspect with your circuit of post #8 that the constant voltage and constant current functions will interact because there is no dead zone before the constant current function operates- am I right? If not, can you describe how the change over is handled? I would be very interested.

About making a lab power supply with the LM2678 (with continuously variable voltage from 0V to maxV, and current limit from 0A to maxA). The only practical solution, in my opinion that is, is to have a negative stabilized supply line. Once again, getting down to 0V is another problem area with all power supplies.

What are your thoughts about all this?

spec

 

[ronsimpson]

What are your thoughts about all this?

On post #8:
Putting a cap across R5 causes a role off in frequency.
T1 makes a "diode OR gate".
Then we are just at the point where the current limit cuts in; The diode is a little 'soft' and shifts from Voltage to Current mode over a small range. I don't know what you mean by "dead zone" but there is a range where both the (V) and (I) feed back. In audio limiters we called it "soft knee"

If you are driving a small motor, where the current varies several times every turn of the motor; you will see the voltage limited to say 12V when the motor is pulling less power, then as the motor pulls more power the voltage drops to keep the current under the limit. (this all depends on if you slow down the current limit amp)

 

[spec]

On post #8:
Putting a cap across R5 causes a role off in frequency.
T1 makes a "diode OR gate".
Then we are just at the point where the current limit cuts in; The diode is a little 'soft' and shifts from Voltage to Current mode over a small range. I don't know what you mean by "dead zone" but there is a range where both the (V) and (I) feed back. In audio limiters we called it "soft knee"

If you are driving a small motor, where the current varies several times every turn of the motor; you will see the voltage limited to say 12V when the motor is pulling less power, then as the motor pulls more power the voltage drops to keep the current under the limit. (this all depends on if you slow down the current limit amp)

Nice explanation Ron- I now understand what your circuit does and how you have integrated the constant voltage/constant current modes.
I had not realized that the VBE of the BJT introduces a 'dead zone' as I call it.

Having said that, I still suspect that there is an unwanted interaction between the constant current and constant voltage modes.

The solution is the have the dead zone referenced to the output voltage, rather than 0V.

Am I right?

spec

 

[ronsimpson]

Am I right?

Don't know. I think it is common for the I and V modes to have trouble at some point. Part of that is to define what you want to happen when.....

Years ago, I would have used a CT (Current Transformer) to measure peak current in the diode. This current matches the output current as long as the IC is switching. A CT is a non-perfect device. (but then aren't we all)

 

[spec]

Perhaps I have not explained my point clearly.

The transfer function for a bench power supply if for the output to be a constant voltage with all currents up to a set maximum current, at which point the output voltage becomes dependent on the output current. Ultimately the output voltage would be zero for a short circuit.

Assume that the output voltage of a bench power supply is set a certain value, say Vset, which produces an equal output voltage, say Vout, equal to Vset for zero current drain. What is not required is for Vout to gradually drop according to the current load on the bench power supply.

I suspect that the latter is the case, under certain circumstances, in the schematic of post #8.

spec

 

[ronsimpson]

What I was aiming for:
IF Iout<Ilimit then Vout=Vset
IF Iout>Ilimit then reduce Vout

Vset=12V Ilimit=1A Rload = 24 ohms then Vout=12V, Iout=0.5A. (FB=1.2V)
Vset=12V Ilimit=1A Rload = 6 ohms then Vout=6V, Iout=1A. (FB=1.2V via the current censor)

 

[spec]

No doubt that is true Ron (I haven't worked it out), but what happens throughout the whole voltage/current range?

spec

 

[Les Jones]

Hi Ron,
Am I correct in assuming that the current limiting starts when the voltage across R5 is greater than the reference voltage of the LM2678 (1.2 volts) plus about 0.6 volts (Vbe of the 2N2222a) ?

Les.

 

[ronsimpson]

Les,

Yes. Under voltage mode, Vout is divided to get to 1.2V. This we all understand.
The current censor turns current to voltage. When this voltage gets to 1.2+0.6 it pulls up on "FB". This will reduce the duty cycle, causing the voltage loop to not be in control.

I should build one in spice and look at the I & V function.

I have built (approximately this) for high power LEDs. There the current loop normally is on and the voltage loop is only there if the LEDs open so the voltage will not get too high and break something. It is common to build a boost, current feedback supply. If the LEDs open the voltage will want to go up "to infinity and beyond", so I add a voltage limit feedback.

I am certain I have built several supplies where the current limit was much like this. Mostly used for "short circuit" protection or for "inrush limiting" on power up.

In bench supplies there is commonly a voltage error amp and a current error amp that are diode ORed together. I think the TL494 IC does that. (I have a memory of using two 1N914 diodes + resistor on the output of the two amps.)

 

[ElectronsFlow]

Hi EF,

Below is an outline circuit for you to experiment with.

The circuit generates a constant current adjustable between 0A and 5A (or up to the maximum current that the SMPS will provide).

The output voltage must be 2.2V or over.

spec

Woah a hand drawn schematic by Spec himself? you must be on holiday?
I might be overlooking something? But I started drawing up a schematic in eagle to include your CC circuit. Is it correct the opa2129 has only 3 pins used?? I see in your drawing you've got the supply volts and GND with a cap across the opamp. And I'm unfurminluar with the term "ORI" above one of the resistors?

P.S I tried the LVD circuit, something went fizz pop bang, and it was the TLV IC. Haven't ruled out the PCB might have a construction error. Or perhaps I heated up the TLV too much during soldering, Just the other day I was given a Pace MBT SV-4 rework station so no excuses for not soldering SMD components now!!

 

[spec]

Woah a hand drawn schematic by Spec himself? you must be on holiday?

no, just too lazy to put the circuit into EAGLE.

I might be overlooking something? But I started drawing up a schematic in eagle to include your CC circuit. Is it correct the opa2129 has only 3 pins used??

Yes, plus the -V and + V supply pins. The OPA2192 has the same pin-out as the ubiquitous LM358 and, in terms of functionality, is the same, in that it comprises two opamps in one package. Connect the +iP and -IP of the spare opamp to opposite supply rails- it does not matter which way around.

I see in your drawing you've got the supply volts and GND with a cap across the opamp.

Invariably you should decouple all chips with a high frequency, low equivalent series resistance (ESR), capacitor. A 100nF disk ceramic X7R dielectric capacitor is good default decoupling capacitor for low power chips, but not for low distortion applications like audio amplifiers. In analog engineering you very often hear the term 'one hundred nan to deck'.

Decoupling capacitors must be mounted as close as possible to the integrated circuit pins and must have as short as possible leads. If you have + and - supply lines you need two decoupling capacitors, one between each supply line and 0V.

Incidentally, an LM723 linear regulator chip needs a default decoupling capacitor across its supply pins.

It is often thought that decoupling capacitors are a kind op optional luxury used by pedantic engineers, but that is not true. Decoupling capacitors are essential for the correct operation of a circuit.

And I'm unfamiliar with the term "ORI" above one of the resistors?

Me not thinking. On schematics it is conventional to use the multiplier as the decimal point. So in this case 0R1 stand for, 0.1 Ohms. By the same token, 2.2 Ohms would be written 2R2 and 470 Ohms would be written 470R.

Things get a bit complicated with high value capacitors. For example 10,000uF should be written as 1mF (one mili Farad) and 2.2 nano Farads would be written as 2n2F, which is the same as 0.0022uF or 2,200pF. It is all quite simple, once you know.

P.S I tried the LVD circuit, something went fizz pop bang, and it was the TLV IC. Haven't ruled out the PCB might have a construction error. Or perhaps I heated up the TLV too much during soldering,

Just the other day I was given a Pace MBT SV-4 rework station so no excuses for not soldering SMD components now!!

spec