Voltage controlled current source circuit needed for high currents

[doncarlosalbatros]

I need to make a voltage controlled current source for nichrome wire. It seems like I need a circuit which translate 0...5V range voltage input to a 0...8A current. I mean a voltage controlled current source. I can use a DC power supply which can supply this current.

Is there a topology or any suggestion circuit for example composed of an opamp and a power transistor? Or any specific IC would work as well.

I would be glad if one can share a circuit diagram or components for this purpose.

Regards,

 

[d123]

Hi,

This is typical, or using a Darlington pair.

Voltage Controlled Current Source - Current Servo

Page 3 of the pdf has a couple of circuits.

 

[crutschow]

What is the resistance of the nichrome wire and what is the supply voltage?

A linear current regulator would be the simplest, but could dissipate a lot of power, depending upon the load resistance and supply voltage.

If so, you may need a PWM type of circuit, which is more complex but would dissipate very little power.

 

[doncarlosalbatros]

@crutschow Heating wire has a resistance 1.5 ohm. I didn't buy the DC supply yet but can be 12V or 24V.

 

[crutschow]

@crutschow Heating wire has a resistance 1.5 ohm. I didn't buy the DC supply yet but can be 12V or 24V.

A 12V supply would minimize the power loss with a linear regulator but the maximum current would probably be about 7.9A due to losses in the circuit.
Is that acceptable?

A 12V supply would give a maximum power dissipation in the regulator of 24W at the quarter-power point, so the regulator would need to be on a good heat-sink, likely with a fan.
Does that sound okay?

 

[RobertFalbo]

Not sure how much regulation he wants, but just a mosfet

They are a voltage controlled current source.


This would be a cheap and dirty driver


https://assets.nexperia.com/documents/data-sheet/BUK92150-55A.pdf

just put this in line with your load and drive the gate with your 0 to 5 V.

 

[KlausST]

Hi,

What precision do you need?
Why a current source?
Do you need true DC current, or is pulsed current allowed, too?

You say heating...therefore I assume you need not very high precision...and pulsed current is allowed.
Then go for a PWM solution.
A DC current 0% / 50% / 100% gives 0% / 25% / 100% heating power.
PWM regulation:
0% / 50% / 100% duty cycle gives 0% / 50% / 100% average current, 0% / 71% / 100% RMS current, 0% / 50% / 100% power.

Please see the different values at 50% setpoint.
Do you want a linear current curve, or a linear power curve?

Klaus

 

[doncarlosalbatros]

@KlausST Yes I can use PWM, since Im going to use a microcontroller with 490Hz freq. I guess this frequancy okay since the load will be resistive?
The wire's temperature is important, so I guess linear power would be better?
Here is the circuit in my mind so far what do you think about it?

 

[KlausST]

Hi,

use a "logic level MOSFET" and you don´t need Q1.

490Hz is good. You could go even lower.

With the PWM you get about linear realtionship: duty_cycle to power.

***
The circuit doesn´t show CURRENT measurement. I recommend to use it if you want true current control.
--> shunt between MOSFET_source and GND --> amplifier with LPF --> ADC

Klaus

 

[doncarlosalbatros]

Here is my final circuit. What do you think about it?:

 

[CataM]

Is there any reason for the mosfet to be in the high side?
There is no LPF after the instrumentation amplifier.

 

[crutschow]

U2 can be a rail-rail op amp, so you wouldn't need a negative voltage.
R1 needs to be larger so the amp input offset voltage isn't a large part of the signal.
1mΩ gives only 8mV max. of signal which would require a very low input offset amp.

Your are driving the N-MOSFET as a source-follower, which means it will have over 2V across the drain-source when on, dissipating a lot of power (over 16W which requires a large heat sink).
That also means you would need about a 15V supply to get 8A through your load.

If you want to drive a grounded load with a high-side switch and minimum dissipation, then you need a P-MOSFET not an N-MOSFET.

 

[doncarlosalbatros]

CataM I took it low because I thought it would be problem wire sitting live. Do you think I should take it low side?
And here is a document https://ww1.microchip.com/downloads/en/AppNotes/01332B.pdf about it but beyond my capacity to conclude.

And the output will be PWM pulses at 500Hz or lower. So the LPF should not distort the pulse shapes. What do you advice?

Thanks

- - - Updated - - -

crutschow Rail to rail opamp causes offset problems thats why I used inamp

- - - Updated - - -

Can you recommend me an N-MOSFET for this application? (Not surface mount please)

- - - Updated - - -

Now I understand why it is better switch is at low side. It dissipates much less power. I will upload modified circuit soon

- - - Updated - - -

How about this configuration? I took the N-Mos low and it dissipates very little power:

 

[crutschow]

The output to the MCU from U2 will be the PWM pulses with a voltage proportional to the peak current.
Is that what you want, or do you just want a DC voltage proportional to the average current?
If you want the average, then you can add an RC low-pass filter on U2's output.

What's the purpose of D2?
The maximum output of U2 is about 1V less than the supply voltage.

 

[doncarlosalbatros]

The output to the MCU from U2 will be the PWM pulses with a voltage proportional to the peak current.
Is that what you want, or do you just want a DC voltage proportional to the average current?
If you want the average, then you can add an RC low-pass filter on U2's output.

What's the purpose of D2?
The maximum output of U2 is about 1V less than the supply voltage.

First time I deal with such application.
You are right I think better DC voltage proportional to the average current.
What R C values for LPF averaging do you suggest for that? The PWM freq is 490Hz.
Should I use an op amp integrator or passive?

Purpose of D2 it is a 4.7V zener to limit possible over voltages to MCU input.

"The maximum output of U2 is about 1V less than the supply voltage."
Oh you are right then I should set the gain such that peak should not exceed 4V right?

Many thanks so far for your great insight and suggestions!

 

[crutschow]

...........
You are right I think better DC voltage proportional to the average current.
What R C values for LPF averaging do you suggest for that? The PWM freq is 490Hz.
Should I use an op amp integrator or passive?

Purpose of D2 it is a 4.7V zener to limit possible over voltages to MCU input.

"The maximum output of U2 is about 1V less than the supply voltage."
Oh you are right then I should set the gain such that peak should not exceed 4V right?
....

A passive filer should be fine.
For 490Hz and less than 1% ripple you want the filer time constant to be a least 50 times the pulse period of 2ms or at least 100ms, giving a low-pass corner frequency of 1.6Hz.
For less ripple you can increase the time-constant at the expense of a slower response time to changes in the current.

If you need a faster response time with a small ripple, then you need to go to a higher order active filter.
To avoid feed-through of the edge spikes of the PWM signal with little overshoot for a step change in signal, a 3-pole, single opamp Bessel filter works well. The component values can be calculated **broken link removed**.

Yes you need to set the gain for about a 4V maximum output.

My point was that, since the maximum output is 4V, D2 is redundant and not needed.

 

[KlausST]

Hi,

Do the current measurement in the source path of the mosfet.
You don't need the driving bjts when using a logic level mosfet.

I recommend a 2nd order butterworth with about 20Hz LPF.
No INA needed.
I recommend to use a slew rate limiting RC at the input of the OpAMP.

Klaus

 

[doncarlosalbatros]

Hi,

Do the current measurement in the source path of the mosfet.
You don't need the driving bjts when using a logic level mosfet.

I recommend a 2nd order butterworth with about 20Hz LPF.
No INA needed.
I recommend to use a slew rate limiting RC at the input of the OpAMP.

Klaus

Klaus

1-) "Do the current measurement in the source path of the mosfet." crutschow says Driving the N-MOSFET as a source-follower will have over 2V across the drain-source when on, dissipating a lot of power (over 16W which requires a large heat sink).

2-) Since I take the MOSFET on low side I cannot use a non inverting op amp since it will not be pulled to ground. I tried a lot in simulation even I take the load between source and ground there is offset.

3-) Can you suggest me a logic level MOSFET for this application?

4-) Why do you suggest to do the current measurement in the source path of the mosfet?

 

[crutschow]

I cannot use a non inverting op amp since it will not be pulled to ground. I tried a lot in simulation even I take the load between source and ground there is offset.

What op amp did you use?
How much offset did you see?

 

[CataM]

3-) Can you suggest me a logic level MOSFET for this application?

What Klauss is trying to say here:

You don't need the driving bjts when using a logic level mosfet.

is that your IRL3803 is already a logic level N channel mosfet (a very good one for this application by the way..), which means, you can drive it directly from the MCU_PWM pin with a properly selected resistor at the gate.

-----

4-) Why do you suggest to do the current measurement in the source path of the mosfet?

That is necessary only if you implement the active low pass filter suggested by Klauss. If you stick with the Instrumentation OP amp, you are good...