Modulated current limiting circuit

[Nyla]

Hi all,

I'm looking for a way to make a circuit that will limit the output current between around 20 and 500mA (200mA will do if it really can't be done). It could be controlled either with a serial interface, a pwm signal, digital parallel signal, analog signal, etc, as long as it's controllable with a microcontroller.

Usually I would use a LM317 with a rheostat, but I want to get rid of the rheostat. I looked into using digital potentiometers, but the maximum continuous current is always around 5mA or less, which is way too little.

I also looked into using a resistor ladder, but I think it's a bit of a bulky solution for something that should be possible otherwise.

The circuit should be able to work at different output voltages between 2 and 5 volts, though it would be acceptable if the components had to be calculated for a fixed voltage or if the system had to be recalibrated for each voltage...

 

[Space Varmint]

Hi all,

I'm looking for a way to make a circuit that will limit the output current between around 20 and 500mA (200mA will do if it really can't be done). It could be controlled either with a serial interface, a pwm signal, digital parallel signal, analog signal, etc, as long as it's controllable with a microcontroller.

Usually I would use a LM317 with a rheostat, but I want to get rid of the rheostat. I looked into using digital potentiometers, but the maximum continuous current is always around 5mA or less, which is way too little.

I also looked into using a resistor ladder, but I think it's a bit of a bulky solution for something that should be possible otherwise.

The circuit should be able to work at different output voltages between 2 and 5 volts, though it would be acceptable if the components had to be calculated for a fixed voltage or if the system had to be recalibrated for each voltage...

PWM would work fine. Filter it and run it to the base of say say a 2n3055.

 

[crutschow]

Can the current sense resistor be in the ground return or does it need to be on the high side output?

 

[Nyla]

I dont exactly understand what you mean, where would I put this transistor?

The circuit I've been working with is this one:

**broken link removed**
Where the current supplied to the load is 1.25/R.

Do you mean I put the 2n3055 in the place of the resistor?

It's for driving a laser diode btw

 

[Roff]

I'm designing something for you.
What settling time do you require from the time the current is told to change until it gets acceptably close to final value (1%, 10%, etc.)?
What supply voltage(s) do you have available?
What does this mean?

The circuit should be able to work at different output voltages between 2 and 5 volts, though it would be acceptable if the components had to be calculated for a fixed voltage or if the system had to be recalibrated for each voltage...

Does this mean the maximum voltage across the load is 5 volts?
Is the load inductive?

 

[saturn1bguy]

See the following appnote from Maxim:

https://www.electro-tech-online.com/custompdfs/2008/11/AN3464.pdf

 

[Nyla]

I'm designing something for you.
What settling time do you require from the time the current is told to change until it gets acceptably close to final value (1%, 10%, etc.)?
What supply voltage(s) do you have available?
What does this mean?
Does this mean the maximum voltage across the load is 5 volts?
Is the load inductive?

The load is a laser diode. Different types use between 2.5 and 4.5v, it would be nice if the circuit could work with any type. Supply voltages would be regulated by a seperate lm317 regulator so they can be adjusted accordingly,t I'm only worried about limiting the current. The settling time should be less than 1ms, it gets updated very frequently. Ideally it could do 20000-50000 updates per second. The most used drivers at the moment are analog and usually driven by a soundcard DAC.

 

[blueroomelectronics]

Do you have a datasheet on the diode? Some have a feedback connection at least the old ones I've seen had this feature, might be different now.

 

[Nyla]

Do you have a datasheet on the diode? Some have a feedback connection at least the old ones I've seen had this feature, might be different now.

All the diodes I've seen do have a feedback photodiode but this is never used as we don't really care. Imho it can be omitted from the circuit.

 

[Roff]

The load is a laser diode. Different types use between 2.5 and 4.5v, it would be nice if the circuit could work with any type. Supply voltages would be regulated by a seperate lm317 regulator so they can be adjusted accordingly,t I'm only worried about limiting the current. The settling time should be less than 1ms, it gets updated very frequently. Ideally it could do 20000-50000 updates per second. The most used drivers at the moment are analog and usually driven by a soundcard DAC.

I think PWM with 1mS settling time would be tough, due to the filter delay and risetime. I'm pretty sure you can't use unfiltered PWM on a laser diode. 20k-50k updates/sec is definitely out of the question for PWM. I think you need a DAC driving the voltage-controlled current source.
That Maxim app note looks promising.

 

[Nyla]

Yeah on second thought PWM doesn't seem very suitable. I'm not entirely sure how to apply the appnote to a circuit that would work for me though...

 

[Roff]

I'm not entirely sure how to apply the appnote to a circuit that would work for me though...

The only potential problem I see is the time required for the serial interface, but that should still allow you to update 50k times per second.

 

[Ubergeek63]

HV9910 LED driver

 

[Roff]

HV9910 LED driver

AFAIK, you can't PWM a laser.

 

[Optikon]

Why not just do a voltage to current converter? One opamp, sense resistor and power mosfet. Then you can program any current you like with a DAC over a serial port. Easy to accomplish 20mA to 500mA. No special reason to use the cruddy 317 right?

 

[Roff]

Why not just do a voltage to current converter? One opamp, sense resistor and power mosfet. Then you can program any current you like with a DAC over a serial port. Easy to accomplish 20mA to 500mA. No special reason to use the cruddy 317 right?

That's basically what is described in the Maxim app note.

 

[saturn1bguy]

Why not just do a voltage to current converter? One opamp, sense resistor and power mosfet. Then you can program any current you like with a DAC over a serial port.

If the OP used a parallel DAC (instead of a serial DAC or the digital pot which sports a serial interface), he might be able to get the update rate he requires...

 

[Nyla]

Ok here's a compromise.

To make the circuit useful to people who require the very fast update rate, it should accept an analog signal between 0 and 5v (this is how another driver works). It can then be interfaced to a soundcard DAC which should provide them with the required speed.

For less high end requirements such as simply fading the laser over a longer period of time, a ~12+ bit DAC could be used with a microcontroller instead.

Is this possible with the circuit from Maxim? If so, can someone please tell me which exact components I would need to achieve this (such as the 'pass transistor')?

Also, for the optional DAC, would I need a voltage output or current output version?

 

[saturn1bguy]

Ok here's a compromise:

To make the circuit useful to people who require the very fast update rate, it should accept an analog signal between 0 and 5v (this is how another driver works). It can then be interfaced to a soundcard DAC which should provide them with the required speed.

Excuse me sir, I'm not sure people are necessarily standing by waiting to design something for you, free or otherwise. Folks are trying to steer you in the right direction.

Also, for the optional DAC, would I need a voltage output or current output version?

The circuit in question is a voltage-controlled current source, as the app note indicates. Thus a voltage output DAC. Besides, an 8bit DAC would suffice for circa 2mA steps for 0-500mA, which is pretty fine.

 

[Nyla]

Excuse me sir, I'm not sure people are necessarily standing by waiting to design something for you, free or otherwise. Folks are trying to steer you in the right direction.

I understand, I didn't mean for the circuit to be actually designed by anyone though I see now it did sound that way.

As for the components, what I'm not sure of is the pass transistor P (the appnote doesnt mention any examples I could use), whether to use a DAC or digipot (the appnote uses a digipot, I wonder why, because DACs can be more accurate and the main advantage of a digipot seems to be that they can be more useful in audio applications if they have a log-taper).

Also, I've been looking at different DACs, and many seem to have voltages ranging from 1.8 to 5.5v. Does this mean that this is the output voltage, ie Vin+ would range from 1.8 to 5.5 in (5.5-1.8)/2^10 = ~3mV steps (for a 10bit dac)? Because if I would make it so that it accepts an input from 0 - 5v (if not using a DAC) then I would simply set Rsense to 10 ohms to achieve a current output of 0-500mA, right? But I would have to make sure the DAC also outputs a signal from 0-5v, rather than 1.8-5.5 (or the minimum current would be 180mA, which is way too much).

All assuming I operate the circuit at 5v.

Am I doing it right? How would I modify the range from the DAC to fit 0-5v?


Also, what is the function of the shunt regulator?
"a shunt regulator (U3) provides a constant reference voltage across the digital pot."

Does this still apply if I use a DAC, or does anything else change? I'm thinking I shouldn't use a DAC.