How complicated can a voltage follower be?

[Rolf Startin]

Hello Everyone

I'm really struggling with something as simple (?) as a current-boosted voltage follower.

Putting a supply-tied transistor or FET into the feedback loop of a unity gain op amp works well every time, but I can't seem to manage to devise a circuit that works with the current-boosting FET tied to 0V.

It is essential that the circuit sinks rather than sources current and that the load is tied to supply and I only have a single 0V/28V DC supply to power the op amp.

I'm missing something here I know - I thought it was that the feedback was of the wrong polarity, and tried to remedy this by inverting the feedback (see 3rd circuit). Now I'm just floored!

Can anyone show me the light please?

 

[audioguru]

You have feedback oscillation because the Mosfet is not a follower, it has voltage gain.
use a P-channel Mosfet and add a negative supply so that the output can go to 0V if you want.

The output of a Mosfet follower is its source, not its drain.

 

[Rolf Startin]

Hello AudioGuru and thanks a lot for picking this up.

There's no way I can provide a -ve supply so that option is out i'm afraid.

All I want to do is to faithfully supply a load which is tied to the +ve rail. My project is to deliver 3 variable voltages (faithful to the input waveform) to an RGB LED strip with a common anode). Is this really impossible given a strictly single supply for both op amp and transistor?

I'd be happy to use a bipolar transistor instead - does this change things?

Thanks a lot

 

[on1aag]

Hi Rolf Startin,

It's very odd to apply a voltage to a circuit that is
referenced to gnd and the output voltage of that
circuit should be referenced to the supply voltage ?????

I also don't understand why you would want to apply a
variable voltage to a led bar, are there current limiting
resistors in series with these leds ? (which can not be removed)
Wouldn't it be simpler to control the current through these
led bars ?

Can you could provide more information about the led bars ?

on1aag.

 

[crutschow]

Your first circuit should work with C1 to control the oscillation. If you reduce the value of C1 you should be able to find a value the eliminates the oscillation without degrading the waveform.

You could use a PNP transitor as an emitter follower (emitter connected to the load). That would allow the output to go to about 0.7V above ground minimum without a negative supply, if that's acceptable.

Make sure the op amp you use is a rail-to-rail device.

 

[audioguru]

You don't need a rail-to-rail opamp. An MC33171, MC34071 or half a dual LM358 opamp have inputs that work to ground and their output goes to almost ground.

 

[Roff]

What is the range of current in your load?
Does the LED strip have a series resistor?
If you are trying to modulate the current, you need a voltage-controlled current sink, not a voltage follower.

 

[Rolf Startin]

Woo Hoo!

Crutschow - you hit it on the head.... working circuit attached. I used a BDT62C PNP darlington in your suggested configuration and it's good for a 1.6A load.

Audioguru - my op amp is LN324N.... I can certainly improve things using a better op amp as you suggested.

On1aag - The strips are by Osram, using tiny current limiting chips rather than resistors. I have a (trapezoidal PWM) system to control single colour LED strips which is practically free of conducted & radiated RF emissions but it's referenced to ground and so I can't gang 3 common anode strips - Osram wisdom, out of my control!

Roff - Maximum drain will be around 2.5A. I looked up the attached current sink (voltage controlled) and built it. It's dreadful - the feedback voltage is load dependent - I don't understand it. The load is current-limited all by itself so I'm staying in the voltage domain!!

Any refinements or further thoughts would be very much appreciated

Thank you all very much!

 

[audioguru]

The output of your BDT62C circuit (it is not a current sink) should follow the input voltage from 2V to 26V perfectly. If the output tries to have a reduced output voltage then the gain of the opamp (100,000 at DC and at very low frequencies) reduces the error so that you won't notice it.

The OP-77 plus Mosfet current sink circuit provides an output current controlled by the input voltage. Its output voltage changes when the load changes to keep the current constant.

The lousy old LM324 has trouble at frequencies above 2kHz so maybe the frequency is your problem. The MC34071 opamp works well up to 100kHz.

 

[MrAl]

Woo Hoo!

Crutschow - you hit it on the head.... working circuit attached. I used a BDT62C PNP darlington in your suggested configuration and it's good for a 1.6A load.

Audioguru - my op amp is LN324N.... I can certainly improve things using a better op amp as you suggested.

On1aag - The strips are by Osram, using tiny current limiting chips rather than resistors. I have a (trapezoidal PWM) system to control single colour LED strips which is practically free of conducted & radiated RF emissions but it's referenced to ground and so I can't gang 3 common anode strips - Osram wisdom, out of my control!

Roff - Maximum drain will be around 2.5A. I looked up the attached current sink (voltage controlled) and built it. It's dreadful - the feedback voltage is load dependent - I don't understand it. The load is current-limited all by itself so I'm staying in the voltage domain!!

Any refinements or further thoughts would be very much appreciated

Thank you all very much!

Hi,

What exactly isnt working? What does the non inverting input connect
to?