Hi Jony130,

Thanks for you idea, I will try your suggestion this weekend, if it's not earlier.

How do you come up with the value of 22K and 1nF? my circuit is oscillating at around 343kHz.

Well I use trial and error method. I simply start from 10K and 100pF and I look on the scope and watched the result.

 

I try your circuit in LTspice and circuit oscillated badly.
But when I add 22K and 1nF oscillations magically disappear.

 

As I previously noted, you don't need A2. If you want some gain in the feedback loop (the apparent purpose of U2), just remove A2, connect R3 between ground and the (-) input of A1, and connect R2 between R1 and the junction of R3 and (-). The transconductance gain will then be (1 + R2/R3) x 1/R1 A/V. That circuit should have no stability issues.

Edit: Of course R2 and R3 will draw off a small amount of current that will add to the MOSFET current so you want to make the resistance sum of R2 and R3 high enough so that this current has negligible effect on your desired accuracy.

I think your logic is flawed. With the original circuit, the transconductance is

\(\bf Gm=\frac{1}{R1}(\frac{R3}{R2+R3})\)

Whereas your proposed modification, without the op amp (and ignoring the loading of R2+R3 on R1), is

\(\bf Gm=\frac{1}{R1}(\frac{R2+R3}{R3})\)

As you can see, the ratios are reciprocal.

 

I think your logic is flawed. With the original circuit, the transconductance is

\(\bf Gm=\frac{1}{R1}(\frac{R3}{R2+R3})\)

Whereas your proposed modification, without the op amp (and ignoring the loading of R2+R3 on R1), is

\(\bf Gm=\frac{1}{R1}(\frac{R2+R3}{R3})\)

As you can see, the ratios are reciprocal.

Where is my logic flawed? Both circuits have gain in the loop. What is the significance of the gain ratios being reciprocals and thus the gains being calculated differently?

 

Where is my logic flawed? Both circuits have gain in the loop. What is the significance of the gain ratios being reciprocals and thus the gains being calculated differently?

Take the case where he wants to get 1mA out. Let's make R2/R3=9.
With his original circuit, the input voltage would have to be 10mV.
With your circuit, the input would have to be 100μV. I think that is counter to what our OP is trying to accomplish.

EDIT: If you consider unity gain as being 1 A/V, then the original scheme has less than unity gain.

 

Take the case where he wants to get 1mA out. Let's make R2/R3=9.
With his original circuit, the input voltage would have to be 10mV.
With your circuit, the input would have to be 100μV. I think that is counter to what our OP is trying to accomplish.

EDIT: If you consider unity gain as being 1 A/V, then the original scheme has less than unity gain.

OK, you got me there. It just didn't occur to me that he was using U2 to reduce the transconductance since it's unnecessary to use an amp for that purpose. You can remove U2, connect R1 directly to U1's (-) input and just use R2 and R3 as an attenuator to the (+) input of U1.

Or am I missing something else?

 

OK, you got me there. It just didn't occur to me that he was using U2 to reduce the transconductance since it's unnecessary to use an amp for that purpose. You can remove U2, connect R1 directly to U1's (-) input and just use R2 and R3 as an attenuator to the (+) input of U1

I have been wondering why it wasn't done that way.

 

I have been wondering why it wasn't done that way.

because at the time I designed it, I didn't think of the other way, and lm358 come with 2 op amps, I thought I could use the other op amp for something

 

because at the time I designed it, I didn't think of the other way, and lm358 come with 2 op amps, I thought I could use the other op amp for something

Well, now you know that putting two op amps in the same loop can lead to instablity.

 

For the case 2 you need to add a resistor between the op amp output and non inverting input.

Also in February I have a similar problem with a voltage control current source.
And I was force to add 22K and 1nF capacitor to prevent circuit from oscillations.

Without 22K and 1nF

And after I add 22K and 1nF.

that works quite well with a lm358 too, thanks Jony130

 

Well, now you know that putting two op amps in the same loop can lead to instablity.

so here is what I did, with different technique mentioned above (those improvements are not in the drawing)

 

here are the waveform I get, I expected the waveform to be better.

• Am I probing the circuit in the wrong way(is there any technique I should be using in low voltage probing, should I being using averaging function in the scope?)
• Is that the waveform you would expected?
• If not above, how can I improve the circuit?

NOTE: I am using X1 probe, the scope is Rigol 1102E, 100MHz