Signal generator curent source 1MHz / 1A 1

[xj25]

Hi,

I need to measure a stray inductance in the range of 5uH to 0.1uH at a current of about 1A.

I thought of designing an small current source amplifier with a signal generator 5Vp input. Freq. to be used is planned fom 100kHz to 1MHz. I will inject a 1MHz/1A waveform So then I will measure the voltage drop in the load to calculate the inductance.

My background is electonic engineering but I don´t usually am in this part of the business, so don´t feel very confident with my steps. Note that in my department everybody look to me with eyebrow up when I tell them to try to do these kind of things...

From my old notes and some google search i reached to this simple opamp current source circuit with output follower stage. RL is the load, I don´t mind not to be ground referenced.

I found by simulating these facts about I would like some comment/feedback:
- I need an opamp about 9MHz BWxG product for getting the voltage follower to work propely up to 1MHz.
- I need transistors also of near ft 9MHz (i.e. with 4MHz ft transistors it won´t work)
- Even so, at 100k-1M freq. I get some nasty ringing overshoots in the crossover point. I avoid these by using diode biasing (kind of class AB amplifier) or a resistor biasing network as seen in the schematic. These resistor netwok driven by the op-amp circuit I don´t know how to calculate it (if possible), I adjusted it by trial and error, but seems to work and seems to me less parameter-sensible than the diode biasing option.

Any comment will be welcomed. Thanks for reading.

Note: the opamp and transistors seen in the drawing are not my current selection, the opamp only work partially to about 3kHz and the transistor up to 100kHz (in the simulation)

 

[LiteYear]

1MHz analog design is tricky. It's entirely possible your design will work, but you might find some seriously subtle prompting is necessary to make it happy. For what it's worth, we've successfully used a much lower tech circuit to measure inductances near this range. Effectively you just change a cap up to, say 10V, put a low inductance resistor in series with your inductance to limit the current, and then capture the waveform at the moment you switch the cap on to the circuit. You have to account for parasitic inductance, but this can be a surprisingly effective way to determine inductance properties.

 

[xj25]

Thanks for the idea Lite, I will try that also as 2th option just in case the 1MHz amplifier fails.
I think that once "hands on" I will continue with the analog circuit just to gain some experience and insight, I was also quite unsure about the MHz performance, that is why I am spending quite a lot of time in doing a simulation as real as possible with the component models.

The current components are in the adjoint circuit. With these the best that I get simulating at 1MHz input is 0.85A peak output (it should be 1A if everything would work ideally)....

 

[LiteYear]

Sure, a lot of experience and insight to be gained there. Keep in mind that your simulation will differ from reality because at 1MHz, parasitics become significant. For example, depending on how your tracks are routed, there will be parasitic capacitance between all tracks. Long tracks will have inductance, and some of your resistors might be a bit inductive too. You can either estimate these parasitics for your simulation, or observe good routing practice to minimise them in your design.

 

[xj25]

Just for info,

a quick prototype following previous schematic (I changed the polarizing network to AB class diode biasing) works well without much distorsion up to 300-400kHz. My wiring is terrible so that undoubtly is affecting.

In LTspice it puzzles me that with AC analysis the bode diagram BW is ok (gain constant) up to several MHz (that makes sense with opamp and transitor BW), but in transient simulation it falls in gain very sharply when you go higher than 1 MHz. Any idea why the difference between the results of both analysis?