I have a primary inductor L1 connected to a voltage source with frequencies varying from 0.5MHz to 15MHz. This primary inductor induces a voltage across a secondary inductor via coupling and I want to investigate the response generated in L2.

So I am trying to amplify a voltage from an inductor using an op amp in non-inverting configuration (image1). Upon doing an AC sweep I find that there is resonant frequency peaking at 3.5MHz which is presumably due to L2 forming an LC tank with the input capacitance(2pf) of the LT1886 amp (image1).

Now this is a problem because I want a reasonably flat band frequency response from 0.5MHz to 15MHz, so I was wondering how can I achieve such a thing if possible?

Here's a few things I've tried:

  1. Place a reasonably large resistor R3 in the non inverting input to decrease the quality factor of the LC tank which should smooth out the frequency response (image2). However looking at the frequency response (image2) the roll off after 3.5MHz is still unacceptable, because I want a flat band response upto 15MHz.

  2. Move the resonant frequency further away by decreasing L2 and then adjusting R3 accordingly (image3). This worked out quite well however I'm afraid that if the inductance L2 is too low the coupling will be insignificant and so the final amplified voltage is dominated by the amplifier's noise due to low signal to noise ratio.

I was also wondering if looking into current sensing would be a better option instead of directly sensing the inductor's voltage?

Image 1: Image 1

Image 2: Image 2

Image 3: Image 3

  • \$\begingroup\$ Just a first question. Is the L1:L2 coupling that low? \$\endgroup\$ – jonk Feb 10 at 2:15
  • \$\begingroup\$ In my application the inductors will probably be separated by one or two cms with air cores. So it's probably not that low \$\endgroup\$ – clostar Feb 10 at 2:28
  • \$\begingroup\$ I am struggling to see how you get such a broad response from a coil pickup. Your "sensor" has, itself, a frequency-dependent reactance. I'm imagining much more fundamentally limited bandwidths. Maybe someone else can teach me how you'd get what you want -- a Butterworth bandpass using an inductive pickup with a fractional bandwidth response of 5.3. I'm struggling, frankly. You can look at something I recently wrote but I don't think it helps much here. \$\endgroup\$ – jonk Feb 10 at 3:16
  • \$\begingroup\$ Exactly how flat does the frequency response have to be, and why? What S/N ratio do you need? L1 has a very large value for 15MHz. What physical component or device does it represent? \$\endgroup\$ – Bruce Abbott Feb 10 at 4:25
  • \$\begingroup\$ Ok thanks jonk I will take a look at the link! There is no specific flatness requirement but I'm hoping for a maximum of 1dB deviation. As for the S/N ratio I need, again I'm not too sure either but the amplified signal will eventually be fed into a 12bit ADC. I'm trying to build a device that detects the resonant frequency of other circuits using the coupling between the L1 and L2 inductors and their interactions of magnetic fields and to do so I need a flatness in their voltage vs frequency response. \$\endgroup\$ – clostar Feb 10 at 5:13

Here's a few things I've tried:

  1. Place a reasonably large resistor R3 in the non inverting input to decrease the quality factor of the LC tank which should smooth out the frequency response

I tried the same idea, but with resistance in parallel with the pickup coil instead of in series.

I used an LT6202 for low noise and a suitable bandwidth. The 2 pF capacitor represents estimated stray capacitance in the pickup coil and wiring.

enter image description here

Frequency response is flat within 1dB from 100 kHz to 15 MHz. Note how the op amp's high frequency rolloff flattens out the peaking coil response.

enter image description here

With an input level of 1 V rms the signal to noise ratio is ~40 dB. That's only about 6.5 bits of ADC resolution, but could be greatly improved by oversampling and/or better coil coupling.

enter image description here

Of course this is only a simulation. Real-world performance will be critically dependent on coil characteristics.

| improve this answer | |
  • \$\begingroup\$ Thanks! You pretty much got bang on what I wanted. Using the op amp's roll off to smooth it out was indeed very clever! I did some simulations on this and I found that the circuit was very sensitive with variations in L2 and C1 and just adjusting R1, R2 and R3 wasn't adequate to get the response I wanted so in reality I don't think I'll be able to use this circuit \$\endgroup\$ – clostar Feb 10 at 19:21
  • \$\begingroup\$ Even simple amplifier circuits are sensitive to small capacitances and inductances at these frequencies. It's the same for a eg. a scope probe, which has to be carefully designed and adjusted for flat frequency response. BTW you might want to do some research on 'grid dip' or 'FET dip' meters which use a loosely coupled oscillator coil to find resonant frequencies in a circuit. \$\endgroup\$ – Bruce Abbott Feb 10 at 21:27
  • \$\begingroup\$ Yep, now that I think about it, you're right. Also, I just realised that my L2 can be signficantly lower than 1mH and still achieve a decent amount of voltage, so I think using a 20uH inductor will be feasible! Yep I have done my research on dip metres and the method of operation for my circuit will be different to them. Just a quick question, how does the passive probe achieve a flat band response across a large range of frequencies? Can't I just use a probe and connect that to an amplifier? \$\endgroup\$ – clostar Feb 10 at 22:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.