INA103: Large DC bias at output from op-amp

Gabriel Altman

Part Number: INA103

Hi,

I'm trying to build an instrument to measure small inductances as part of a school project. The circuit drives a sine wave through a sense resistor and an unknown inductor. The INA103 amplifiers are used to convert the differential voltage across a 10 ohm sense resistor and across the unknown inductor (range of 50-80nH) into single ended measurements so that they can be digitized with an ADC. I've simulated the circuit in LTSpice, and the model yielded the desired outputs. I used Altium to design the PCB and had it manufactured by JLCPCB.

When the board arrived, I measured the outputs from the INA103 instrumentation amplifiers using an oscilloscope. The input signal is a sine wave, 1MHz, 900mVpp with a 460mV DC offset. I was expecting to see sine waves with amplitudes in the range of of 1.5V to 1.7V and a DC offset of 0.75V to 0.9V. Instead, I see sine waves with an amplitude of roughly 160mV and a DC offset of 12.3V. I don't understand where the large DC offset is coming from.

I'm supplying the amplifiers with +/- 15V.

Where did I go wrong? Is the SPICE model inaccurate? Did I fry the amplifiers when I soldered them onto the board? Did I make some sort of hard error in my PCB? Images attached. I can provide many more if it would be helpful.

LTSpice Model

Schematic image

Altium Schematic Image

Thanks for your help!

over 2 years ago

0 kai klaas69 over 2 years ago

Guru 140200 points

Hi Gabriel,

the INA103 is not fast enough with that high gain:

gabriel_ina103.TSC

As you can see the frequency response shows a dampening of 1.8dB at 1MHz. This also means that the INA103 has no linearizing gain reserve at all. Even with a gain of 0dB you would only have a gain reserve of about 16dB, which is not sufficient. And at higher output signal amplitudes the slew rate limit of 15V/µs of INA103 would come into play, finally. So your INA103 isn't fast enough for your application.

I would go for another approach:

When using an inverting amplifier, you would not have to deal with differential signals, neither for the current and nor for the voltage. And you wouldn't need to measure the measuring current either, because R1 is connected to virtual ground. The measuring current is directly Vg1 / R1, provided your OPAmp is fast enough.

The output of OP1 directly delivers the AC voltage drop across the inductance to be measured. Eventually a dampening resistance should be connected in parallel to the inductance to dampen the self resonance of inductance. The discussion about this I leave to you, as this is a school project.

Take a really fast OPAmp, though. A 100MHz OPAmp would provide a gain reserve of 40dB at 1MHz. Just enough to guarantee a true virtual ground at the -input of OPAmp

Kai

0 Gabriel Altman over 2 years ago in reply to kai klaas69

Prodigy 20 points

I think I found a large error in my design: the Sense and Output pins are not connected. I will connect them and let you know the results.

0 kai klaas69 over 2 years ago in reply to Gabriel Altman

Guru 140200 points

Hi Gabriel,

yes, this would explain a lot.

Kai

0 Tamara M Alani over 2 years ago in reply to Gabriel Altman

TI__Genius 10850 points

Hi Gabriel,

Thanks for your post. Please let us know if you run into any issues moving forward with your design.

0 Gabriel Altman over 2 years ago

Prodigy 20 points

I bridged the Sense and Output pins on the INA103. The output waveforms are now a very close match to my LTSPICE simulation. The circuit works!

The reason for the initial failure was my error in reading the datasheet and reconciling the pins in the SPICE model with the actual pinout of the amplifier.