# Amplifier Specifications for Lock-in Detection

Since a lock-in amplifier is effectively a high-precision AM demodulator, what are the most critical characteristics for amplifiers in the high-frequency signal chain? Take a 50kHz lock-in circuit for example where 40dB of gain is applied before performing synchronous detection by an AD630 or an ADC for digital mixing with and objective dynamic range >80dB. It is easy to find op-amps with <10nV/√Hz @ 50kHz so noise can be made negligible for a narrow output bandwidth (basically the rational for using lock-in detection). It seems to me that 1/f noise of the system may be dominated by amplifier power supply rejection, gain stability of feedback network, offset drift, etc.

Which amplifier parameters should be optimized in this case where we need to minimize introducing AM noise on a high-frequency narrow-band signal? Will some of these things (such as offset drift) be eliminated entirely by AC coupling?

Will some of these things (such as offset drift) be eliminated entirely by AC coupling?

Yes, coupling eliminates low frequency noise

You don't need to use an AD630, a circuit like this will do, then you can choose a suitable amplifier for your application with the right bandwith and noise. Source: http://www.analog.com/en/analog-dialogue/articles/synchronous-detectors-facilitate-precision.html

Another way to do the lock in amp is like this:

simulate this circuit – Schematic created using CircuitLab

Which amplifier parameters should be optimized in this case where we need to minimize introducing AM noise on a high-frequency narrow-band signal?

The amplifier parameters will be determined by how much SNR you need for your application. The bandwidth of the amplifier will be determined by the lock in frequency (it will need to go faster and much faster than the next cycle). Offset is usually not a problem since it shifts the DC and the ADC by a few bits.

• Say I need 80dB dynamic range in <1Hz BW and provided enough gain for 1Vpp full scale at the lock-in frequency. Then I only need 0.1mV stability at the detector, but if I had 50dB gain then my amplifiers offset and gain better not drift more than 0.3μV, right? So I probably need pretty good amplifiers and low TC parts, even for an AC signal.
– Mike
Commented Sep 12, 2018 at 23:20
• That is a helpful article. I think I got distracted by the shiny ADA2200 and missed the circuit above the first time I read it.
– Mike
Commented Sep 12, 2018 at 23:22

Yes, that is the rationale.

You can consider offset drift as a component of low frequency noise.

The amplifier noise in the bandwidth is what is important. By the way, 10nV√Hz is far from negligible in many applications. It's not just the amplifier noise, but external noise and Johnson-Nyquist noise in resistors in the circuit and so on. Be sure to consider both voltage and current noise and the spectrum of each.