Noise in Lock-Ins

There is no standard definition for lock-in amplifiers as to how to measure the input noise figure. This makes nonsense of other manufacturers figures and certainly makes comparisons difficult. Many of Scitec’s competitors simply put the noise figure of their input stage op-amp down on their data sheets. This is easy to do but is incorrect as it is only valid for very high gains (x100000) and is not applicable for the lower first stage gains that are typically used (x10).

Taking actual measurements is difficult as well. Where should you take the measurement? after the first op-amp? (which will have a different gain from instrument to instrument) after the input gain stage? (if so at what gain setting?) or at the output of the instrument (which has far more noise sources than just the input).

Finally, we know some of competitors’ figures are nonsense (or at the very least misleading) due to the circuits they use. Most lock-in's have AC coupled input which is produced by a simple RC filter. The resistor used is typically very high so as not to affect the input from a high impedance source and may have a value of 100M ohms. If you calculate the noise generated from this resistor you find it produces thermal noise of 1.27 µV / Hz1/2. This is a couple of orders of magnitude larger than then figure specified in the data sheets we are criticizing. The Scitec 400 series of lock-in's gets around this problem by DC coupling the input stage and does not use any resistors larger than 470k ohms.

Unfortunately, none of the above helps the customer make a comparison between Scitec and competitors regarding noise figures. We are not in a position to be able to test our competitors units so can not state that we are better or worse than they are. We can only say that the figures they give at best require a good deal of justification as to how they were produced.

• Back to Signal Processing
• Back to Lock-In Amplifiers