Understanding noise RMS and picking an appropriate current source for Hall probes

Thread Starter


Joined Nov 2, 2020
Hi Everyone,

I am trying to define the most suitable current source for my application. Essentially, I need a high stability current source to feed Hall probes sensors. The probes' nominal current is 10 mA. A stability of 100 ppm (i.e., 1 uA) for short term (a few hours) is desired. Usually, I can see two main information in the datasheet: the accuracy and the RMS noise. Accuracy is associated with systematic errors, so it should not affect the stability. On the other hand, RMS noise seems to be a good starting to evaluate stability.

Let me use the Keithley 2450 datasheet as an example (see the image attached). For a nominal current of 10 mA, let's suppose I use the range 10 mA. For that range, the noise RMS (< 10 Hz) is 40 nA. My first question is: what exaclty this information means? Does it mean that the output current would have a standard deviation of 40 nA if a sample it in 10 Hz? What if I use a sampling rate of 50 Hz? Would the noise RMS be higher? Can we estimate it?

Last but not least: how would you evaluate whether the instrument meets the 100 ppm stability? Which variables (besides price) would you take into account to make a decision?

Thank you all.


sparky 1

Joined Nov 3, 2018
TI presentation on noise conciderations. It is possible you may need precision ultra low noise instrumentation op amp.
when you pull out a Keithley with fA range the room goes quiet. The pcb layout can make a difference that is why some would rather
purchase an evaluation board, it depends on which range you are working and if you decide digital will suit your needs than analog.
The LTC2400 modele, usb module and things and stuff, stable current supply, the data file might be a useful set of values on a pc.
Having a module calibrated is sometimes necessary when publishing important results.

It would be interesting to know the real criteria and evaluate the actual noise output. 100 ppm current seems doable?
let's say you put together a ratiometric hall sensor with 10 mA LM334, an OP847 and a few 9V batteries.
It could tell you something about the real time criteria depends on the bench work.
Last edited: