0.1 Hz to 10 Hz Passive Filter

[Alex Nikitin]

This is not a replacement for a "proper" noise measuring amplifier/filter as discussed in other thread(s). However I made this filter to measure and compare LF noise on HP3245A and other sources with not the lowest noise levels and it is perfectly useable for this purpose. The filter is connected to the input of a (slightly modified, hence "M") Keithley 2015M set at 1NPLC and 100mV DCV. The self-noise of Keithley with a shorted input is below 1uV p-p, with the passive filter connected it increases to about 2uV p-p - or about 0.2ppm at 10V DC input. The components are what I had at hand, a pack of 2.2uF polypropylene capacitors, a small diecast box and a 3-position switch (to defeat the filter fully or only the LPF part, the switch and the protection diodes are not shown on the circuit).

The advantages of a relatively small PP capacitor are very low leakage (picoAmps) and very fast settling time if the DC voltage is changed. The disadvantages are  higher noise, first order filtering and a slight attenuation (about 2%, so can be considered benign).

Cheers

Alex

[jfet]

This  will reveal my age, but we did the same thing back when we used strip chart recorders on low level measurements.  You would have a fast, high level transit come along and it would make the recorder peg for several seconds.  You reduced the bandwidth and it would make the signal clean enough for us to see the signal at its low level.  I made a box that had an strait thru position, then 30 hz , 10, hz, 3 hz and 1 hz.

[Alex Nikitin]

Some further results with a better voltmeter. The noise floor with the filter is now about four times better (~500nV p-p at NPLC1) and a 9V battery doesn't add to it. The noise measured from the Fluke 731B 10V output is about 1.5uV or 0.15ppm p-p. Good enough for me!

Cheers

Alex

[HighVoltage]

I agree, this is an amazingly stable volt meter.

You probably need to find some lab that is capable to calibrate a 3458A.

I have only done comparison measurements, especially to one of my 7 1/2 digit Keithley 7510
And both are just rock solid to each other.
And both of these instruments need only a fraction of the time to warm up, compared to a 3458A.

[Alex Nikitin]

The Keithley is three times the price of Hioki and if all you need is the DC precision, the Hioki can be a great building block as it has all required functions.

I think we better to discuss Hioki in the relevant thread!

Cheers

Alex

[Alex Nikitin]

Here is my attempt to measure the settling time of the filter, the source is Keithley 263 starting at 0V output , switched to +10V, switched to -10V and back to 0V. The horizontal axis in samples, about 30 per second, so 1800 per minute. The second graph is just zoomed in version of the first one. The voltage jumps are limited by protection diodes on the output of the filter, and the output settles to <50uV in less than 3min even for 20V jump with a polarity change.

Cheers

Alex