• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

filter design problem

Status
Not open for further replies.

solis365

New Member
Hey everyone

I've got kind of an interesting problem. I'm working with some very low frequency and low noise measurements, and we need a low pass filter with a cutoff frequency of somewhere in the range 0.1Hz to 10Hz (not really critical). I'm not really concerned with the steepness of the filter, nor the topology (chebychev, butterworth, elliptical... doesnt matter) The point of it is to filter out 60Hz harmonics noise that is coupling into the measurement lines (and other noise in the ~10Hz order of magnitude).

The trouble is, since it must be a low-noise filter, large resistors arent viable and neither are inductors (it would be best if NO inductors were used.) Also, the filter cannot be active, it must have passive components only.

I'm new to filter design and was wondering if anyone has any ideas to get around these problems.

thanks!
 
Last edited:

solis365

New Member
50 ohms for the source/load

i mean, I'm not necessarily looking for a specific design down to component values and such (if youve got one, thats great, ill try to figure out how it works), but if anyone has information, things i might want to consider, reading material (its tough to find anything talking about the types of filter i am interested in)
 

ericgibbs

Well-Known Member
Most Helpful Member
50 ohms for the source/load

i mean, I'm not necessarily looking for a specific design down to component values and such (if youve got one, thats great, ill try to figure out how it works), but if anyone has information, things i might want to consider, reading material (its tough to find anything talking about the types of filter i am interested in)
Look at this for starters.
 

Attachments

Nigel Goodwin

Super Moderator
Most Helpful Member
That pretty well rules out anything apart from an active filter.

You could try a simple capacitor across the input of the load, but it's going to need to be a LARGE capacitor, and will only give a really crappy roll off.
 

crutschow

Well-Known Member
Most Helpful Member
Why are you eliminating active filters as a solution? If you are concerned about noise, there are low noise op amps that have less noise than a 50 ohm resistor.

And why no inductors? Inductors do not contribute noise.
 

solis365

New Member
Look at this for starters.
I am familiar with bode plots, simple filter topologies, transfer functions, and amplifiers.

The trouble with this filter is that its picky, and cannot be active.

Nigel Goodwin said:
That pretty well rules out anything apart from an active filter.

You could try a simple capacitor across the input of the load, but it's going to need to be a LARGE capacitor, and will only give a really crappy roll off.
dont really care about roll off, if it starts at 1Hz corner frequency, by 60Hz its rejecting ~30dB (assuming ~20dB/decade, or a single order filter), which is plenty. by 120Hz its definitely enough rejection.

A large capacitor is probably what we're looking at. Does impedance matching matter a whole lot when you're looking at frequencies of .0001 to .1 Hz?

Do you have any feel for low noise, large-value capacitors?

the whole filter circuit will be contained in a shielded box, so hopefully noise wont couple into the capacitor itself.
 

Nigel Goodwin

Super Moderator
Most Helpful Member
A large capacitor is probably what we're looking at. Does impedance matching matter a whole lot when you're looking at frequencies of .0001 to .1 Hz?

Do you have any feel for low noise, large-value capacitors?
Where are you going to find low noise, non-electrolytic, capacitors in the 100's of uF?.

As a clue, at 20Hz the reactance of 160uF is about 50 ohms.
 

solis365

New Member
Why are you eliminating active filters as a solution? If you are concerned about noise, there are low noise op amps that have less noise than a 50 ohm resistor.

And why no inductors? Inductors do not contribute noise.
We are examining the 1/f noise spectrum of a particular FET.

we are setting the GATE voltage at a particular DC level and examining the noise spectrum at the drain. The filter is to keep noise out of the gate (particularly low frequency noise, which would then be seen in the low-freq spectrum of the DUT and invalidate the results) by introducing another active device, it will be very difficult to separate the 1/f noise contribution of one device from that of the device under test. the test is extremely low power (output on the order of femtoamps), so any 1/f noise contributed to the gate will definitely be visible on the output, which is gained up by an extremely low noise transimpedance amp.

essentially the only active devices in our test setup are the device under test and the low noise amp, which is a pretty specialized device. the filter also needs to be located near the DUT so noise doesnt couple in AFTER the filter.



and inductors generate fields. so any inductor would have to be indivudually shielded and isolated from the rest of the circuit. but it cant be far away from the rest of the filter, as then noise will couple into it.
 
Last edited:

fernando_g

New Member
---Thinking Outside the Box----

If you have 60Hz being induced or conducted into your D.U.T., could you supply the D.U.T. and lo-noise amp from batteries, and have everything inside a shielded box?

This is not so hare-brained as it seems. Once I was having hum-pickup problems feeding the phantom power on some very expensive Neumann microphones. No matter what I did, I could not completely get rid of the hum. Solved it by connecting five 9 volt batteries in series to provide the phantom power.
 

crutschow

Well-Known Member
Most Helpful Member
You may have to put the whole setup in a shielded box to remove all noise. The box should be of magnetic material (mu-metal is good) to shield the magnetic as well as electric fields.
 
Status
Not open for further replies.

Latest threads

EE World Online Articles

Loading
Top