What is considered a HIGH & LOW impedance?

[ANTALIFE]

Hi All

Quick question about impedances and filters. When using a T or 𝜋 or any combination of RLC filter, what is considered a HIGH or LOW impedance?

Am I correct to say that something like a OPAMP/GPIO pin... is a HIGH impedance; while a power supply, load cell... is a LOW impedance?

Also am asking this as I want to figure out when it's appropriate to use a  T/𝜋/RLC filter
https://www.emcstandards.co.uk/files/module_3_filtering_for_emc_emc11ms_v3_4_taster_jan_2021.pdf

[bob91343]

There is no specific value that makes an impedance high or low.  It depends on the situation.

For instance, with audio gear, anything 600 Ohms or lower is considered low impedance; anything in the 1000 or higher range is high.  With RF, 50 or 75 Ohms is low.  High impedance for RF is probably 450 to 600 Ohms.

In general, high impedance is very low power, such as in microphones or voltmeters.

[TimFox]

A typical audio signal connection is a low impedance source (600 ohms or less source resistance) connected to a high impedance load (say, 10 k to 1 megohm input impedance).
For such a connection, the load voltage will have negligible variation over the ranges I stated.
For audio power, the actual output impedance of the power amplifier is very low (due to negative voltage feedback from the output terminals), and the amplifier is designed to operate efficiently into a defined load impedance (e.g., 4 or 8 ohms).

[CatalinaWOW]

Just repeating what has been said in previous posts in different words so that you get another window into the meaning.

There is no set answer.  These characterizations are always meant when attaching one circuit to another though sometimes the attachment is implicit, not explicit.  If the source circuit has a significantly lower impedance than the load the output from the source will not change meaningfully when the load is attached.  But meaningfully is not clearly defined.  Go into the volt nuts zone and changes of parts per million are considered important.  In their world high impedance is defined in Megohms or even Gigohms.  In the audio world changes less than a couple of dB are usually negligible and the ratios mentioned earlier for high impedance are perfectly fine.  In the high power automotive audio world where some are attempting to draw hundreds or even thousands of Watts from a 12 volt supply low impedances are measured in tiny fractions of an ohm.

At some times and places attempts to partially define this have been made.  In some portions of the audio world 47k was the standard for high impedance, but this never became universal and now is not even widespread.

[T3sl4co1l]

For purposes of filters it's gonna be something like 50-200 ohms.  Less being "low" and more being "high".

The tradeoff is whether it's easy to get good components for some given performance -- value, Q factor, strays, physical size and cost, etc.

The threshold is ultimately defined by the impedance of free space, 377 ohms; it's easy to get impedances around here, or more usually some fraction thereof (like 50 ohm coax or 100 ohm twisted pair, where that fraction is 1/5th or thereabouts, determined by geometry of the conductors).  How much depends on geometry of the conductors.

Strays being what they are, this is a whole lot less critical at low frequencies, where very different impedances are practical (like 1 ohm, or thousands).  Until frequencies are low enough that other strays are just too annoying to deal with.  Or frequencies so high that trace capacitance and such start to take over.

And yep, impedance matters to the system the filter's a part of.  If it's driven by an opamp say, maybe it doesn't tolerate such a low impedance, and you need a filter of a kohm or more.

Also if you're using opamps, active RC filters are probably practical -- they can be done at much higher Z than LC filters, so saves on the size of capacitors, while not needing bulky/expensive inductors.  They aren't so practical in the 10s of MHz, where opamps get expensive.

Tim

[Shock]

Multimeters 10M and 1G is high, 3K is low.

[Doctorandus_P]

10G Ohm is a low impedance for an electrometer.

Resistances in the mili ohm range can be considered high for high current applications.

[TimFox]

Another example of a question where the only appropriate answer is "it depends".
A more useful answer requires more information when the question is posed.

[ejeffrey]

So the OP specifically asked about the context of passive filters when choosing between series first and shunt first topologies.

The way a filter works is that you have alternating series and shunt impedances.  In the passband the series impedances should be low and the shunt impedances high (relative to each other).  In the stop band the reverse is true.  Near the cutoff frequency they will be about equal.

For the elements in the middle this is relatively clear. For the ones at the end, they are in series or parallel with the source or load.  If the source or load is high impedance compared to the series element (an inductor for low pass filters), having a series inductor first won't do much.  If the source/load is low impedance a shunt first won't do much.

So the way to determine the answer is to figure out the source impedance and determine the size inductance or capacitor you need to have the same impedance as the source at your edge frequency.  If the capacitor is huge but the inductor is sensible you have a low impedance source and you should use a tee filter.  If the capacitor is sensible but the inductor is huge you have a high impedance and should use a pi network.

The other thing to look at is the parasitics.  What is meant by a "huge" inductor or capacitor is not just cost and size but that the parasitic shunt capacitance or series inductance dominate the behavior at your transition frequency.

[ANTALIFE]

Thanks for the help everyone, filters just got a bit clearer for me

Another example of a question where the only appropriate answer is "it depends".
A more useful answer requires more information when the question is posed.

How else are people meant to learn ;^)

[Shock]

I always find it amusing that specifications can be low but measuring high at the same time. Another is low voltage is 1000V, 30V or as close to DC as possible depending on who you are talking to.

[Doctorandus_P]

If you're designing a passive filter then both the input and output impedance become a part of the filter.

If the output impedance of what drives the filter is 100x higher then the impedances used in the filter itself, then you get an  extra attenuation of approximately 1%. Same between the impedances used in the the filter and the input impedance of the next "stage".

Often such differences can be neglected, but not always.
Sometimes the output impedance of the previous, and the input impedance of the next step are are deliberately designed into the filter. Tak for example Hifh Frequency stuff, or other circuits that work with controlled impedances.

[T3sl4co1l]

Note that a filter must have at least one terminated port, and the other can be open, short, or any other resistance.  There are different coefficients for each combination; values are tabulated in reference books.  They are not generally known/used by the common calculators, which use the equal, system impedance case (except for even order Chebyshev, which by its nature must have some impedance ratio; odd Cheb can have symmetrical or other impedances).

When a port is shorted, it must have a series branch (since, you can't parallel anything with zero ohms), and when a port is open, it must have a parallel branch (same thing in reciprocal form).

Also, odd pi (shunt C ports for LPF) tend to be preferred for practical purposes, because again, they use one less inductor than capacitor, and inductors tend to stink.

As to the matter of tee/pi, it doesn't matter so much with respect to impedance; component limitations and costs are the bigger factor.

Tim