# Can someone explain how to use this filter calculator?

http://www.calculatoredge.com/electronics/bw%20pi%20low%20pass.htm

Cutoff Frequency: I think I understand this. If I want to pass everything below 1khz, then that is the frequency I enter in the box. Correct?

Impedance Z0: I don't understand this at all. Impedance of what? My load? How do I know the value? Is it just R = V/I for the voltage and current draw of my load?

Number of Components: I think I understand this. More components = larger, better filter. Correct?

Thanks for any help

Cut off frequency, sometimes referred to as the 3db corner point.
Enter the frequency you want pass and those below.
Impedance changes with frequency, but here it is what you are needing at the cutoff frequency.
It is made up of resistance and reactance components at cutoff.
Lots of info on the WEB.
.

Impedance Z0: I don't understand this at all. Impedance of what?

The filter output impedance. For best performance it must match the load impedance.

larger, better filter. Correct?

Larger does not necessarily mean better. The filter should be designed according to your requirements. What are those requirements?

LarryD:
Impedance changes with frequency, but here it is what you are needing at the cutoff frequency.

.

I don't think I know what I need. Will have to think about it, and try to understand how the filter works with my load.

jremington:
The filter output impedance. For best performance it must match the load impedance.

My load is just a sensor that needs a stable power supply. I've no idea what the impedance is. It needs DC, does that mean the impedance is low or high?

jremington:
Larger does not necessarily mean better. The filter should be designed according to your requirements. What are those requirements?

The requirement is to remove the ripple (caused by a piezo sounder) from the power supply for a sensor.

The requirement is to remove the ripple (caused by a piezo sounder) from the power supply for a sensor.

Those problems can usually be solved by just adding a 10 uF or 100 uF 6V capacitor across and as close as possible to the power and ground terminals of the sensor.

In more severe cases, you can add a small resistor (say 100R) or coil (10 mH) between the capacitor and the sensor power supply. Look up "power supply decoupling" for more info.

jremington:
The filter output impedance. For best performance it must match the load impedance. be as low as possible.

Impedance matching is only necessary if you want maximum power transfer. If you want maximum signal fidelity or maximum efficiency, you want low driving impedance and maximum load impedance.

Beedoo wants maximum power transfer.

jremington:
Those problems can usually be solved by just adding a 10 uF or 100 uF 6V capacitor across and as close as possible to the power and ground terminals of the sensor.

In more severe cases, you can add a small resistor (say 100R) or coil (10 mH) between the capacitor and the sensor power supply. Look up "power supply decoupling" for more info.

I already have a 10uF electrolytic capacitor close to the sensor. I found the value by trial and error, fitting different capacitors to get the best result. Unfortunately the sensor stability is still not as good as when completely removing the piezo sounder from the circuit. The best improvement so far was by adding a 100 ohm resistor in series with the piezo sounder and a 10uF capacitor across the sensor. The downside is that installing a 100 ohm resistor reduces the volume of the sounder by quite a lot. I will try your suggestion of adding a resistor or inductor to the sensor.

I had considered trying many different values of capacitors and inductors, but the number of combinations and time required to do this soon adds up. I was hoping to employ a more scientific approach of calculating what I need, and then shifting a little in either direction to see if I have the best design. Seeing as I know what frequency my piezo runs at it seemed like the obvious thing to do. However, analogue electronics is a dark art to me, so I can't do this on my own.

what sort of filter do you want?

frequency ? highpass? lowpass? bandpass? and what ripple and phase/delay response ....

that's a very wide question!

regards

Allan

The best improvement so far was by adding a 100 ohm resistor in series with the piezo sounder and a 10uF capacitor across the sensor.

That makes sense, the resistor reduces the load created by the piezo sounder, reducing the ripple on the power supply line.

Adequate power supply decoupling is what you need and a full analysis starts with analyzing the problem. Do you have access to a scope, so that you can see the ripple on the power supply line?

If not, try adding the 100 ohm resistor in series with the sensor (between the capacitor and power supply).
In more severe cases you can use a PI network: cap, resistor, cap. Here is an overview.

jremington:
Beedoo wants maximum power transfer.

I do not see anything in this thread to suggest that that is the goal.

This does suggest that you misunderstand the MPT theorem. I offer you these two problems:

For a fixed load resistance of 100 ohms, what value of source resistance will provide maximum power to the load?

For a fixed source resistance of 100 ohms, what value of load resistance will provide maximum power to the load?

The answer to these two questions are very different.

Beedoo:
I already have a 10uF electrolytic capacitor close to the sensor. I found the value by trial and error, fitting different capacitors to get the best result. Unfortunately the sensor stability is still not as good as when completely removing the piezo sounder from the circuit. The best improvement so far was by adding a 100 ohm resistor in series with the piezo sounder and a 10uF capacitor across the sensor. The downside is that installing a 100 ohm resistor reduces the volume of the sounder by quite a lot. I will try your suggestion of adding a resistor or inductor to the sensor.

If you buzzer is causing noise, how about you try and stop it at the source? Do you have any power supply filtering near the buzzer?

Take a fairly large capacitor (10 - 100 uF to start) and put it across the buzzer. You may also want to put a small value inductor in series with the buzzer's power.

allanhurst:
what sort of filter do you want?

frequency ? highpass? lowpass? bandpass? and what ripple and phase/delay response ....

that's a very wide question!

regards

Allan

Unfortunately the more questions are asked, the more ignorant I appear!

jremington:
Adequate power supply decoupling is what you need and a full analysis starts with analyzing the problem. Do you have access to a scope, so that you can see the ripple on the power supply line?

No, but I'd like a scope. If anyone knows a good PC scope kit that will work with Ubuntu then I'm all ears.

I think the ripple comes from the piezo. Certainly the sensor stability improves a lot when I disconnect the piezo.

Jiggy-Ninja:
If you buzzer is causing noise, how about you try and stop it at the source? Do you have any power supply filtering near the buzzer?

Take a fairly large capacitor (10 - 100 uF to start) and put it across the buzzer. You may also want to put a small value inductor in series with the buzzer's power.

So far I tried two things at the piezo:

1. Putting a capacitor close so that the piezo has a good source of current when the NPN on the other side of the piezo switches on.
2. Putting a diode in parallel with the piezo in case there was any back emf from the piezo.

I will try the capacitor idea, should it be electrolytic or unpolarised? What value of inductor would count as 'small'?

I have a good stock of capacitors and resistors, but I don't have any inductors so will have to try and buy something. I usually use either Maplin or ebay so hopefully there is something available from one of those.

I offer you these two problems:

@J-N: You are merely cluttering Beedoo's post with irrelevant comments.

Beedoo:
No, but I’d like a scope. If anyone knows a good PC scope kit that will work with Ubuntu then I’m all ears.

I’m not sure how good PC scopes are, but ebay has 2-channel Hantek’s for under 250 USD. 2 channels is the minimum you want, but if you have room in your budget Rigol (a good brand) has the DS1054Z for 400 USD.

I think the ripple comes from the piezo. Certainly the sensor stability improves a lot when I disconnect the piezo.

I’m certain that it does, based on your observation and the fact that it’s going to be taking huge current pulses when it switches on.

So far I tried two things at the piezo:

1. Putting a capacitor close so that the piezo has a good source of current when the NPN on the other side of the piezo switches on.

What capacity? If it’s your standard 100 nF ceramic it probably won’t be enough. What where the leads connected to? It needs to be across the supply rails, not just the piezo element, like this:

For the filter to give you the designed frequency response, both the source and load impedances must equal the value of impedance that you enter as Zo in that online calculator. Anyway it’s better to consider this as an issue of power line decoupling.

It’s very surprising if the current taken by the piezo sounder is causing the problem because the current taken by a piezo element is typically less than 4 mA. What power supply are you using? However if your sounder is powered by DC, then it must contain electronics to generate the alternating waveform required by the piezo element. I wonder if this electronics is causing the problem.

I have driven this piezo sounder directly from an Arduino Micro. It does not contain any electronics so needs to be driven with an alternating voltage. You can use a PWM “analogue” Arduino output. Use analogWrite(pin,127) for maximum volume. The frequency of the PWM does not correspond with the resonant frequency of the piezo element but the sound output level is remarkably adequate for most purposes.

This thread is a classic instance of the xyproblem: http://xyproblem.info/

You should have asked how can I stop interference from my piezo buzzer affecting my sensor readings...

Designing a linear analog filter is not the best answer to this question(!)

MarkT:
Designing a linear analog filter is not the best answer to this question(!)

Quite right it is not any sort of answer to this question.

You need to:-