Filter between digital and analog circuits

Go To Last Post
8 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I want to add a filter between my AVR and the analog circuits. But I do not know what to use. I want it to be small and cheap. Small is most important. The filter should stop the high frequency from the AVR to reach the analog parts.

This is my options so far:
- Only resistors, maybe 1 k. (higher will decrease signal speed). The resistor should decrease the energy of the signal noise but without capacitor, will it have any effect?
- Resistor + capacitor = low pass filter.
- Multilayer chip bead. This is a wide band inductor.
- Ferrite bead array. Very easy to use as it fit the port pins well. Smallest. But how much would the ferrite do by itself? Should I add capacitors also?

I understand that there is a typical answer to this: "It depends". (On how much filter I need). But some advice would be great from what experience you might have. I will use separate ground planes and a very well filtered power supply.

My favorites:
1. My oscilloscope, Yokogawa DLM2024.
2. My soldering iron, Weller WD2M, WMRP+WMRT.
3. JTAGICE3 debugger.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Do you want to filter (isolate) power? Do you want to filter logic or control signals? Do you want to filter a PWM signal? We need a little more information.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I said power is very well filtered so obviously it is not power. Also I said that greater than 1 k would limit the speed of the signals so obviously we are talking about MHz range here.

It is SPI and control signals between the AVR and analog circuitry. I want to stop the CPU noise from reaching my amplifiers that are controlled throught SPI and some I/O.

My favorites:
1. My oscilloscope, Yokogawa DLM2024.
2. My soldering iron, Weller WD2M, WMRP+WMRT.
3. JTAGICE3 debugger.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I said power is very well filtered so obviously it is not power. Also I said that greater than 1 k would limit the speed of the signals so obviously we are talking about MHz range here.

Jim, He like to make us to think, for shure all information is obvius.. hehe
Take it easy bengtr...

I think the best options for you will be resistor + capacitor and ferrite bean + capacitor, that will depend of characteristics of the ferrite bean, inductive or resistive, that will form the L-C or R-C filter.
Use your favorite TDS3014B to see how it will perform on your circuit.

Brunomusw

Regards,

Bruno Muswieck

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jim,

I was going to ask the same questions you did. I actually had it typed out. But I decided not to do so. You see why now. :D

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Is the noise audible? That is, can you hear it? If so, then it is probably NOT microprocessor clock.

An important point that bengtr may not be aware of: having a good power supply will do little to keep power supply noise out of the audio circuits. How you deal with it depends on how the power flows.

At the very least, there should be a good capacitor or two (say a 0.1uf ceramic and a 100uf electrolytic) where the power enters the audio circuit.

Also a resistor or inductor (on the power source side of the caps) in series. The resistor might be 47 ohms or so. A ferrite bead might also work, but you need to choose one with as high an impedance at LOW frequencies as possible.

A ferrite bead in each control line might also help.

There COULD also be ground issues that are related to the processor and how external devices (signal sources and loads) are connected.

All of this is hard to provide a solution without a lot more information such as how do you detect the noise, how is your system arranged, are these circuit boards or prototype boards, and on and on!

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thank Jim and sorry if I was rude but there are just a little to many posts like that here. Many of them by good reason though. (Without mention if it was so here...)

Anyway, good advice from you as usual. But I am aware of most of what you say. What I mean with saying that power supply is well filtered is that the hole power supply system is built more or less as good as it can possible be. This is built on a 4 layer PCB with GND planes split in separate parts for analog an digitl. Also a spearate layer for power that is built in a star configuration. Only linear and PI-filter after the supplies. Every circuit of course have its own capacitor and on the star node there is a OSCON cap that have extremely low ESR. All is battery powered so the only thing I want to filter from the power line are noise from digital power reaching analog. This parts of design I have long experience from, not as long as you but 10+ years.

The noise is not in the audio range, some may be but I think there are higher frequencys as well. In this case I am looking into general advice about digital line filtering. In this subject I have not enough experience. I now the principles but I have do not know what to choose. Inductor + cap should be the best or rather R+L+C. But maybe that may be waste of space and cost? The board is small, 30x50 mm so it is very short paths. If you have split ground planes, I think you should do something of the signal lines between digital and analog as otherwise you will have a large current loop between the ground planes.

So what do you think? Will just a resistor do any good for this type of noise or should I go for a better filter?

I should mention that to measure with an active probe with very short gnd wire.

My favorites:
1. My oscilloscope, Yokogawa DLM2024.
2. My soldering iron, Weller WD2M, WMRP+WMRT.
3. JTAGICE3 debugger.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If the noise is coming in the SPI lines, and that is what I believe you are saying, I would think that some ferrite beads would do the job for you. They would just attenuate the frequencies you are having problems with, if those frequencies are well above the SPI frequency and also if those frequencies are well into the frequencies that the ferrite beads attenuate.

Depending on the speed of the SPI, you might be able to opto couple the SPI signal to the analog board, getting rid of any galvanic noise problems.

Oh, and I do hope I did not upset you with my previous post. I really did not know exactly what you meant, honest. :?

Hej då!