Help to design a passive low pass filter

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

i am using wireless (RF) to communicate with 2 mc, the rx-er using is NT-R07A (taiwan product), i only get some odd ascii code. i am advice to build a low pass filter that has a cut of frequency of 9600 Hz so i would able to get the correct characters i send.

but the basic formula of filter is f(cut off)=1/2piRC, so the RC=1.66*10^-5,

i try many diff value of R and C to match this value but still cannot works. any suggestion ?

thks for reading ^_^

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

Remember that C is measured in Farads.

There are Java calculators out there on the web. I don't have a link right now.

But do you only want a first order filter? It has a pretty slow cut-off. 2nd order would be sharper. Functionally, you could have 2 1st order filters in a row, but other topologies might be better.

Active filtering (using an Op-Amp) might allows even higher order filtering without requiring an excessively high input impedance.

-Tony

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

For 9600 Baud data you may need a little more than 9600 Hz upper frequency limit. After the filter you have an analog signal so you have to put some kind of analog to digital conversion (e.g. comperator) after the filter. This stage also has some adjustable parameters.

So you probably have to look at the recieved signal with an scope not only as a digital signal.

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

Diehardfans,

Can you tell us what XTAL frequency and AVR model are you using?
These low-cost RF Data modules have quite low RX power, so the baud rate should be 100% correct and a good antenna should be added.
Are you connecting it to computer serial port? If so, use some proper XTAL frequency in AVR to get correct BaudRate, like 1.8432MHz, 3.6864MHz, or 4.9152MHz...

When baurate is not 100% correct, this is a typical malfunction.

You do NOT need any filters, I dare to bet on it. :D

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

Quote:
i am advice to build a low pass filter that has a cut of frequency of 9600 Hz

This is not enough information to build a filter. Roll off rate, in-band and stop-band characteristics are also necessary in its design.

Can you suggest why a filter will solve your problem? It would help people to give you good guidance.

There may be other reasons for the odd ascii code. Incorrect baud rate is right up there, along with a radio protocol not suitable for your wireless link.

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

Kleinstein wrote:
For 9600 Baud data you may need a little more than 9600 Hz upper frequency limit. After the filter you have an analog signal so you have to put some kind of analog to digital conversion (e.g. comperator) after the filter. This stage also has some adjustable parameters.

So you probably have to look at the recieved signal with an scope not only as a digital signal.

You are right! I did not think of the digital nature of the modem. Edges have extremely high freqency elements (100"s of MHz).

Filtering will round-off those edges considerably. You are right that the signal will need to be "squared up" again. For 9600 baud, I'd guess that a cut-off filter would need to be above 19200Hz.

I also agree with the Broxbourne that it is likely that something else is wrong rather than a noisy signal.

-Tony

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

What distance are you working with, and are you using these devices for bidirectional communication?

Jim

If you want a career with a known path - become an undertaker. Dead people don't sue! - Kartman

Please Read: Code-of-Conduct

Atmel Studio6.2/AS7, DipTrace, Quartus, MPLAB user

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

The cut of for 9600 baud depends of how the data is modulated. If it's GMSK where the max freq is 4800 Hz 9600 is ok.
But like sulake I don't think you need a filter. It will make more phase error than it will help. First get it to work without a filter 'close' and then see if you need anything extra.

Jens

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

Redundancy or forward error correction are used to overcome noisy rf links. Send 0xAA then one char three times. Look for 0xAA, then take char that occurs 2 times. This is 4 chars per char (2400 equiv) but might give better results?

Imagecraft compiler user

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

So here goes again a lot discussed question:

Is it a "raw" RF-link or does it really have an USART interface?
I bet it is a "raw" RF-link like they usually are.

Does the datasheet say that USART data should pass the RF-link when directly fed to it, or should you use a preamble and modulate the data with Manchester etc to get it through the RF-link?

Why you need modulation? Raw RF-links should be thought as AC-coupled wires. No DC passes it. So long periods of signal being in one state will get distorted, but short periods of signal being in one state gets through quite nicely. Which means you could transmit 0x55 or 0xAA nicely but 0x00 or 0xFF would be lost or wrong.

- Jani