a newbie like to made 20kByte/s data sender by atmega8/16

hi friends,
i am try to learn RF and make a one-way data sender for my favorite microcontroller.
i don't like use a rf module or any all-in-one chip.
my challenge is at first for receiver part.
here is my idea:
transmitter:
base is on a pll fm transmitter for carrier and 8 predefined reference frequency for data bits 0..7 (for example: 1khz,2,4,8,16,32,64,128khz)
for each byte i will mix it's equal frequancy on carrier.

receiver:
i use a superhet to output of IF stage 2 (455k) then i use a digital cmos frequncy meter for send data to microcontroller.

is it possible? please don't laugh!
thanks a lot!

Added after 1 hours 5 minutes:

idea 2 :
can i design a radio for use pwm or usart port of micro for rf tranfer data for 200kbps in serial 1 bit row?

which idea is easy for build the rf part and lower noise and errors?

I'm not laughing - honestly!

What you propose will work but in practical terms it will be very inefficient. A similar multi-tone technique is used for low speed radio data communication in systems like RTTY, Olivia and STANAG. The top speed of these systems is usually only a hundred characters per second or so.

The problem is that you have too many stages, each dependent on the output of their predecessor. It would be much easier to use serial data rather than sending the bits in parallel as you propose. Serially, you only need two frequencies and the main carrier can be used instead of providing modulating signals of your own. Also consider that the frequencies you chose, which are easy to generate, will be prone to errors at the receiver. Any second harmonic distortion would make low order frequencies detectable as higher order bits. Trying to get 256KHz bandwidth at 455KHz center frequency just to get 128KHz tone through would be tricky too.

Using relatively narrow band FM with pre-conditioned data will give you far better throughput with MUCH simpler and more reliable circuitry.

Brian.

It sounds almost like the early radio control models - they used a different (audio) frequency for each servo (well, bang-bang control). Before cheap, lightweight electronics they used a sort of tuned reed system which detected the frequencies I think.

Anyway, I think you need to re-think your approach. As suggested, a narrow band FM would be a better approach. Even AM would be worth trying - just make sure you transmit the data serially. .

Keith.

thanks
now, i know that must make a serial data tranfer radio.
what about receiver type?
how many stage for convert?
superhet or direct coversion?
another question is about carrier frequency.
i like my radio work between walls,rooms.
also i don't like use low frequency because long antenna is needed.
in the very high frequency amplify will be difficult and semiconductors will be high price.

i like a solid state with few channels.
which frequency range do you accepted?

Added after 1 minutes:

ic MC3361 - MC3371 or MC3357 is enough solution?

It sort of depends on how much building you want to do and what parts you can get for a reasonable price. Local regulations on transmitting may also affect the choice. If I was trying to do it cheaply I would probably use a blank spot in the AM radio band and then use a normal AM radio to do the receiver part - just pull a signal from the AM demodulating diode. The antenna wouldn't be too long - use a ferrite rod, as in an AM radio.

A better design would be FM and the chips you mention would be fine for that assuming they are still available.

The other way is to look at the 433MHz or 868/915MHz bands - there are lots of chips for that sort of frequency which should be cheap and need very few external components.

Keith.

Take a look at http://www.decibit.com/index1.html where you can get 2.4GHz transceivers (transmitter and receiver on one module) for just over S8 !

They are very easy to use, have built in antennas for short range (<100m) use and easily available from many sources. The band they use is license free in most countries and they are incredibly small, about postage stamp size. Maximum speed is 2Mb/s which is much faster than you need and if you use them properly, they have the ability to automatically re-send data if it doesn't arrive safely.

Brian.

what about trasnsmitting to range 500meters.

Added after 48 minutes:

i found out RF9901-RF9902
now i need a pll for coonet to them.

You won't get 500m with any legal, license free system at those sorts of frequencies. You need to be looking at lower frequencies (<100MHz) or more power (probably illegal, even where you are).

Keith.

here i live, there is one law only : no laws!
please tell me how can i connect a pll to RF9902.
thanks

Added after 6 minutes:

if i find a pll ic ,then i must calculate and design a LPF for connect between them?
can i connect them directly?

Added after 30 minutes:

hello guys,
i found LMX2306 , that works up to 550Mhz and it is enough for me.
please help me to connect RF9902 and LMX2306.
thanks

I would look at something simpler assuming you can buy the chips where you are. Look at the Melexis TH72031
http://www.melexis.com/Assets/TH7203...heet_4806.aspx
It just needs a crystal - no synthesiser.

Or some of the Nordic VLSI products http://www.nordicsemi.com/
- similar products and again, single crystal operation.

I would pick the lowest frequency possible (maybe 433MHz) - better chance of longer range, although you will still need a good antenna to get 500m, and even then may have to add a power amp.

Keith.

i can't find another pll or synth i my city.
please someone draw and give me a schematic for connect rf9901 & LMX2206 for the range they are match.
i like make a digital solid state radio with few fixed channel.
i know calculator-websites but for me it take some days for learn about plls and loop filters and more. also i no experience about RF.

this is a handy job for someone in this site.

god bless you