On November 27, 2005 05:25 am, Olin Lathrop wrote:
> Tag changed to EE as it should have been in the first place.
>
> Jose Da Silva wrote:
> > An AGC is needed unless you do FM
>
> Not necessarily.  If it's simple binary AM modulation, AGC is only
> used for the purpose of deciding where the threshold between on and
> off should be. There is no need to actually normalize the full
> carrier on level to some preset value, only to determine what it is.=20
> This is commonly called a "data slicer" circuit.

I guess an AGC isn't needed if you are dealing with 100% on carrier and=20
100% off, but I thought the topic at this point included regenerative=20
receivers, in which case, it might be a good idea to still have a=20
signal present, say 20%, 100%  ?    now, you need an AGC.

> Of course this can also be done in software with very little
> electronics. One of my current projects is a receiver for 434MHz ISM
> band low power instruments.  The modulation scheme is simple binary
> carrier on/off at 10Kbit/second with machester encoding.  The
> previous design used a bunch of opamps and filters and comparator to
> convert the output of the RF receiver into a digital signal.  My new
> design does this all in software on a dsPIC. The dsPIC samples the RF
> signal strength at 100KHz, and the rest is software.  In the process,
> it also measures the full on RF signal strength used further
> downstream in other parts of the system.

Impressive  ;-)

quad opamp, FSK, clipped and filtered signal, sent to ordinary 6811=20
input pin, approx 10kbit too... ...designed in 1991 ;-)
opamp circuitry, less than $1, substitute a cheap pic for 1200baud,=20
$1+$1 approx $2

dsp solution starts at above $4
http://www.microchip.com/ParamChartSearch/chart.aspx?branchID=3D8071&mid=3D=
14&lang=3Den&pageId=3D75

I win!  ;-)

> > Radio passes through walls.
> > Your 100mW may still be overshadowed by a transmitter a mile away
> > pumping out 5W.
>
> Very unlikely.  *Think* about it instead of just blurting something
> out.  -snip-

Okay....   ;-)
... so 1 mile was picked out-of-the-blue as a "distance outside that=20
room", but it is a random number.  Just as random as if the transmitter=20
would be next door to the room or across the street.

> ....=A05W/100mW =3D 50 power ratio. =A0SQRT(50) =3D 7.1, ....

....it's probably safe to say the 100mW is omnidirectional, but your=20
numbers fail if the 5W is sent via a 9db gain antenna and I'm sure=20
antennas with more gain than 9db exist because a friend of mine bought=20
a 14db gain antenna for their internet connection.
It really depends on the frequency in question. Garage door openers &=20
car locks, unlikely to find directional antennas. Industrial/commercial=20
settings, probably. Satellite/microwave, definitely.

I'll ignore your blurt if you ignore my blurt.  ;-)

> Now of course in practise received signal strength doesn't go down
> smoothly with the square of the distance, but that's the general
> trend.  Obsorption of the signal by material between the transmitter
> and receiver will make this fall off even more quickly.  One study
> showed that in a hospital instead of falling off with the distance to
> the power of 2 it's more like 2.8.  Multipath distortions can cause
> hot spots and cold spots at particular locations, but these are
> essentially noise on top of the overall attenuation with distance
> trend.

If you don't want to calculate the numbers, it's basically black magic.

> > I'm not sure why you need freq stability.
>
> In a product you definitely do.  There are a lot of demands on the RF
> spectrum, so keeping everyone to their frequency is important.  In
> practice, transmitters need to be crystal based.  Good frequency
> stability and accuracy of the transmitter also allows the receiver to
> discriminate better between the signal and noise.

Good answer in reference to a transmitter, but off-topic since the=20
question was directed at the regen receiver.  ;-)

Anyhow still worth "not-snipping-as-off-topic" since it is still useful=20
for readers to keep this in mind  ;-)

> > AM is what I thought you meant from your original posting.
> > FM FSK PSK is much much better, so I'm glad you chose it.
>
> That's a pretty irresponsible statement without more qualification.

:-/

> FM is absolutely not "better" than AM in all circumstances.  As with
> everything, it's a tradeoff and each has their strong and weak
> points.

correct.

> FM circuits are more complex and therefore more expensive. =20

clip it. amplify to 0v/5v, look for crossings. cheaper than the dsp.

> It will also take more power for the same number of bits sent.  On
> the other hand, FM has more immunity to interference from weaker
> sources.  I don't remember what the OP wants to do, but both could be
> viable choices.

correct.

> >> The simplified requirements are: I need two black boxes that would
> >> let me connect two terminals 10 meters apart, and allow
> >> communication at 1200 baud. I know you guys tend to assume there's
> >> a pic on each end that I can squeeze some modem code into, but
> >> that's not the case here. :-)
>
> Ah.  You must always think of a radio link as an error-prone channel.

correct.

>  At only 10 meters and 1200 bits/second, simple circuits will work
> fine most of the time.  The only way to guarantee reliability is to
> have the bi-directional communication.  The higher level protocol
> uses ACKs and NACKs and retries until the data gets thru.  If you can
> tolerate some gaps in the data stream, then you can just use a simple
> transmitter and receiver with packets and checksums.  The receiver
> tosses out any packet that isn't right.

this was being determined at the time, so it was an unknown. if the=20
receiver can clean it up, then no smarts are needed, but if the=20
"terminal" was incapable of determining what is/isn't noise, then you=20
need to do it at the link/black-box.

:-)

--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist