PSN-L Email List Message

Subject: Re: Digging in...
From: ChrisAtUpw@.......
Date: Mon, 15 Aug 2005 16:55:41 EDT


In a message dated 14/08/2005, gcouger@.......... writes:

Hi  Chris,

I don't think we are to the point that it either easy or cheap  to run out 
and get a wireless link for home built toys. But it time to start  thinking 
about them and including them in thoughts and  plans.
    We seem to getting nearer, slowly, but I'm not  interested in toys.

Doing a  wireless link is a large undertaking. I did the first work for 
Datalink  Systems www.rfdata.net wireless stuff and the power supplies were 
designed to  for 5 volts at 3 amps to accommodate all the various stuff we supported. 
 However 5V 1A 
would meet most installation  requirements. 
    5 watts average would require quite a large size  battery and solar cell 
array.
 
>>    Going wireless some noise problems  improve. Using long lever arms made 
of light beams can mechanically amplify  the signal but it change the noise 
problem from electronics to  mechanics.
 
    This stopped being a viable seismic technique by  the early 1970s and got 
replaced with sensors which had far better resolution.  Like down to 1/100 
nano metre. You still had photocell noise with the light  beams. Semiconductor 
lasers are very noisy as well as having divergent beams.  This was OK for 
photographic recording, but who wants to use it these days?  particularly when you 
can both filter and display a chosen digital  recording? 

Using  microwave particularly the 2.4 GHz band heavy rain and wet foliage is 
always  going to be a problem. There is probably not a worse frequency for 
rain  interference in the low GHz band than they chose for ZigBee. That can be 
over  come with higher gain antennas and more power.  
    Or you could try a hamming code.

I would  use Zigbee because it is cheap, I know it works and the data format 
is not set  in stone 

Today I can buy 60 MHz ARM computer board from New Micros off  the rack for 
$29 bucks high speed SPI or I2c A/D coverers, memory are cheap  and 30 yard 
ZigBee link that I can be programmed to use a modification of SLIP  
http://www.faqs.org/rfcs/rfc1055.html to implement a limited to store and  
forward method for capability for digitized data to insure data integrity.  Made 
from components it could be very cost effective about $100 for the  computer 
and about the same or a bit less for the solar  power.
    And about the same for the ADC and.....$$$
 
    A 12 bit ADC with noise is not much use  for a seismometer. The 16 bit 
ADCs have two bits of noise on them as supplied.  This could be averaged out, 
but it hasn't been done.

Adding  GPS for time somewhere in 
the system would cost around $50 bucks. 
    For just the receiver. Then you have to buy the  aerial and also an 
interface board. 
    The cheapest that I know about is $140 from  psn.

Less for  an old Rockwell Jupiter but that needs to be off the solar power as 
it draws  200 ma without a amplified antenna but the receiving station could 
use it to  send a heart beat time message every second to update all the 
clocks on the  wireless network. If you make if from off the shelf parts it will 
run about  500 hundred to a thousand dollars.
    I may be getting old, but I am not getting  stupid.
 
    It would be great to see timing systems available  which use WWVB and a 
trimmed quartz crystal, like on the small radio clocks.  Preferably for 
somewhere about $30. There is absolutely no requirement for  microsecond accuracy 
with 10 Hz seismic waves.
 
    Alternatively, it would be nice to see the  computer bodgers put a 
somehere near decent clock on their boards. 

Going to  UHF radios limits your band width to the IF frequency of the radio. 
 
    ?? While you can get superhet modules, many of them  just use a surface 
wave filter.

Of  course with ARM computers being a cheap as they are you can put more  
compression and computing power at the remote site than I ever dreamed  of.

You won't get any praise from me for making things  complicated. I look for 
reliability, well written software and overall  performance, but I don't see 
much of it yet. One radio comms programme requires  30 MB of RAM. With systems 
like that about, it looks like real progress is going  to be slow.
 
    One technique which I have used with success in  the past for distances 
up to 100 yds line of sight is pulsed IR optical links  with Fresnel lenses.
 
    Regards,
 
    Chris Chapman. 









In a message dated 14/08/2005, gcouger@.......... writes:
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Hi=20 Chris,

I don't think we are to the point that it either easy or che= ap=20 to run out and get a wireless link for home built toys. But it time to sta= rt=20 thinking about them and including them in thoughts and=20 plans.
    We seem to getting nearer, slowly, but I'm= not=20 interested in toys.
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Doing a=20 wireless link is a large undertaking. I did the first work for Datalink=20 Systems www.rfdata.net wireless stuff and the power supplies were designed= to=20 for 5 volts at 3 amps to accommodate all the various stuff we supported.=20 However 5V 1A
would meet most installation=20 requirements. 
    5 watts average would require quite a large siz= e=20 battery and solar cell array.
 
>>    Going wireless some noise problems=20 improve. Using long lever arms made of light beams can mechanically amp= lify=20 the signal but it change the noise problem from electronics to=20 mechanics.
 
    This stopped being a viable seismic technique b= y=20 the early 1970s and got replaced with sensors which had far better resolutio= n.=20 Like down to 1/100 nano metre. You still had photocell noise with the light=20 beams. Semiconductor lasers are very noisy as well as having divergent beams= ..=20 This was OK for photographic recording, but who wants to use it these days?=20 particularly when you can both filter and display a chosen digital=20 recording? 
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Using=20 microwave particularly the 2.4 GHz band heavy rain and wet foliage is alwa= ys=20 going to be a problem. There is probably not a worse frequency for rain=20 interference in the low GHz band than they chose for ZigBee. That can be o= ver=20 come with higher gain antennas and more power. 
    Or you could try a hamming code.
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>I would=20 use Zigbee because it is cheap, I know it works and the data format is not= set=20 in stone

Today I can buy 60 MHz ARM computer board from New Micros= off=20 the rack for $29 bucks high speed SPI or I2c A/D coverers, memory are chea= p=20 and 30 yard ZigBee link that I can be programmed to use a modification of=20= SLIP=20
http://www.faqs.org/rfcs/rfc1055.html to implement a limited to store=20= and=20 forward method for capability for digitized data to insure data integrity.= =20 Made from components it could be very cost effective about $100 for the=20 computer and about the same or a bit less for the solar=20 power.
    And about the same for the ADC and.....$$$
 
    A 12 bit ADC with noise is not much u= se=20 for a seismometer. The 16 bit ADCs have two bits of noise on them as supplie= d.=20 This could be averaged out, but it hasn't been done.
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Adding=20 GPS for time somewhere in
the system would cost around $50 bucks.=20
    For just the receiver. Then you have to buy the= =20 aerial and also an interface board.
    The cheapest that I know about is $140 from=20 psn.
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Less for=20 an old Rockwell Jupiter but that needs to be off the solar power as it dra= ws=20 200 ma without a amplified antenna but the receiving station could use it=20= to=20 send a heart beat time message every second to update all the clocks on th= e=20 wireless network. If you make if from off the shelf parts it will run abou= t=20 500 hundred to a thousand dollars.
    I may be getting old, but I am not getting=20 stupid.
 
    It would be great to see timing systems availab= le=20 which use WWVB and a trimmed quartz crystal, like on the small radio clocks.= =20 Preferably for somewhere about $30. There is absolutely no requirement for=20 microsecond accuracy with 10 Hz seismic waves.
 
    Alternatively, it would be nice to see the= =20 computer bodgers put a somehere near decent clock on their boards.  <= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Going to=20 UHF radios limits your band width to the IF frequency of the radio.=20
 
    ?? While you can get superhet modules, many of=20= them=20 just use a surface wave filter.
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Of=20 course with ARM computers being a cheap as they are you can put more=20 compression and computing power at the remote site than I ever dreamed=20 of.
    You won't get any praise from me for making thi= ngs=20 complicated. I look for reliability, well written software and overall=20 performance, but I don't see much of it yet. One radio comms programme requi= res=20 30 MB of RAM. With systems like that about, it looks like real progress is g= oing=20 to be slow.
 
    One technique which I have used with succes= s in=20 the past for distances up to 100 yds line of sight is pulsed IR optical link= s=20 with Fresnel lenses.
 
    Regards,
 
    Chris Chapman.
 

[ Top ] [ Back ] [ Home Page ]