Searching \ for '[EE] Radio Synchronisation' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: techref.massmind.org/techref/index.htm?key=radio+synchronisation
Search entire site for: 'Radio Synchronisation'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Radio Synchronisation'
2006\10\13@063623 by list

flavicon
face
I'm currently working on a design with a specific problem, which I'm
hoping someone will have come across before.

Essentially I have several remote circuits, which must remain unconnected.
They all need to be synchronised within 100 Micro Seconds.

Essentially we are looking at deploying a beacon transmittor which will
send out a pulse at known intervals and have low power recievers on each
circuit which come out of hibernation at the required interval, to reset
the onboard clocks appropiatly.

We are looking at achieving 100m range (indoors), however power at the
beacons is not a significant problem (we have 11Watts available, and these
are not battery powered). Power at the recievers is crucial, each circuit
is powered by battery and must be capable of running for a minimum period
of a year. Physical battery size is not crucial but small is better :)

Does anyone have any other suggestions how to achieve the required
synchronisation, or can anyone suggest any suitable radio IC's/circuits
which could be used as I am stuggling to find anything useful via google
(must be using the wrong keywords)

Any suggestions or comments would be greatly appreciated.

Mat


2006\10\13@065756 by Jack Smith

picon face


spam_OUTlistTakeThisOuTspampiccoder.co.uk wrote:
{Quote hidden}

One common way I've seen this done is that the receiver sleep period is
relative to the last received signal.

Perhaps this will be clearer with an example. Suppose that the receiver
needs to be active for 1 second every 100 seconds. Instead of trying to
maintain a highly accurate time base so that the receiver is activated
at exactly the right time based on some initial clock setting, the
receiver strategy is to turn off for 100 seconds after it receives a
good update. This means that even with a rather poor clock on the
receiver control, the 100 second sleep period will be more than accurate
enough.

The difficult bits are deciding what to do if the receiver misses an
update. One solution would be to power up the receiver 100% of the time
until a valid signal is received. This will give the fastest resynch
time. A more power efficient, but not the fastest re-synch is (if a
valid signal is not received) to shift the sleep period slightly, say to
99 seconds. Eventually, the receiver on interval will coincide with the
transmit period and the receiver will resynch and return to the desired
100 second period.

As far as the specific receiver design goes, there are lots of ticks to
save power. A radio pager, for example, can run for a month with a
single AA alkaline cell. A good part of the power savings relates to
turning the receiver off except during a time window related to the ID
sequence.

Jack
>
>  

2006\10\13@073258 by Jinx

face picon face
part 1 4189 bytes content-type:text/plain; (decoded 7bit)

> a minimum period of a year. Physical battery size is not
> crucial but small is better :)

A nW part (eg 12F675) with the 32kHz module running will
consume around 20uA at 3.6V, => a lithium with more than
20x10-6 * 24 * 365 = 175200uAh capacity will last a year,
excluding power needed for the receiver

> Does anyone have any other suggestions how to achieve the
> required synchronisation

So, you keep the PIC running at 32kHz, which will be the
rough time-base, with an instruction cycle time of 30.5-ish
microseconds. Less than the accuracy you require from the
synch clock pulse. At the required time, which you can now
measure fairly accurately due to the crystal, turn the receiver
on long enough to pick up a pulse

> can anyone suggest any suitable radio IC's/circuits which
> could be used as I am stuggling to find anything useful via
> google (must be using the wrong keywords)

I'm quite happy with the Keymark Rx-Tx pair, as mentioned
in the recent thread [EE] Aerial for data link

Transmitter, 212kB (NB, they say 3V supply but actually 2-5V)

http://www.jaycar.co.nz/products_uploaded/ZW3100(mod).pdf

Receiver, 779kB

http://www.jaycar.co.nz/products_uploaded/ZW3102(mod).pdf

At present it's what the datasheet suggests. I have a 173mm length
of 22B&S (0.63mm) enamelled wire as a 1/4 wave at both ends.
This works fine up to at least 100m line-of-sight

Richard Prosser has made suggestions re aerials

Wire Diameter:
Will effect the Q of the antenna an also slightly effect the resonant
length. As long as you are reasonably close to the datasheet, small
changes in diameter won't really have a big effect. However, note that
the rf energy is concentrated near to the surface of the antenna so if
the diameter gets too small, the volume of conductor carrying the RF
current will decrease and the losses will increase.

Material
You want it conductive. Copper is good - but heavy. Al is slightly
less good but much lighter and may be easier to support. Magentic
materials should be avoided in general. Watch out for corrosive
problems if it is outdorrs - esp if exposed to salt spay etc.

Configuration
Basically the larger the antenna the bigger the gain possible. But
dimensions and matching become more critical, and you experiance the
law of decreasing returns. A 1/4 wave antenna on a ground plane
radiates in all directions (horizontally) but fires a fair amount of
energy up into the air. The radiation angle can be reduced by
increasing the size of the driven element (e.g a 5/8) or matching into
a 1/2 wave section etc. If you only want to send to/receive from one
direction a reflector type will work well. Yagis can give very useful
gain but may need a bit more care in setting up. There are a number of
good designs available on the internet though. For single frequency
operation you can get quite high gain if you are able to play around
with the right equipment. Don't forget that gain antennas will assist
both the transmitter and receiver - as opposed to increasing the
output power which is a transmit only option.

One type that may be worh looking at in the longer term is the
J-pole.  Omnidirectional but has useful gain over a 1/4 wave, doesn't
require a ground plane and has a reasonably low radiation angle. Can
be made using old-style TV ribbon & placed inside PVC pipe for
external protection if required (although the lengths need to modified
slightly).

http://en.wikipedia.org/wiki/J-pole

http://www.hamuniverse.com/slimjim.html

Dennis Crawley made these suggestions

For directivity, use a yagui, with a parabolic plane as reflector,
builded with 1,5mm bronze wire,(separation of elements < 1/16
lamda),... mount the receptor near the antenna.

For a omnidirectional:
Choose rigid wire or little coper tube (1/4 lamda < 17cm).
Mount the receiver in a sealed metal box and use a BNC to
connect the anntena for the experiment.

Then build a pi tank.
Make the transmitter send a continous 0xAA at the desierd
baudrate. Try to See the 0xAA with a scope on the RSSI pin,...
or linear pin. Adjust the pi tank until the pulses have a good
shape. Vary the cap values


part 2 1669 bytes content-type:image/gif; (decode)


part 3 35 bytes content-type:text/plain; charset="us-ascii"
(decoded 7bit)

2006\10\13@080006 by Jinx

face picon face
> good update. This means that even with a rather poor clock on
> the  receiver control, the 100 second sleep period will be more
> than accurate enough

That could be done with SLEEP and WDT. Quiescent should be
about 5uA. Maximum WDT period of eg the 16F88 is 268s, so
that would mean a low duty cycle for the receiver power

> The difficult bits are deciding what to do if the receiver misses an
> update

That's going to be a trade-off of WDT vs a crystal. The crystal
option uses more power, but it's stable. WDT uses less power,
but is vulnerable to Vcc and temperature changes and variation
compared to the nominal can be significant

2006\10\13@091358 by list

flavicon
face
Thanks for the email, I'm not too bothered about the PIC/uProcessor side
of things, I know I can keep track of the pulse fairly easily, and with in
the power requirements. The aerial information is very handy though. The
main issue is how to ensure that all the radio's have the same latency
recieving the pulse through the circuitry, to ensure they are all in sync.

We've come across these
http://www.rfm.com/products/data/tr1100.pdf
they look pretty good, and the price isn't too bad either, does anyone
know what sort of range can be expected from them?

Mat

{Quote hidden}

> -

2006\10\13@101217 by David VanHorn

picon face
You might look into the low power VLF receivers for WWVL.
I'm not sure if it meets your accuracy requirement, but it would remove the
need for any transmitter.



--
Feel the power of the dark side!  Atmel AVR

2006\10\13@101223 by Marcel Duchamp

picon face
Whatever scheme you end up using, keep in mind that with radio links,
you will fail to get receive signals from time to time.  Interference is
a certainty occasionally.  So whatever the function is that you are
implementing, you must allow for missed sync updates once in a while.

.....listKILLspamspam@spam@piccoder.co.uk wrote:
{Quote hidden}

2006\10\13@101245 by alan smith

picon face
if cost is not an issue, then zigbee is designed to be a very low power sensor type interface, i the only reason to have a RF link is to sync the clocks?  If your using RF to send data back, then thats another issue of each device waking at the same time and trying to transmit data back to a host.
 
 What about just putting on a RTC for each device?  Those should be very accurate?

listspamKILLspampiccoder.co.uk wrote:
 Thanks for the email, I'm not too bothered about the PIC/uProcessor side
of things, I know I can keep track of the pulse fairly easily, and with in
the power requirements. The aerial information is very handy though. The
main issue is how to ensure that all the radio's have the same latency
recieving the pulse through the circuitry, to ensure they are all in sync.

We've come across these
http://www.rfm.com/products/data/tr1100.pdf
they look pretty good, and the price isn't too bad either, does anyone
know what sort of range can be expected from them?

Mat

{Quote hidden}

>

2006\10\13@104520 by list

flavicon
face
Last time I checked RTC's weren't accurate to micro seconds, unless anyone
knows of any which are? We had considered these but decided a more active
approach would probably be needed as the sync has to be kept for over a
year.

We are only concerned about the sync at the moment and have other methods
for transmitting data around, which regretably cannot be used for syncing.

Also I have looked into zigbee and the best it can achieve is 100mS
syncing. Good suggestion though.

Mat

{Quote hidden}

>> -

2006\10\13@104555 by list

flavicon
face
thanks, we are essentially looking at a flywheel system to cope with
missing pulses, however due to the length of time it is likely to be
running, some active sync will be needed to ensure the flywheel effect is
kept on course. (In our opinion).

Mat

{Quote hidden}

> -

2006\10\13@104823 by list

flavicon
face
thanks, Ive investigated this route before, but regretably we cannot
assume that the WWVL signal is available in our area, and it most likely
will not be.

Mat


> You might look into the low power VLF receivers for WWVL.
> I'm not sure if it meets your accuracy requirement, but it would remove
> the
> need for any transmitter.
>
>
>
> --
> Feel the power of the dark side!  Atmel AVR
> -

2006\10\13@133348 by Neil Baylis

picon face
Another possibility is to put a GPS timing receiver (e.g. Trimble
Resolution-T) at each remote site. These are very easy to interface,
and will synchronize to within well under a microsecond. GPS signals
are available everywhere in the world (unless you're underground, or
something).

You don't need to have the GPS contiouously powered either. You can
use it to discipline your local clock. For example, you could power up
the GPS once per week if you like, to resynchronize your local clock.
You can also wake up the GPS if there's ever a power failure, to
reacquire the absolute time. Or if you have enough power, just leave
it running all the time. Or run it whenever there's enough sunlight
for solar power.
--
http://www.pixpopuli.com

2006\10\13@143149 by Paul Hutchinson

picon face
> -----Original Message-----
> From: piclist-bouncesspamspam_OUTmit.edu On Behalf Of @spam@listKILLspamspampiccoder.co.uk
> Sent: Friday, October 13, 2006 9:14 AM
>
> Thanks for the email, I'm not too bothered about the PIC/uProcessor side
> of things, I know I can keep track of the pulse fairly easily, and with in
> the power requirements. The aerial information is very handy though. The
> main issue is how to ensure that all the radio's have the same latency
> recieving the pulse through the circuitry, to ensure they are all in sync.
>
> We've come across these
> http://www.rfm.com/products/data/tr1100.pdf
> they look pretty good, and the price isn't too bad either, does anyone
> know what sort of range can be expected from them?

Get the "ASH Transceiver Designer's Guide"
http://www.rfm.com/products/tr_des24.pdf Section 1.3 covers operating
distance. Note, the values assume you are constrained by FCC limits for
power output, their modules can easily exceed FCC limits giving longer
ranges.

I use the TR1000 (on the DR3000 module) at work and have been satisfied with
the performance. BTW - you might want to consider the module versions of the
device, DR3300, we found that the module version simplified the application
enough to justify the higher cost.

Paul  

>
> Mat

2006\10\13@170931 by Art

picon face
There are data transmitters and receivers available from several
vendors that are data ready. They are used in short range automotive
applications, such as low tire pressure warning.

You might google for those. They use very low power and have agc so
they can be woken up by a brief high power transmitter pulse, then
can transmit data for extended periods at a much lower transmitter
power, which saves the batteries.

Some have globally unique serial numbers embedded in them which
identifies which transmitter has been activated.

Enjoy.

Art



At 10:12 AM 10/13/2006, you wrote:

>You might look into the low power VLF receivers for WWVL.
>I'm not sure if it meets your accuracy requirement, but it would remove the
>need for any transmitter.
>
>
>
>--
>Feel the power of the dark side!  Atmel AVR
>

2006\10\13@172508 by Jinx

face picon face
> The main issue is how to ensure that all the radio's have the
> same latency recieving the pulse through the circuitry, to
> ensure they are all in sync

I can comment on the Keymarks but I guess this applies to
others as well. Without any 433MHz present, the output
of the receiver is random noise, because of the AGC, and
it does take a little while (several milliseconds) for the o/p
to settle down on reception of RF

If this is also the case with whatever receiver you use, you'd
want to have the receiver powered up for a short time before
the expected pulse transmission, and the transmission itself
may need to include a preamble (like a balanced bit pattern
such as a series of 0xAA or 0x00FF) to stabilise the receiver.
You'll have to test the latency of modules for yourself

In my experience with the Keymark modules, the falling edge
of a '1' bit is within 20us of expected, ie a 1.000ms pulse
comes out of the Data pin as 1.000ms +/- 0.02. The bit-length
inacuraccy rends to be in a '0' - '1' sequence. Although the
'0' and '1' bit add up to 2ms, the rising edge in the middle is
+/- more than the falling edge

2006\10\13@180739 by Jinx

face picon face
> Last time I checked RTC's weren't accurate to micro seconds,
> unless anyone knows of any which are?

Maxim DS3231, +/- 2ppm, 0C - 40C (1 minute per 11.57 days)

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/4627

Taking 50us variation as 1ppm gain/loss and maximum drift,
the time would be out by 100us after

1,000,000 * 50us * 2 = 100s

A minutely refresh would guarantee synchronisation



2006\10\13@183948 by Art

picon face
However.......

The RTC can be tweaked in software to hold it to a much much better
long term accuracy. I forget where I read the paper, but its output
is compared to a GPS and characterized over a period of some number
of hours (and possible temperature range variations if needed).

Every so often, an extra pulse is added to speed up the clock or the
clock is inhibited for a number of clock cycles to slow down the clock.

I think someone was using the RTC's for underground earthquake
sensors in very remote locations where the ground temperate was very
constant. Complicated and power hungry oven controlled quartz
oscillators were not practical. By running a wire to the surface,
they could collect data and tweak the inexpensive RTC as needed, each
session refined the clocks accuracy until it could eventually run for
many months without any attention.

The GPS is not needed after the modules smart tweaking procedure is completed.

Don't quote me on this, but I think they ended up with less than 5
seconds of error per year and were working on a similar scheme to
tweak the RTC's clock to account of aging of the quartz and for the
change in speed of the oscillator due to voltage drop in the batteries.

Keep up the great work, this is a great list!

Art


At 06:06 PM 10/13/2006, you wrote:

{Quote hidden}

>

2006\10\14@002204 by Denny Esterline

picon face
Maybe I missed something - this sounds too easy.

Power at the transmitter is effectively unlimited, right? (o.k. 11 watts, but that's still lots) Why not just send a time signal out continuously, and let the receiver read it whenever it wakes up? The transmitted data stream could be a time packet followed by some form of sync pulse - something like $5A5A, then the reciever snychronizes with the next rising edge.

I'm more interested in why the units need that level of sync.

-Denny




{Quote hidden}

> --

2006\10\14@102245 by list

flavicon
face
Thanks, fantastic idea. It wont be possible to install GPS recievers on
every node, (will be 200+) however a couple of these in the network could
solve some of our problems. Regretably some of the locations this system
will be installed are underground (hench why WWVL wasnt an option) however
I should be able to get a GPS receiver to the surface at a handful of
locations around the network, and use this as the time sync to relay to
the other nodes.

Thanks for a fantastic idea!!

Mat


{Quote hidden}

> -

2006\10\14@102357 by list

flavicon
face
Thanks for the info, I hadnt come across remote wake up devices before, I
shall look into this further.

Mat

{Quote hidden}

>>-

2006\10\14@103254 by list

flavicon
face
Thanks, we were looking at using a preamble signal for exactly the reasons
your stated. I've looked at the Keymarks, and the range seems a little
under what we are hoping for. The accuracy you quoted is far better than I
expected though, I must have missed this in the datasheet. I shall
certainly give then another look.

Mat

{Quote hidden}

> -

2006\10\14@104614 by list

flavicon
face
Thanks, not sure I follow your calc's, I need to dig out some old notes on
ppm stuff and recheck them, but i will certainly give it more thought. I
will no doubt end up using a RTC to flywheel a signal from something like
a GPS reciever.

Mat


{Quote hidden}

> -

2006\10\14@104959 by list

flavicon
face
thanks, thats a really great idea. something I think I will no doubt end
up using. We are in a similar situation with regards to the GPS signal and
will no doubt end up have to lay a special surface cable to recover a
valid signal and repeat it throughout the network.

If anyone has any more info on this technique I would be very interested.
I will have a google around later on for it.

Mat

{Quote hidden}

>>-

2006\10\14@105414 by list

flavicon
face
Stupid me. Although particularly obvious, this had totally passed me by,
I'm so use to saving power where I can, it comes automatically to me now!!

I think this combined with a GPS reciever could be the solution to all my
problems!

As to why they need to be that synced, (thats a secret :P). I've
questioned the spec a few times, but I'm told it's necessary by the powers
that be.

Mat


{Quote hidden}

>> --

2006\10\14@122900 by Art

picon face
I'm looking for a timing GPS now.

You didn't mention your budget.

But, all of the GPS solutions require lots of power, although it's
easily accurate to within 1 us. OEM modules can be had for 25 to 70
dollars each though, so perhaps this is viable.

GL

Art





At 10:54 AM 10/14/2006, you wrote:

{Quote hidden}

2006\10\14@173937 by Jinx

face picon face
> > Why not just send a time signal out continuously, and let the
> > receiver read it whenever it wakes up?

> I'm so use to saving power where I can, it comes automatically
> to me now!!

Same here. I immediately thought "What's the best way using
the least power ?" (I'm in the middle of a battery-powered Rx-
Tx project and sure that influenced me)

It's a little bit of a wake-call about not being so narrow-minded

2006\10\14@180554 by Mat

flavicon
face
Glad its not just me, I've got this annoying instinct that I am missing
something with this at the moment, and there is an even simpler way of
achieving this.... Now if I could just put my finger on it.

Mat

{Original Message removed}

2006\10\14@180902 by Mat

flavicon
face
Budget isn't so much of an issue, sure it's a factor, but at $25-50 a unit,
this is a fraction of the overall cost, and probably easily justified. The
power and boot up times however is a bigger problem. The best solution could
possibly be a hybrid of both, using the gps on fixed nodes and relaying this
via the radio to the battery node. Still pondering this as a concept...

Mat



{Original Message removed}

2006\10\14@200058 by Bob Axtell

face picon face
Mat wrote:
> Budget isn't so much of an issue, sure it's a factor, but at $25-50 a unit,
> this is a fraction of the overall cost, and probably easily justified. The
> power and boot up times however is a bigger problem. The best solution could
> possibly be a hybrid of both, using the gps on fixed nodes and relaying this
> via the radio to the battery node. Still pondering this as a concept...
>
> Mat
>
>  
I have been busy- I think I missed most of this thread. If the task is
having a unit "check in", I have done
it with a pseudo-random generator, and have each one check in at
deliberately different times. There
will be an occasional collision, but if the number of units are few, it
works quite well. Just have the unit
check in twice as often, but have the interval quite random, and it
works well; you can't get a collision
every time..

If that wasn't the subject, then pay no attention to the man behind the
curtain.

--Bob
>
> {Original Message removed}

2006\10\14@203528 by Ray Warren

flavicon
face
>
> -----Original Message-----
> From: KILLspampiclist-bouncesKILLspamspammit.edu [RemoveMEpiclist-bouncesTakeThisOuTspammit.edu] On Behalf Of
> Jinx
> Sent: 14 October 2006 22:40
> To: Microcontroller discussion list - Public.
> Subject: Re: [EE] Radio Synchronisation
>
> > > Why not just send a time signal out continuously, and let the
> > > receiver read it whenever it wakes up?
>
> > I'm so use to saving power where I can, it comes automatically
> > to me now!!
>
> Same here. I immediately thought "What's the best way using
> the least power ?" (I'm in the middle of a battery-powered Rx-
> Tx project and sure that influenced me)
>
> It's a little bit of a wake-call about not being so narrow-minded
>
On Sat, Oct 14, 2006 at 11:05:55PM +0100, Mat wrote:

> Glad its not just me, I've got this annoying instinct that I am missing
> something with this at the moment, and there is an even simpler way of
> achieving this.... Now if I could just put my finger on it.
>
> Mat
Do the outlying units have to be synced to universal time or just to each
other? If they just need to be synced to each other then a centrally
broadcast time operating from an RTC should be sufficient and
would eliminate the added complexity of the GPS sync.
Ray Warren

2006\10\15@071347 by Mat

flavicon
face
Essentially being synced to each other is critical and being sync to
universal time would be an added bonus, which I would like to include, but
not essential.

Mat



{Original Message removed}

2006\10\15@094347 by Vasile Surducan

face picon face
11W inside of a building means suicide for the people living in the
office where the transmitter is installed.
I doubt he read the standard before claiming he can use 11W.
I even bet.

Vasile

On 10/14/06, Denny Esterline <spamBeGonefirmwarespamBeGonespamtds.net> wrote:
{Quote hidden}

2006\10\15@103135 by Jake Anderson

flavicon
face
Vasile Surducan wrote:
> 11W inside of a building means suicide for the people living in the
> office where the transmitter is installed.
> I doubt he read the standard before claiming he can use 11W.
> I even bet.
>
> Vasile
>
>  
i think he means he has 11W of power *availablle* ie from a plugin power
supply type thing

2006\10\15@111850 by Mat

flavicon
face
11W is the power supply limit, not the radio power limit! Sorry that wasn't
very clear at all, however we do have another limit of 2W transmitted energy
on the radio, amongst other legal limits I still need to work through.

Mat

{Original Message removed}

2006\10\16@002030 by Vasile Surducan

face picon face
Oh, understood now.
For 100m link inside the building, with a right antenna and an
american office bulding style (cubicles, concrete or iron structure
and a lot of glass) you need no more than 15-20dbm power at 2.4GHz.
And still don't recommend to any human living permanently near the router...

Vasile

On 10/15/06, Mat <TakeThisOuTlistEraseMEspamspam_OUTpiccoder.co.uk> wrote:
> 11W is the power supply limit, not the radio power limit! Sorry that wasn't
> very clear at all, however we do have another limit of 2W transmitted energy
> on the radio, amongst other legal limits I still need to work through.
>
> Mat
>
> {Original Message removed}

More... (looser matching)
- Last day of these posts
- In 2006 , 2007 only
- Today
- New search...