On 9/8/2011 10:43 PM, Dave Platt wrote:
You could build a very nice full duplex repeater on a single
frequency that way.

Yes, except that the ARRL has decided to only petition for a waver for
single time slot TDMA, which can't be used for a repeater.

You're scheme would certainly work, and I too am wondering why nobody
has bothered to do it. Possibly because nobody really wants full
duplex (with echo, reverb, feedback, etc).
Not so much full duplex, but single frequency half duplex, with
negligible time delay (implying 100ms frame time) between Rx and Tx.



I wonder whether you may not also have to be really careful with your
transceiver/receiver switching design. You'll really need to be able
to trust (and drive) those PIN diodes properly... goof up on even a
single time-slice and you could put enough TX power into your receiver
to turn its front end into a pile of smouldering char in a millisecond.

This isn't a problem with normal split-frequency repeaters, thanks to
the isolation in the duplexer cans.

Do any of the commercial TDMA systems use the same frequencies for
base-mobile and mobile-base? My recollection is that TDMA cellphone
systems operate with split uplink/downlink frequencies.


Sure.. 802.11 is half duplex on a single channel, for instance.

Lots and lots of radars have fast and reliable T/R switching at pretty
much any frequency you care to name from DC to light.

TDMA cellphone uses split bands probably because it was on top of
existing AMPS systems. There is also a frequency allocation issue (e.g.
no need for new licensing). Having separate forward and reverse bands
also helps with frequency reuse and near-far issues. I hardly think
that hams are going to carpet the country with repeaters to the extent
that cell sites do.

On 9/8/2011 10:05 PM, Noskosteve wrote:
On Aug 31, 11:38 am, Jim wrote:
One wonders why someone isn't pushing for a digital TDMA scheme for
amateurs. You could build a very nice full duplex repeater on a single
frequency that way. ...

No filtering, much less intermod issues in multi station at onee site
systems... all kinds of good comes of it.

Digital schemes on HF to replace SSB I can see having real trouble (the
biggest is the lack of a "party line" capability, the other is the long
propagation delay on HF paths), but on VHF and up FM, you already have a
"one person talks at a time" by virtue of the standard FM demodulator.

Uhhh. It's been a long time since I worked on such a system (1975 I
think),
but IIR the prop delay through space for moderate distances kills the
idea.
Rough calculations gives a round trip delay, at 10 miles from the
repeater,
of about 0.1 ms. For two stations at that distance that's 0.1 ms not
available for sampling, bit width and processing.

You don't need that big a guard time if you keep track of how far you
are from the repeater and adjust your timing appropriately (that's what
a lot of systems do, and it's what was used for coarse position finding
in the phase 1 E-911 systems). That was an ordeal in 1980s to
implement, but today, it's in the piece of cake area, at least from an
implementation complexity and hardware standpoint. There is probably off
the shelf IP for it, too.


Keeping the BW down
also needs
rise and fall time as well as guard times. It added up quiclky back
then.
The vocoder becomes very important to reduce the data rate.

Yes, but on the other hand, standard cellphones use 8kbps and while the
quality isn't great, it's good enough. Of course, that gets us into that
whole "any decent codec is tied up with licensing problems" rat's nest.

On Mon, 12 Sep 2011 09:19:29 -0700, Jim Lux
wrote:

Not so much full duplex, but single frequency half duplex, with
negligible time delay (implying 100ms frame time) between Rx and Tx.

Most of the delay will not come from the flight time or mux switching.
It will come from the necessary audio compression. It won't work
without audio compression, which means that some types of
uncompressible pre-randomized data is not going to work (no big deal).
Also, the more compression, the longer the latency.

TDMA cellphone uses split bands probably because it was on top of
existing AMPS systems.

Yep. The problem was that TDMA (IS-54/IS-136) had to be compatible
with the then existing analog cell systems. Therefore, all early TDMA
phones had to offer analog compatibility. Digital only phones weren't
available until about 2003.

There's also little justification for making the change. It will not
double the number of available channels as some pundits have
suggested. Since the return audio now has to be sqeezed into the
previously transmit only channel, the number of users per channel is
cut in half. The result is no capacity change.

There is also a frequency allocation issue (e.g.
no need for new licensing). Having separate forward and reverse bands
also helps with frequency reuse and near-far issues. I hardly think
that hams are going to carpet the country with repeaters to the extent
that cell sites do.

I've looked into butchering cellular handsets into something usable on
ham radio. I have internals on some of the old Motorola flip phones
and bag phones and could probably modify the firmware sufficiently to
turn it into a conventional radios. The fatal flaw was the fixed
45MHz T/R offset. There were simply too many components that would
need to be replaced in order to operate on the smaller offset
available to hams, or on simplex. In addition, it's usually fairly
easy to go down in frequency, but the phones would require going up
from the 850MHz cellular bands to the 915MHz ham band. I gave up on
the idea.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Mon, 12 Sep 2011 09:24:00 -0700, Jim Lux
wrote:

You don't need that big a guard time if you keep track of how far you
are from the repeater and adjust your timing appropriately (that's what
a lot of systems do, and it's what was used for coarse position finding
in the phase 1 E-911 systems). That was an ordeal in 1980s to
implement, but today, it's in the piece of cake area, at least from an
implementation complexity and hardware standpoint. There is probably off
the shelf IP for it, too.

Yep. The problem was that GSM had a built in distance limit at about
35 km. Any furthur and the timing would get mangled. That was
changed with adaptive timing:
http://en.wikipedia.org/wiki/Timing_advance

Keeping the BW down
also needs
rise and fall time as well as guard times. It added up quiclky back
then.
The vocoder becomes very important to reduce the data rate.

Yes, but on the other hand, standard cellphones use 8kbps and while the
quality isn't great, it's good enough. Of course, that gets us into that
whole "any decent codec is tied up with licensing problems" rat's nest.

Codecs are incredibly important. A 1% increase in channel capacity
translates to adding thousands of additional users to a system. Nobody
uses fixed rate codecs these daze. The current fashion is variable
bandwidth schemes, such as EVRC, SMV, 4GV, etc (for CDMA). The
challenge is to get something that sounds decent with low latency, but
doesn't blow up with weak signals, high error rates, lousy SNR, etc.
When someone succeeds, it's immediately patented, creating the
predicable licensing mess.

Drivel: I once worked on a codec that required the receiving end to
have a library of the speakers phoneme sounds in storage. That
drastically reduced the amound of information that needed to be sent.
It would have worked and possibly sold, except that it was far too
easy to impersonate someone by simply switching phoneme libraries. It
was a fun project while it lasted.




--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Sun, 11 Sep 2011 16:05:44 -0700, Jeff Liebermann wrote:

On 10 Sep 2011 18:27:58 GMT, dave wrote:

Cell sites are a different animal . We were talking about 2-way,
point-to-point, VHF/UHF broadcast type sites.

Perhaps you didn't notice but several of the examples of PIM and rotten
coax induced intermod were for non-cellular systems. The problems are
much the same with any service type. If you have moderate TX power,
magnetic materials in the connectors, and sensitive receivers, PIM might
be a concern.

It's still a math problem. You can predict intermod products from known
frequencies whether the non-linear device is active or passive.

On 9/19/2011 12:48 PM, dave wrote:
[ dribble snipped ]

I see you're over here trying to sound impressive too.

Jeff


--
"Everything from Crackers to Coffins"

On Mon, 19 Sep 2011 12:48:09 -0500, dave wrote:

On Sun, 11 Sep 2011 16:05:44 -0700, Jeff Liebermann wrote:

On 10 Sep 2011 18:27:58 GMT, dave wrote:

Cell sites are a different animal . We were talking about 2-way,
point-to-point, VHF/UHF broadcast type sites.

Perhaps you didn't notice but several of the examples of PIM and rotten
coax induced intermod were for non-cellular systems. The problems are
much the same with any service type. If you have moderate TX power,
magnetic materials in the connectors, and sensitive receivers, PIM might
be a concern.

It's still a math problem. You can predict intermod products from known
frequencies whether the non-linear device is active or passive.

Yep. And after I've done the math, I still have to get rid of the
intermod. The problem is not the math. That's well known and easy to
do. The problems a
1. Finding which of the hundreds of signals found on a typical
mountain top is causing the problem.
2. Finding where the likely culprits are located (i.e. which
building).
3. Finding any and all sources of non-linearity that are producing
the mixes. That could be anything from a gold on nickel connector to
insufficient reverse power protection on a broadband power amp.
4. Site management and politics.

It's no longer single "known frequencies" causing the intermod. In
these days of broadband everything, it's fairly wide swaths of digital
noise that's causing the intermod. For example, CDMA phone is 1.25Mhz
wide, WCDMA is 5Mhz, and CDMA2000 is up to 25Mhz wide. The worst part
is that most of the culprits can't be decoded on my service monitor,
so I can't tell for sure if they're causing the intermod.


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Mon, 19 Sep 2011 14:59:53 -0500, Jeffrey Angus wrote:

On 9/19/2011 12:48 PM, dave wrote:
[ dribble snipped ]

I see you're over here trying to sound impressive too.

Jeff

I don't need to sound "impressive". I have been paid to conduct many
intermod studies using proprietary software.

On Tue, 20 Sep 2011 00:16:29 -0700, Jeff Liebermann wrote:


Yep. And after I've done the math, I still have to get rid of the
intermod. The problem is not the math. That's well known and easy to
do. The problems a
1. Finding which of the hundreds of signals found on a typical mountain
top is causing the problem.
2. Finding where the likely culprits are located (i.e. which building).
3. Finding any and all sources of non-linearity that are producing the
mixes. That could be anything from a gold on nickel connector to
insufficient reverse power protection on a broadband power amp. 4. Site
management and politics.

It's no longer single "known frequencies" causing the intermod. In
these days of broadband everything, it's fairly wide swaths of digital
noise that's causing the intermod. For example, CDMA phone is 1.25Mhz
wide, WCDMA is 5Mhz, and CDMA2000 is up to 25Mhz wide. The worst part
is that most of the culprits can't be decoded on my service monitor, so
I can't tell for sure if they're causing the intermod.

Cellular phones are a different animal. I worked on fixed and mobile,
mostly analog, mostly FM radios. Theory and practice are quite different.

The tower owner should have an inventory of every transmit and every
receive frequency, plus all the standard I.F., plus nearby external high
powered sources. The owner should have cleared each frequency before it
went on the air, and should not add a tenant if doing so would create a
harmful spur to existing users. This is site management 101.

I don't care how the WL people run their data streams. Cellular folks
don't like high mountains (except for backhaul). I know they use very
advanced techniques to hear signals below the noise floor; keeping that
noise floor as low as possible is of paramount importance when you are
looking at 100 mW devices in people's pockets 5 miles away.

FWIW, Tek has a real nice analyzer that will reverse engineer TDMA spurs.
make time-lapse spectrum analysis, and can even write on a map for you.

On Tue, 20 Sep 2011 07:16:40 -0500, dave wrote:

Theory and practice are quite different.

One day, you're going to eat those words, when you have to decide
whether to follow theory or practice. When I find that they're
different, it's usually because I'm doing something wrong. Also, if
you understand the theory, you can probably figure out the practice
(what to do). However, if you know the practice (i.e. seat of the
pants engineering), you're highly likely to fumble somewhere.

The tower owner should have an inventory of every transmit and every
receive frequency, plus all the standard I.F., plus nearby external high
powered sources. The owner should have cleared each frequency before it
went on the air, and should not add a tenant if doing so would create a
harmful spur to existing users. This is site management 101.

You almost made me spill my hot chocolate. You're correct. Site
managers should do all that. The problem is that all but one of the
site managers that I know of are business types, not engineers. They
hire engineers, tower jockeys, construction crews, and generally run
the business. It's not unusual for me to get a call or email with "I
just signed on to have [insert name] company put their radios in the
building. I'll let you know if anyone complains". This translates to
"Don't burn any billable hours doing calculations until AFTER someone
experiences interference. In short, I get paid to clean up the mess,
not to do the planning. If I want to enforce any engineering
standards, it's also done post mortem. At best, I would get an email
asking where in the building and tower I would guess the new radios
should be installed, usually without telling me the frequencies or
equipment. Interrogating the prospective new customer is something I
try to do, but often they contract out the repeater service to a comm
shop, which claims that they don't know anything because they're
afraid I might steal the customer. I don't wanna talk about
licensing, HAAT calcs, and coordination. Hopefully, your operation is
a bit closer to theory than practice.

I don't care how the WL people run their data streams. Cellular folks
don't like high mountains (except for backhaul).

Generally true. The CDMA crowd doesn't like high mountains for the
same reason they don't like CDMA operation in airplanes. The noise
floor is much higher up high and there are not enough channels
available to handle all the potential users if in a metro area.
However, they do like medium high mountain tops with fairly well
controlled coverage areas. They also like to share site ownership and
management with public agencies to reduce costs.

I know they use very
advanced techniques to hear signals below the noise floor; keeping that
noise floor as low as possible is of paramount importance when you are
looking at 100 mW devices in people's pockets 5 miles away.

100mw is about the maximum that a cell phone can belch. Power control
will usually keep that down to about 30-50mw.

FWIW, Tek has a real nice analyzer that will reverse engineer TDMA spurs.
make time-lapse spectrum analysis, and can even write on a map for you.

Well, the 20+ year old P25 radios are finally being forced into
service by FCC edict, along with various incompatible TDMA
implementations. Meanwhile, cellular is heading towards various CDMA
spread spectrum technologies (CDMA200, WCDMA, LTE, etc), which makes
TDMA look kinda dated. Anyway, I can't afford much in the way of
expensive test equipment and usually borrow or rent what I need. I
haven't actually seen a spur, mix, intermod, or noise on a spectrum
analyzer for many years as the receiver sensitivities are well below
the analyzer noise floor. Same problem with PIM (passive intermod).
It takes quite a bit of power to produce PIM making it almost
impossible to measure PIM while the xmitters are in operation. Trying
to see PIM on a spectrum analyzer is futile.



--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558
# http://802.11junk.com
# http://www.LearnByDestroying.com AE6KS