| |
PA2AGA > HDDIG 25.09.99 02:27l 172 Lines 7892 Bytes #-9766 (0) @ EU
BID : HD_99_239E
Read: GUEST
Subj: HamDigitalDigest 99/239E
Path: DB0AAB<DB0ZKA<DB0ABH<DB0SRS<DB0MW<DB0ERF<DB0SHG<DB0SM<PI8DAZ<PI8GCB<
PI8HGL<PE1NMB<EA7URC<PE0MAR<PI8VNW
Sent: 990924/2006Z @:PI8VNW.#ZH2.NLD.EU #:2284 [HvHolland] FBB7.00g $:HD_99_239
From: PA2AGA@PI8VNW.#ZH2.NLD.EU
To : HDDIG@EU
Received: from pa2aga by pi1hvh with SMTP
id AA19764 ; Fri, 24 Sep 99 19:50:08 UTC
Received: from pa2aga by pa2aga (NET/Mac 2.3.67/7.5.3) with SMTP
id AA00015847 ; Fri, 24 Sep 99 21:07:17 MET
Date: Fri, 24 Sep 99 21:01:20 MET
Message-Id: <hd_99_239E>
From: pa2aga
To: hd_broadcast@pa2aga
Subject: HamDigitalDigest 99/239E
X-BBS-Msg-Type: B
the coast in a 50 mile strip between sea and mountains. That
makes it suitable for rail transit (which much of the country, not
constrained to a tight linear corridor, is not), and also makes it
suitable for a well designed fast coherent backbone network.
But California has neither at this time. IMHO that's mostly a matter
of lack of desire rather than insurmountable physical obstacles.
In much of the rest of the country, there are insurmountable
(with amateur level resources) physical barriers to a national
level network.
>Yes, there are some small cities, and a lot of small (very small)
>towns in between, but few of them have enough hams (or
>any hams for that matter) to support intermediate link sites.
>So we have fair network from Vancouver, BC south to Portland.
>Then it gets pretty thin for a long way. The reason for the long
>haul links is to bring places like Eugene, OR into the network.
>That took a 100 mile hop. The next hop south from Eugene
>is very tough; the transverse ranges north of the Siskyous chop
>the paths up into 10-20 mile chunks. For a couple hundred miles.
>Then you hit the CA Central valley, and all becomes easy again.
>Um ... you did have to cross Siskayou Summit and Mt. Shasta
>to get there though. If one could accomplish all the above, then
>one has a network roughly following I-5 down the coast.
>But what about inland? More big mountains in the way.
Yes, you make my case for me very well. Cross the Rockies
and picture spanning the Great Plains, the prospects for a
coherent fast national packet network become even more
bleak.
>So we use a lot of long haul to very long haul 2M 1200 baud links.
>It's that or nothing.
Yes, and for distances beyond a couple of hops, you effectively
have nothing. You're down to a couple of characters a second
net, or worse you've got constantly breaking links, and a net that's
fully congested by reconnect attempts and retries. At one time,
we had a 1200 baud net that reached from Maine to the tip of
Florida. It was virtually useless most of the time.
Gary
Gary Coffman KE4ZV | You make it |mail to ke4zv@bellsouth.net
534 Shannon Way | We break it |
Lawrenceville, GA | Guaranteed |
>.
------------------------------
Date: Wed, 22 Sep 1999 12:43:05 -0400
From: Gary Coffman <ke4zv@bellsouth.net>
Subject: Let's look at real numbers for TNC software sales
On Mon, 20 Sep 1999 20:02:33 -0700, "Hank Oredson" <horedson@att.net> wrote:
>Gary Coffman <ke4zv@bellsouth.net> wrote in message
>news:DHrmNx+s2SaofDUoViT9sz9jVUYr@4ax.com...
>> >So you are saying these modules do not need any tune up?
>> >Just plug everything together, hook to an antenna, it works?
>> >Could you be more specific about the test gear required
>> >beyond your basic solder gun, screwdriver and pliers?
>>
>> Netting the various transverters together on a common frequency
>> is the only adjustment absolutely needed. That can't realistically be
>> done by ear. You can do it by monitoring the eye pattern (with a scope)
>> of another station known to be on frequency. If one of the group has an
>> IFR or the like, you can use that without needing a priori to have one
>> transverter spot on frequency.
>
>Ok, I need a scope.
An amateur without a scope is working blind.
>> Typically, we find that the transverters are a few kHz off frequency
>> when first crystalled up, so they have to be netted. You don't have to
>> get absolutely spot on, the modems use a tracking data detector
>> that can handle up to about 5 kHz frequency error without impairing
>> BER. But we'd like to reserve that as margin for thermal drift and
>> component aging.
>>
>> It is also nice if you can monitor the eye pattern with a scope so you can
>> adjust the transverter PA drive of other systems for best linearity (and
>> monitor their received eye pattern to do the same for your PA). That'll
>> make you a better spectral neighbor, and lower BER a bit too. But we've
>> found that with the MMT transverters, turning the drive down to get about
>> 4 watts out always gives us a clean spectrum, so you can do it with just
>> a Bird. (You can't get away with that with the Hamtronics units, however,
>> for them you *really* need access to a spectrum analyzer. Nasty things
>> unless they're tuned *just* right.)
>
>Ok, so I need to buy a scope.
Yeah, but you don't need much of a scope. You need dual trace, but
even an audio bandwidth is sufficient. A suitable scope costs two
or three hundred dollars, surplus or import. (Of course you'll never
regret buying a better scope, but you don't have to spend thousands
either.)
>> No external test equipment at all is needed for the RF modems. There are
>> some switch settings to configure (if the defaults don't suit you). You
>can
>> select one of 10 frequencies (100 kHz spacing), select whether you want
>> to operate duplex or half duplex, select TTL or RS422 on the computer
>side,
>> select whether you want FEC or not, etc. But the modems come configured
>> to work right out of the box and shouldn't need any adjustments unless you
>> want something other than the default settings.
>>
>> The stuff that had to be tweaked in the first generation modems has all
>> been put into ASIC for the current modem, so there is no tune up involved.
>> You can peak the IF filter if you like, but the modem has an internal
>method
>> of doing that which doesn't require any external test equipment. You just
>> set a jumper, then tweak for max on the signal bargraph display. It should
>> be fine right out of the box, however.
>>
>> >Where does one get these various modules?
>> >What is their cost?
>>
>> Paccomm (and GRAPES) sell the W&T RF modems. The cost is $349.
>> Transverters are available from their manufacturers, or at hamfests. I
>> usually find MMT transverters for around $150 at most of the hamfests
>> I attend. They're usually crystalled for the wrong segment of the band
>> (the satellite or weak signal segment), so you need to spend $15 for a
>> new set of crystals to move them into the data bandplan. New transverters
>> range in price from under $200 to over $600 depending on make, model,
>> and band. Virtually any linear transverter will work with the modems.
>
>> You need a HDLC card in the computer too. I've personally used both
>> the PI-2 card ($98) and the Gracilis PacketTwin (don't know the current
>> cost on that one, but it is more expensive since it is full duplex DMA).
>
>DMA probably overkill (or underkill, see some of the assembler
>newsgroups for interesting measurements of DMA I/O vs. programmed
>I/O, I was surprised at the results). Modern machine ("Pentium anything")
>can easily handle many channels of 56k data. I personally do not like
>the SCC chip, but that is "hacker bigotry" and nothing else.
Yeah, you *could* bit bang one HDLC interface. But if you're running
a modern multitasking OS and have other applications that need to
run, or want to service several packet modems at a node site, then
the hardware assist is very welcome.
>> There are others. A Mac can work directly from its serial port since it
>> comes stock with chips that can be programmed to handle RS422 and
>> HDLC directly (about the only case I can think of where having a Mac is
>> an advantage).
>
>Hmmm ... my Imsai I/O chips can do HDLC ... back to CP/M ?
That's not as droll as it sounds. A dedicated packet switch could be
built up on a CP/M base. It isn't the way I'd do it, what with commodity
To be continued in digest: hd_99_239F
Read previous mail | Read next mail
| |
