[aprssig] Q: Paths and the D700 and Digipeating in General (LONG)

Tue Jun 7 13:37:38 EDT 2005

noskosteve at yahoo.com wrote:

>
> I'm trying to undrstand the *basics* of the path designators.   I've 
>  had a D700 for about 2 years and some of this is starting to stick, 
> but... 
>  
> I get the WIDEn-n concept, but it is the series of XXXX,YYYY path 
> designators (aliases)  and precisely how they are handled that I have 
> fog.  If there is a clear explanation of just how packet paths are 
> handled, on the net, point me there and forget the rest of this as 
> this may all be handled by one paper or answer...  Here goes.
>  
> In excerpt 2 (below), the *Bob* recommends a  D700 digipeater path of:
>                Set APRS-UIDIGI to TEMP,WIDE1-1
> Now, it seems that the tracker should have a path of only TEMP,
> Except Bob says:
>                Set pocket tracker to use path VIA WIDE1-1,WIDE2-1
>  

The Pocket Track will have a path of "TEMP, WIDE1-1" (or perhaps TEMP, 
WIDE2-2 depending on location -- urban or rural).

The D700 that repeats the packet will have a "digipeat on" alias of only 
"TEMP".

> SO.. FIRST  basic UIDIGI stuff:
> Will a digi set for TEMP,WIDE1-1 digipeat a packet with _either_ a 
> path of TEMP  *OR*  WIDE1-1   in the first slot?
>   I think I am asking ;  Is it the case that a Digi will 'peat a 
> packet with *any* of the alias's (in the currently tested spot of the 
> packet) in its own UI list?
>  

YES    I assume that you mean in the path above "TEMP, WIDE1-1" that you 
really mean "TEMP or WIDE1-1" Digipeater ALIAS (the path name it will 
respond to) lists can include more than one alias.  These can be either 
generic "fake" callsigns like "RELAY" (no longer desirable) or 
"WIDE2-2",  geographic designators like STHMTN, or real callsigns like 
W6XYZ,.  Anything can be used as long as it conforms to the packet 
convention of up to six letters/numbers (i.e. callsign-like) optionally 
followed by "-" and one or two digits.

The hops in the path statement are processed SEQUENTIALLY.  Until the 
first hop is "used up", the second one won't be acted on.   Thus if you 
want to use your own mobile as the first hop and then a "real" digi as 
the second hop, you have to use something like TEMP, WIDE1-1     The 
D700 is a relatively dumb  digipeater. To prevent it from needlessly 
repeating other mobiles and fixed stations, you set it's digipeater 
ALIAS  to TEMP.  By then setting PocketTracker paths to start with 
"TEMP", you ensure that the 700 will only respond to PTs, and not to 
full power stations.

> Then, for a station originating packets:
> Are the aliases handled only sequentially by digis, left to right?  
> like this...
> With a path of :      XXXX,YYYY
> Is it true that this packet will ONLY be 'peated by a Digi with XXXX 
> in its list?  In other words if no XXXX Digi hears it, it dies, 
> forever lost in the ether, even if a YYYY Digi hears it?
>  

 Precisely!     The WIDE1-1, WIDE2-2 construct is a way of allowing home 
stations (acting as first step digipeaters) to take advantage of the 
"magical" dupe supressing properties of WIDEn-N type paths that "real" 
digipeaters use. The traditional home path of "RELAY" doesn't do this. 
   (Digis based on KPC3+ TNCs handle duplicate supression of WIDEn-N 
paths very effectively but fail completely on RELAY.) 

The problem arises when a  "real" digipeater hears the packet 
initially.  "Real" digis are now being set to ignore "RELAY" 
(traditionally they DID accept RELAY -OR- WIDEn-N) to reduce channel 
congestion from duplicate packets generated by inefficient handling of 
RELAY. 

The initialial WIDE1-1 is a path construct that will work with either a 
new "real" digi or with a  home station acting as a "first hop" digi  
The average home digi software or TNC doesn't know how to decrement a 
real WIDEn-N path;( ie. WIDE2-2 in should result in WIDE2-1 out). 
Instead the typical home setup with "dumb TNC will mark it as used; i.e. 
WIDE2-2 in will be repeated as "*WIDE2-2".  As a work-around,  you throw 
the home station an expendable WIDE1-1 that it will "use up" while 
leaving the following WIDE2-2 or WIDE3-3 for the smarter "real" digis.

> Then the XXXX Digi transmits it  (marking XXXX with  *) and only a 
> YYYY digi will then 'peat it?
>  
> So, Digis look for the first alias without  * and if it gets a match 
> with *any* in its list it  repeats it?
>  

YES!  Exactly!

>  
> ===  Part two  GATE question  ===
> In the HF case (1st excerpt below).  The recommended path
>                GATE,WIDE2-1
> means that ONLY a GATE will take this packet off the RF network, 
> right?______
> Then, after the GATE gets it, it will only then be digipeated by the 
> true WIDE (assuming it has the WIDEn-n software)?___
>  
>   Except,  is this from the HF side *to* 2M or  from 2M to HF -- or 
> does it matter -- ?_____
>  

YES.   This is usually done internally by dual-port TNCs like the 
Kantronics KAM which can be connected simultaneously to an HF rig at 300 
baud, and to a VHF rig at 1200. [Some software can also  simultaneously 
multiple single-port TNC on different serial ports as well.]  

However, NEVER set such a setup to gate VHF back to HF!!!
1)     The TNC will overflow and possibly lock up since the VHF traffic 
will come in potentially four times as fast as the HF side can repeat 
it,  (due to the higher baud rate).
2)      More seriously, you will be monopolizing 30M over a third or a 
half of the entire US, with almost continuous transmission of  traffic 
from the small area covered by a VHF digi or two. This will render 30M 
unusable by HF mobiles over a large part of the country.

>  
> Moving to (the final) PART 3 a related topic "Call Substitution"
> Without this, I believe each Digi that  'peats a packet puts an 
> asterisk (*) in the field of the path designator it "used" ??   Hmmmm  
> Peter piper 'peats a peck of 'peated packets...
>  
> *With* Call Substitution, we will see (within the 'peated packet) a 
> list of calls of the digis that handled this packet, right?  I think 
> the last one will have the *   I look @ the findu raw packets and next 
> hope to follow the q construct information there...'cept I don't 
> really know wht a "client" is...
>  

CLIENT = User's APRS software also acting as igate; i.e. the program 
that took the signal off RF and injected it into the Internet system.

The key element to callsign substitution is the way WIDEn-N paths are 
decremented rather than marked as used on each sequential hop:

WIDE3-3 into first digi results in WIDE3-2 out -- not marked as used up.

WIDE3-2 into second digi results in WIDE3-1 out -- not marked as used up.

WIDE3-1 into third digi results in WIDE3-0 out
    (actually it shows as just plain "*WIDE3"  ) and finally "dies" 
(i.e. no more digipeats). 

Note that on each digipeat the generic "callsign" (including the fake 
"SSID") actually changes.

Only a KPC3+  (or a TNC2 with UI-DIGI ROMs) can do this stand alone.  No 
other TNC (such as a PK-232 or MFJ with OEM firmware) can do this unless 
you use an attached computer and program to provide the n-N smarts. 

Stephen H. Smith             wa8lmf (at) aol.com

Home Page:                   http://wa8lmf.com

New APRS Symbol Chart
  http://webs.lanset.com/wa8lmf/miscinfo/APRS_Symbol_Chart.pdf   

New/Updated "Rev G" APRS     http://webs.lanset.com/wa8lmf/aprs
Symbols Set for UI-View,
UIpoint and APRSplus:

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tapr.org/pipermail/aprssig_lists.tapr.org/attachments/20050607/39754f83/attachment.html>