home *** CD-ROM | disk | FTP | other *** search
- TNC PARAMETERS
- --------------
-
- (With acknowledgements to G8TIC, G0RLG and others, who first sent out this
- information)
-
- It has been recognized for some time now, that the default parameter
- settings in virtually all TNCs are far from optimum. I have been using a
- different set of parameters for over a year now, and I find that it gives a
- significant improvement.
-
- The first impression is that the TNC has gone to sleep, as it seems to wait
- long periods without doing anything. Normally, the old PTT lamp is flashing
- away merrily, and the TNC is obviously busy doing something useful, right?
- WRONG! A great deal of its activity is completely wasted, and only serves to
- clog up the frequency. Also, you are liable to frequent disconnection due to
- the TNC being too 'impatient'.
-
- I can draw two crude parallels, which might serve to get the message across.
- Have you ever been in a queue, when people start pushing from the back? The
- whole queue gets squashed into a smaller length, so those towards the back
- move forward some considerable distance, and therefore feel that they are
- making useful progress. In fact, the queue still has the same number of
- people in it, and the crushing and jamming at the front will make the queue
- move more slowly once the gates/doors etc. are finally opened. Ie it was
- completely useless activity, just like on packet.
-
- Also, imagine talking on a trans-Atlantic satellite phone link to someone,
- so that there is a slight delay. An impatient person will expect an instant
- reply to a question, and so will not give the other person a chance to reply
- (remember the delay) before 'polling' eg saying something like "hello, can
- you still hear me"? In all probability, the other person will start speaking
- at that moment, and you have a 'collision'. If this goes on long enough, the
- impatient person may get frustrated and hang up ie disconnect.
-
- So, I have found that although the TNC seems more 'lazy', far fewer of its
- transmissions are wasted ie the 'hit' rate is higher. Also, because it is
- more patient, I almost never get disconnected from the mailbox due to
- exceeding the retry count, even when things are very busy indeed. Traffic
- slows right down as the channel gets very busy at peak times, but at least
- the link holds slow but sure! At quiet periods I can download stuff from
- my local BBS just as fast as ever.
-
- If more and more users change their parameters, (and mail forwarding is kept
- off user access freqs like 144.650, 432.675 and 433.675 MHz), the situation
- would improve dramatically. Sure, there will always be antisocial 'hogs' on
- packet, but then you find the same type on HF, flattening all the opposition
- in an attempt to work the DX. The same type of person pushes into queues
- too! That is no reason for the majority to follow their example, and so I
- have avoided all temptation to change my TNC parameter settings back.
-
- OK, enough of the hard sell, here are the changes to make. Don't worry if
- you don't understand the explanations; they are there for interest only.
-
-
-
- >> MAXframe 2
- A packet sent over the air is made of separate frames, each one complete in
- itself and containing information, (eg text). You can think of it as a train
- with several carriages, each one carrying passengers. In the case of a
- 'live' user, a frame is generally one line of text, as each frame normally
- ends when RETURN (CR) is pressed, unless you're one of these people who
- forgets!
-
- By limiting the TNC to 2 frames per packet, it will ensure that even though
- you might have typed, (or uploaded off disc), many lines of text ahead of
- what has actually been sent, the TNC will never try to send more than 2 at a
- time. Short packets stand more chance of getting through without umpteen
- retries etc..
-
- >> Paclen 80
- This defines the maximum length of an individual packet frame, not the
- entire packet. The setting is rather academic, as in normal use a RETURN
- will terminate the frame before Paclen is reached. However, in case you tend
- to forget to press RETURN, or if you have a PMS facility, it is advisable to
- reduce the setting from the default 128. Do not exceed 128, as some PMSs
- have software bugs that will cause corruption of forwarded messages!
-
- >> FRack 7 (ie 7 seconds)
- After a TNC has sent a packet, it waits a certain time for a response. If no
- acknowledgement is forthcoming within that time, it will 'poll' the other
- station to check if the link is still good. The default time of 2 or 3
- seconds is far too short on a busy frequency, as the BBS or Node may have to
- wait 5 seconds or more for the channel to go quiet, before it gets a chance
- to acknowledge. Meanwhile, your TNC keeps polling impatiently, thus
- increasing the channel congestion, and risking disconnection due to
- exceeding the retry count.
-
- FRack 7 is a much more suitable setting, and even though it may appear to
- make the TNC very lethargic, it is much more efficient. This is the single
- biggest improvement you can make to the network!!
-
- BAYCOM uses 100ms units for the FRACK timer, so you should set it to 70
- rather than 7. According to the v1.5 manual, the FRACK setting should adjust
- during connection, but it is wise to have a sensible value right from the
- start.
-
- >> RESptime 30 (ie 3 seconds) This is a figure used by the TNC to calculate
- the delay before sending an acknowledgement to a packet ie it is the
- opposite to FRack. The default setting is very antisocial, and causes the
- collision of a very high percentage of packets. By increasing the setting to
- 3 seconds, you will not notice any change, but you will be helping a lot.
-
- >> PPersist ON (note the double 'P')
- Many TNCs have two sets of timing parameters, and the newer ones are more
- effective than the old ones. Therefore, by setting this ON you will enable
- the new ones. Don't worry if you get an error message in response to this
- command, eg Eh? or ?unknown command.
-
- >> PErsist 38 (only one 'P' this time)
- This is a figure used by the TNC to determine how much it tends to 'grab'
- the frequency. Ideally it is 256 divided by the number of users on the
- channel; perhaps around 25. However, due to a bug in some TNC firmware, the
- lowest you can use reliably is 38. Using a high figure is very antisocial,
- so please don't!
-
- If your TNC gives an error message in response to this command, then it is
- using old firmware. In that case, you need to enter an appropriate DWait
- setting see below.
-
- >> DWait 0 (or 16 on older firmware)
- Having set Persist 38, you should set DWait 0 in order to complete the
- enabling of PPersist. Some TNCs do not accept the PPersist ON command, but
- do accept the PErsist 38 command. In this case, setting DWAIT 0 auto-
- matically sets PPersist ON.
-
- If you have older firmware that does not accept the PErsist 38 command, DO
- NOT SET DWait TO ZERO!!! Set DWait 16 instead, otherwise your TNC will try
- and hog the frequency all to itself, and you won't be at all popular!
-
- >> SLottime 10 (or 100 with BPQ code) This is a timer value, used in
- conjunction with Persist. The 'standard' value used is 100ms, ie SLottime 10
- with most firmware, but 100 with BPQ due to different units. Some newer AEA
- firmware has a default of 30 for some strange reason. Older firmware will not
- recognize this command.
-
- >> TXdelay 23 (or as needed)
- This is the delay between the transmitter being keyed up, via the PTT line,
- and any actual useful data being sent. This is to give the TX circuits time
- to settle down, and for the squelch in the receiving station's RX to open.
- Most modern rigs can manage with typically 20 to 25, so try 23 as a starting
- figure. Even the oldest and most relay-filled rig shouldn't need more than
- 50! Using less than 20 might prevent the other stations' squelches opening
- in time, even though 15 is theoretically possible on many rigs.
-
- SUMMARY
- -------
- MAXframe 2
- Paclen 80
- FRack 7 (70 in the case of BAYCOM) ie 7 seconds
- RESptime 30 ie 3 seconds
- PPersist ON (may not work on some TNCs)
- PErsist 38 (may not work on some TNCs)
- DWait 0 (16 if PErsist 38 doesn't work)
- SLottime 10 (may not work on some TNCs) ie 100 millseconds
- TXDelay 23 (or as needed)
-
- Go on give them a whirl! Think of the warm glow of smug satisfaction that
- you will feel, by being so virtuous, hi hi. 73...
-