Date: Mon, 6 Sep 93 04:30:07 PDT
From: Advanced Amateur Radio Networking Group <tcp-group@ucsd.edu>
Errors-To: TCP-Group-Errors@UCSD.Edu
Reply-To: TCP-Group@UCSD.Edu
Precedence: Bulk
Subject: TCP-Group Digest V93 #229
To: tcp-group-digest
TCP-Group Digest Mon, 6 Sep 93 Volume 93 : Issue 229
Today's Topics:
More than 3 packet drivers ? (2 msgs)
Problems with pa0gri 2.0p
RE PI CARD PROBS SOLVED
scc&hi speed packet , part 1/2
scc&hi speed packet , part 2/2
Subscribe
Send Replies or notes for publication to: <TCP-Group@UCSD.Edu>.
Subscription requests to <TCP-Group-REQUEST@UCSD.Edu>.
Problems you can't solve otherwise to brian@ucsd.edu.
Archives of past issues of the TCP-Group Digest are available
(by FTP only) from UCSD.Edu in directory "mailarchives".
We trust that readers are intelligent enough to realize that all text
herein consists of personal comments and does not represent the official
policies or positions of any party. Your mileage may vary. So there.
----------------------------------------------------------------------
Date: Mon, 6 Sep 1993 10:31:40 +0930 (CST)
From: Rob Mayfield <mayfield@guest.adelaide.edu.au>
Subject: More than 3 packet drivers ?
To: tcpgroup@ucsd.edu
Hi All,
Just wondering if anyone has had success with 5 or more packet drivers
in some of the later nos versions (jnos 1.10x?). Ive managed to get what
looks like a working compile, but previously have had stability
problems. The user wants 5 ports on bpq, and doesnt want 2 images of nos
running (for obvious reasons).
I'd be interested in hearing from anyone who has this running. We have
an application where they guy wants 7 drivers, but we'll work on the 5
first :-)
73's ... Rob
--
rob mayfield vk5xxx/vk5zeu po box 46, henley beach, south australia, 5022
packet radio vk5xxx@vk5xxx.#adl.#sa.aus.oc wire +6183487713w +6182341007h
i-net mayfield@wattle.itd.adelaide.edu.au or xtasc@levels.unisa.edu.au
------------------------------
Date: Sun, 5 Sep 93 23:06:11 -0400
From: Jim De Arras <jmd@cube.handheld.com>
Subject: More than 3 packet drivers ?
To: Rob Mayfield <mayfield@guest.adelaide.edu.au>
> Hi All,
>
> Just wondering if anyone has had success with 5 or more packet drivers
> in some of the later nos versions (jnos 1.10x?). Ive managed to get what
> looks like a working compile, but previously have had stability
> problems. The user wants 5 ports on bpq, and doesnt want 2 images of nos
> running (for obvious reasons).
I have 5 packet drivers working, no problem! You just need to edit 3-4 files
to increase the amount available. The .h file alone is not enough, you have to
add the needed copies to the .asm and the .c files.
>
> I'd be interested in hearing from anyone who has this running. We have
> an application where they guy wants 7 drivers, but we'll work on the 5
> first :-)
>
> 73's ... Rob
> --
> rob mayfield vk5xxx/vk5zeu po box 46, henley beach, south australia, 5022
> packet radio vk5xxx@vk5xxx.#adl.#sa.aus.oc wire +6183487713w +6182341007h
> i-net mayfield@wattle.itd.adelaide.edu.au or xtasc@levels.unisa.edu.au
>
>
Jim
------------------------------
Date: Sun, 5 Sep 93 12:23:13 CST
From: invitado@speedy.coacade.uv.mx (Cuenta Publica)
Subject: Problems with pa0gri 2.0p
To: tcpgroup@ucsd.edu
Hi!! I installed the pa0gri nos with a jnos configuration yesterday, but when
i try to do an 'attach asy' command the program hangs. Somebody knows what
happens?? I am using pa0gri 2.0p . The configuration works fine with the
wg7j code.
Also i have problems connecting two machines via the internal slip drivers.
I can do it, but only with the clarkson packet drivers. I cannot connect
two machines by a dialup connection by slip,i can only by ppp. Somebody knows
how to do it ???
Thanks for any response
Roman
Roman Torres
Programmer
invitado@speedy.coacade.uv.mx
------------------------------
Date: Sun, 5 Sep 93 11:00:00 -0600
From: karl.ebel@drig.com (Karl Ebel)
Subject: RE PI CARD PROBS SOLVED
To: TCP-GROUP@UCSD.EDU
hello all,
thanks for all your replies. All of the ideas are indeed
potential problem areas.
We followed the advice from Dave:
120 ohms across jumper W6 at DDIOW position
220pf from U2 PIN 6 to ground.
It works perfectly with a 386DX-40MHz board. And it still
works with the former inuse 486DX-25MHz board.
Thanks to Dennis for his inside description of the problem.
NOS in a DOS-Box of OS2 V2.1 is still crashing. We are
studying the problem.
So thanks to all helpers from the group:
DF5KO, DJ8JG, DL5KAW and me DK8GD.
73 de Karl.
+-------------------------------------------------------------+
| Internet ..: karl.ebel@drig.com |
| AMPRNET ...: charly@dk8gd.ampr.org 44.130.32.1 |
| BBS Mail ..: dk8gd@db0iz |
+-------------------------------------------------------------+
It works with a 386DX-40MHz. And it
still works with the forme inuse board 486DX-25MHz.
Components are 120 ohms and 220pf.
Thanks from the group:
DF5KO, DJ8JG, DL5KAW and me DK8GD.
73 de Karl.
+--------------------------------------------------------------+
| Internet ..: karl.ebel@drig.com |
| AMPRNET ...: charly@dk8gd.ampr.org 44.130.32.1 |
| BBS Mail ..: dk8gd@db0iz |
+--------------------------------------------------------------+
------------------------------
Date: Sun, 05 Sep 93 08:42:25 UTC
From: pa3aou@pa3aou.ampr.org
Subject: scc&hi speed packet , part 1/2
To: tcp-group%pa2aga@pi8ehv.UCSD.EDU
Hi Folks
I recently compiled a small report on SCC card modifications
for hi speed packet . Guess you may be interested .
Split it up in 2 parts , here is part 1
SCC card modification for use with hi speed baudrates
-----------------------------------------------------
1- Purpose:
To improve performance of time response scc card so that
speeds up to 38400 baud can be used
Exchange of experience and knowledge on hi speed packet with
scc card .
2- Introduction:
The pa0hzp scc card is standard equipped with 8530 scc chips
and performs very well with baudrates 4800bd - 9600 bd .
However some mod's are needed to increase usable speeds up
to 38400 .
This document is not a spic and span upgrade manual but
describes problems and modifications personally
encountered .
If you want to have the best , buy the most recent state of the art
stuff somewhere .
If you want to get the most out of you exsisting scc card ,
and have fun , keep on reading .
3-Areas of improvement
3.1- Optocouplers
Optocouplers are usable up to 9600 bd or so but cause problems
at higher speeds . This due to optocoupler response but also
cable capacitance . Using standard 10 wire shielded 'serial port'
cable with a length of 3 meters did not work at 38k4 .
Shorten this cable to a minimal length ( 30 cm or so ).
Experiments with flatcable may be usefull too ( didn't try)
Also optocouplers have been removed at the hi speed ports.
A 74LS00 in the 38k4 modem now directly drives the TTL ports
on the scc card .
For this, the txd and rxd tracks on the scc card, going
to/fro the modem were cut and directly connected to the
74ls14 gates .
Even better is to use a DIL solderable adapter which can be
inserted in the IC socket . Keeps the scc pcb original .
Modify your modems so that it is TTL driven ( remove the
resistors etc ) .
As a consequence, from now on modems may only be disconnected
with the power on ALL parts of your system shut down !
Also connect the return path ; mount a wire from the 'earth'
connection at the modem side of the 37 pens D shell connector
to the earth on the scc card .
Verify with an Ohm meter ( to pen 7 of any 74ls14 ) that it is
the earth and not the +5 Volts line ....
Doublecheck all mod's made since there is a real risk of
blowing your scc or pc now ....
3.2- (E)SCC chip
The SCCchip ( 8530 ) is a standard chip used for synchronous
datacommunication ( packet ) as the 16450 is standard used
for asynchronous communication ( serial port ) .
Both generate interrupts to your CPU for every character received ..
So if the serial port receives a block of data with lenth of 1024
bytes , 1024 interrupts are generated
Any other activity in the pc generating interrupts during reception
of this data block may cause that characters are missed .
With synchronous transfer , this leads to eg Rx Overruns and
Tx underrruns .
Solution : use of fifo buffers .
The 85230 from AMD or Zilog incorporates 8 byte receive and 4 byte
transmit fifo buffers .
This allows data to be stored in the fifo so that the cpu is able
to fetch the datablock complete when returning from another task .
( compare with improvement replacing 16450 with 16550 usart ) .
Installation ; just remove the 8530 from the scc card and insert
the 85230 ( zilog or amd ) . Observe static discharge precautions .
You may replace only one or both chips if you like .
The escc is supported with net from pe1chl
Use a recent version( 921018lt.exe or upwards should support the
escc ) .
In the autoexec.net the ' attach scc ' must be replaced with
' attach escc '
for those channels equipped with the 85230 .
Tests run on a 40 Mhz 386 gave no space or overruns at 19k2 but
did do so on 38k4 .
After 'attach escc ' no overruns or space errors anymore
( tested with 921018lt.exe )
Recently also the SCC driver for Linux ( pe1nnz ) was adapted on an
experimental basis to support escc, first tests reduced tx overruns from
roughly 300 / Megabyte back to nil ( 0 ) or occasionally one
( tested at 38k4 ). RX improvements still under construction .
Also NOS may support escc, a version made by pe1mnl also
supports escc on experimental basis ( is not the final version yet,
same trick used as with the scc driver for linux )
End of part 1
----------------
73 de harry pa3aou@pi8hvh
PLEASE reply to the list, NOT to the From: address
because this mail is sent through a one-way gateway!
------------------------------
Date: Sun, 05 Sep 93 08:43:19 UTC
From: pa3aou@pa3aou.ampr.org
Subject: scc&hi speed packet , part 2/2
To: tcp-group%pa2aga@pi8ehv.UCSD.EDU
Improvements for use of scc card on hi speed packet part 2 :
------------------------------------------------------------
3.3- Clock speed SCC card
Increasing the clock speed on the scc card will also
improve the response .
The clock speed must be a multiple of 2 in order to generate
the proper baudrates . With an 8530 usually a 4.9152 Xtal is used .
Using the 85230 escc will allow to use higher clock speeds .
The amd am85c230a is specified to run on 8.192 MHz scc clock speed .
This requires both chips on the scc to be escc .
Also modify the clock speed given in the ' attach escc '
statement used to initialise the scc card .
Maybe it is possible to directly mount a 8.912 xtal , I didn't
test this .
I used parts ' available' which happened to be a 19.660800 MHz
oscillator module .
It is mounted at the place of the 4.912 MHz quartz .
__________
| 1 2 | top view : x = dot
| | 1 = + 5 Volts
| x 3 | 2 = TTL output
---------- 3 = earth
Before, remove the 4.9 MHz quartz as well as R20,21 ( 1 kOhm ) and
C 1 ( 1 nF ) .
Cut the track from the oscillator output going to the 74LS368 .
Cut the track coming from the ESCC chips at pen 20 ( clck ) going
to the 74LS368 .
An extra 74LS93 is mounted near the edge connector ( empty socket)
The 14 pens IC socket is mounted closest so that pen 7 is free
and pen 14 at the +5 Volts position .
Before inserting the IC, lift pen 14 !! .
Unlike most TTL IC's , the 74LS93 does NOT have its +5Volts
supply on pen 14 ..........
Connections as follows:
pen 1 : connected to pen 12
pen 2 : to earth
pen 3 : to earth
pen 5 : to +5 Volts
pen 10 : to earth
pen 14 : input from Oscillator block
Assuming frequency = input from oscillator block on 74LS93 pen 14 ,
available output signals are now :
Qa = pen 12 = frequency /2
Qb = pen 9 = frequency /4
Qc = pen 8 = frequency /8
Qd = pen 11 = frequency /16
At present the scc runs on 4.9 Mhz ( Qb ) since only one ESCC
is present .
Make sure to specify this frequency in the 'attach scc '
initialisation .
Performance improvement here unknow yet ..
Note , use 74 LS 93 , the 74 L 93 has a different pin layout ...
For sure you may want to use another IC , no problem . Just
observe that it will run on the higher frequency ( cmos may be
insufficient )
74LS74 ( dual D-type flip flop ) , 74LS290 or 74LS293
( binairy counters ) are some alternatives .
Be creative and check your databook for exact pen layout !
3.4- Stop unwanted interrupts
Depending on the transceiver you have the Rx part may produce
noise even when no carrier is received .
The FT290 for instance leaves the receiver on when it transmits .
Often I can then trace my transmitted packets .
Another set miraculously produces extra noise in the lf receiver
partduring transmit .
As result Rx interrupts are generated during transmit
Kill those interrupts !
Use an extra 74 LS 00
original added
.............. --------------------------------------------------
74 hct 132 o to rest of original part
__ ___ |
PTT-|--| | | o rx data from modem
|--| |o-| |
| | | ----- | ------ ------
--- | 1| | | 5| | 9| |
|--| N1 |3 --| N2 |6 |--| N3 |8
|--| * |o------| * |o---|--| * |o---R 1---o
2| | 4| | 10| | to scc card
----- ------ ------ rx data
+5 Volts at pen 14
ground at pen 7
* = added 74 LS 00 gates
R 1 = 470 Ohm for scc cards with optocoupler
discard R 1 for direct TTL driven scc card ( modified )
The modem cannot be used for fullduplex operation after this change !
The circuit can be simplified by offering a PTT/NOT directly to
pen 4 of N2 ( low state when transmitting ) and
leaving out N3 ( not needed in for AX25 HDLC coding , usually NRZI )
I just put them in for level conversion to / from TTL .
3.5- Check your PC
Not only focus on the scc and software .
Doublecheck your PC setup .
Experiments with reducing wait states to 0 and increasing
ISA AT bus speed may cause troubles or improve
the overall performance .
Check it ! It's worth trying .
4- Thanks
Thanks to those helping me with supplying material en spiritual
support .
If you find bugs in this article , send it to
ax25: pa3aou@pi8zaa.#nbo.nld.eu
smtp: pa3aou@pi8ehv ( local ) or pa3aou@pi8hvh
73 de Harry
3-9-93
PLEASE reply to the list, NOT to the From: address
because this mail is sent through a one-way gateway!
------------------------------
Date: Mon, 06 Sep 93 06:50:48 GMT
From: ian@g3nrw.demon.co.uk (Ian Wade)
Subject: Subscribe
To: tcp-group@ucsd.edu
Subscribe
+===============================+========================================+
| Ian Wade | E-mail: ian@g3nrw.demon.co.uk |
| DOWERMAIN LTD | [158.152.16.226] |
| 7 Daubeney Close, Harlington, | AX.25: G3NRW @ GB7BIL.#27.GBR.EU |
| Dunstable, Beds LU5 6NF, UK | AMPRnet: g3nrw.ampr.org [44.131.5.147] |
+===============================+========================================+
.
------------------------------
End of TCP-Group Digest V93 #229
******************************
******************************