| |
VK2ZRG > TECH 08.12.05 08:36l 100 Lines 5300 Bytes #999 (0) @ WW
BID : 1714_VK2ZRG
Read: DL1LCA GUEST OE7FMI
Subj: A puzzling fault
Path: DB0FHN<DB0RGB<OK0PPL<DB0RES<DK0WUE<SP7MGD<VK7AX<VK2TGB<VK2IO<VK2WI
Sent: 051208/0444Z @:VK2WI.#SYD.NSW.AUS.OC #:41 [SYDNEY] FBB7 $:1714_VK2ZRG
From: VK2ZRG@VK2WI.#SYD.NSW.AUS.OC
To : TECH@WW
VK2ZRG/TPK 1.83d Msg #:1714 Date:08-12-05 Time:4:35Z
Hello All,
My packet radio system developed a rather puzzling fault last Sunday.
I found that I couldn't up-load a file to a remote station. After sending 600
to 700 bytes, the file transfer stopped as the TX wasn't being keyed on. If
I then sent a Ctrl Z from the keyboard, the partially sent file was closed at
the remote station. If I deleted the file and then resent it, it always stopped
at the exact same byte.
I made a test file that had 12 identical lines of text and tried to send
that. Same result...always stopped at the same byte after sending 9 lines.
This proved that it wasn't something in the text that was the problem. Funny
thing was that I could send any number of lines OK if I typed them directly
from the keyboard instead of sending a file.
I could send any size file to my mailbox, up to the limit of the mailbox
size of course. And I could down-load files from the BBS OK as well. All
very curious.
I did a loop back test of the com ports, tried using a different com port
to access the TNC, a different version of TPK and another installation of
TPK on a different partition of the HDD, all to no avail. I even tried another
radio, and again, no difference.
It looked like the TNC was the cause of the problem. So I did a soft reset
of the TNC but again no change. Next I took the cover off my KPC-3 TNC and
did a hard reset with J7. After that I was only getting garbage characters
from the TNC so I thought that the TNC had finally carked it.
I then decided to use my homebrew Baycom style modem with TPK 1.83. I got
this working at the second attempt, so was back on packet. Also tried the SP6
software that I first used back in 1992. It took a bit of head scratching
and reading of the DOC file to remember how to operate SP6, I can tell you.
While doing all of this, I remembered that the KPC-3 manual mentioned
something about a baud rate change after a hard reset. I re-read the manual
and sure enough, the baud rate between TNC and PC was 1200 following a hard
reset. TPK was still set to 9600 bauds, so no wonder that I was seeing
garbage from the TNC. Anyway, after a bit of fumbling trying to enter a *
at the correct moment, I got the TNC reset to the factory default settings
and operational.
I tried sending my test file to the BBS and, joy oh joy, it up-loaded OK.
So then I went through the various TNC settings, putting them back to the
previous setting and trying my test file up-load. All went well until I
changed my mailbox size from the 5 kb default size to the maximum size
that it could be set to (24 kb). Then the original problem was back!
I found that if I set the mailbox size to 10 kb, I could send a 4.5 kb
to the BBS OK. So with this 10 kb size, I then tried up-loading a 8 kb
file to the BBS. Well, it got to just over 4.5 kb and bombed out just as
the smaller files had done earlier.
What became apparent when sending the 8 kb file, was that it was all
sent in one go, without the TNC pausing for breath, so to speak. That's when
the penny dropped; there was no control of data flow from the PC to the TNC.
I measured the supply voltages on the 14C88 RS 232 driver IC, and found
there was only -2.2 volts on the negative rail. This seemed rather low,
so I swapped the 14C88 for a MC1488 from an PC old serial port card.
(The designer of the KPC-3 had thoughtfully used a socket for the 14C88
and a 74HC14 IC that had connections to the RS 232 connector. I suppose
that he knew something like this would happen one day.)
That fixed the original problem but the MC1488 is drawing more current than
the 14C88 and the negative rail was only -2.5 volts. The supply rails for
the 14C88 come from a charge pump system running at 13.5 kHz. The schematic
of the KPC-3 doesn't show what the normal supply voltages are for the 14C88.
I would be most grateful if someone with a KPC-3 could measure these
voltages on pins 1 and 14 of U1 14C88 (with the KPC-3 connected to the PC)
and post the info on packet radio.
In the meantime, I've upped the charge pump frequency by around 6 times,
which has raised the -2.5 volts to -3.75 volts. This will have to do until
I can get a new 14C88.
One thing that I learnt from this fault is that setting the mailbox size
to the maximum isn't too clever. It reduces the TNC buffer size, which of
course, limits the number packets that the buffer can hold. I've had my
MAXFRAME number set to 7 for a long time but my packet system would only ever
send three or four 256 byte packets at one time regardless. With my mailbox
now set to 20 kb, It will send 7 packets at once and the file transfer speed
between stations is 30% greater than before.
You're probably wondering why my TNC should have failed. Well I put it
down to good old Mother Nature. There have been quite a few thunderstorms
around here in the past two weeks and one lightening strike was very close
to my QTH. (I heard the thunder almost simultaneously with the lightening
flash.) So I suppose that the 14C88 was just zapped by the lightening.
73s from Ralph VK2ZRG@VK2WI.#SYD.NSW.AUS.OC
/ack
Taglines Version 1.00, by Colin Coker G4FCN
C Program run, C Program Crash...ReWrite in Pascal!
Read previous mail | Read next mail
| |
