[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Re: MFJ 1270B + 9600 Modem

In a message dated 10/21/98 8:31:46 PM Central Daylight Time,
kd2bd@atc106.ml.org writes:

> While were on the subject, has anyone on AMSAT-BB successfully doubled the
>  1270B's clock speed to 4 MHz and replaced the CPU, SIO, and EPROM with
>  faster components to operate at the higher clock speed?  If so, what was
>  the part number of the SIO used?  My reason for asking is because I
>  purchased a Z80 SIO/2 that does not function in the TNC, even at the 
> standard
>  clock speed.  The original component was a Z80 SIO/0.
>  When operating at an RF speed of 9600 bps, and a terminal rate of 19.2k,
>  I have observed that the 1270B occasionally drops characters when receiving
>  data from KO-23.  My thinking is that doubling the TNC's clock speed may
>  solve the problem.


I am attaching a modification that I documented on the TNC-2 compatibles
that address the serial port speed.  At the end of the attached document
you will find some Z80 part numbers/descriptions that will help you determine
the speed of the various Z80 chips.  It does not cover all of them, but it
cover the ones I have seen in the TNC-2 compatibles including PAC-COMM
and MFJ type units.

I have modified dozens of these units to work at 38.4K on the serial port.
They are capable of highert speeds, but we did not need anything higher
for our stacked nodes.  Sometimes I found that it was not necessary to
change the SIO to a higher speed as a faster CPU was all that it took.
However, I would recommend changing both to the same speed when
doing so.

I used this modification with my PAC-COMM TNC200 with a G3RUH
modem installed for the 9600 baud birds.  It works great.


Tim - N8DEU
Huntsville, Alabama

TNC-2 Compatible 38.4K baud serial port modification

Compliments: Tim Cunningham - N8DEU
Internet   : n8deu@aol.com, cunninghamt@usa.net
Date       : 26-APR-98


This modification details what is necessary to make a TNC-2 compatible
packet modem capable of a 38.4K baud data rate on the serial
communications port.  This is not a modification for the radio data
rate.  This modification has worked on MFJ1270, MFJ1270B, and PACCOM
TNC-200 units.  It should work on similar TNC-2 compatibles.

Why I chose the 4800 baud computer serial data rate switch
I chose to use the 4800 baud switch for the computer serial
communications data rate on the back of the TNC unit.  You may choose
other switches if you desire, but this one made the modification
simpler and will not affect the other switches that you may want to
use later.  Besides, if you look at the schematic, some of the other
signals going to this switch are shared with the radio data rate
switches.  Since I did not want to affect the future operation of the
300, 1200, 2400, and 9600 baud radio data rates, it made logical sense
to select the 4800 baud computer data rate switch.

The Serial data rate clock generator circuit
In order to understand how the serial data rate clock generator
affects the computer serial data rate, it is necessary to understand
the electronics on the TNC-2 compatible modem.

The original 4.9152 MHz crystal (Y1) in the circuit is the source of
the clock signal for the Z80 microprocessor and the serial
communications clock.

The output of the clock source is fed into U4 (74HCT393).  The output
of U4 generates a divide by 2, divide by 4, divide by 8, and a divide
by 16 output at pins 3, 4, 5, and 6 respectively.

This gives us output frequencies of 2.4576 MHz, 1.2288 MHz, 614.4 KHz,
and 307.2 KHz respectively on U4 pins 3, 4, 5, and 6.

Component U4 serves 2 purposes in the TNC.  One purpose is to provide
a divide by 2 function as an option to supply the CPU with a
2.4575 MHz clock signal at U4.3.  This pin is tied to JMP2 so that the
clock frequency to the Z80 CPU can be selected between the source
clock frequency of 4.9152 MHz or 2.4576 MHz at the output of U4.3.

The second function of component U4 is to provide the base frequency
to derive the radio and computer serial port data rate at U4.6.  The
output of U4.6 is fed to U1.10.

Component U1 (CD4040) takes the output from U4.6 and divides this
frequency even further to create the necessary signals to generate the
clocks required for selecting the radio and computer serial port data
rates via switch SW2.  The purpose of SW2 is simply to switch the
proper divided frequency output of U1 and route it to the proper path
for the radio and computer data rate paths.  It may be interesting to
note that the outputs from U1 represent the corresponding 16X signal
required to produce the proper radio and computer serial port data

For example, the output of U1.9 represents the 9600 baud data rate.
The frequency present at U1.9 is actually 153.6 KHz.  Since I
mentioned that it represents 16X for the data rate, then a simple
division of 16 will yield the expected 9600 baud data rate.  Also
notice that 153.6 KHz is exactly half of the frequency that is
supplied to the input of U1.10 at 307.2 KHz.  Since U1 is being used
as a divider, then you should understand that pins 9, 7, 6, 5, and 12
represent the divide by 2, divide by 4, divide by 8, divide by 16,
divide by 64, and divide by 512 functions respectively.  You may
wonder why there are a few skips in the division.  Don't worry,
because that is just the way the CD4040 component is constructed.

The output at U1.12 provides a 600 Hz sync clock source to the SIO
chip for the computer serial data rate.

Since this modification is only for the computer serial data rate,
then the rest of the discussion will focus on this topic.

I have never seen a modification that allows the serial port on a
TNC-2 compatible that allows the computer serial port data rate to
run any faster than 19.2K baud.  So, one day after looking at some
Grapes 56K baud modems we had in our HUNTSPAC (the Huntsville Amateur
Radio Club packet group) stash, I thought it was pretty silly to have
a 56K baud radio modem that was limited by its 19.2K baud computer
serial port.  That is when I pulled out the schematics and said there
has to be a faster way to do this.

If you have ever performed the 19.2K baud computer serial port
modification to your TNC-2 compatible, then you know that you just
simply take the 30.2 KHz signal at U1.10, cut a trace to one of the
pins on SW2, and add this signal so it can be switched to the SIO
(U21.27 RXTXCB signal).  Thus, you have 19.2 KHz available for use
on your computer serial port.  Of course, it was not that simple,
because you were limited by an OPAMP at U3 that was too slow to
allow for the higher data rate.  If you replaced U3 with a new TL084
component in place of the slower LM324, it usually worked.  In some
extreme cases, you had to change the jumper at JMP2 so that it was
set to change the CPU clock speed to 4.9152 MHz instead of the
2.4576 MHz setting.  This doubles the CPU clock speed.  However,
some of the early Z80 CPU and SIO chips would not work at the 5 MHz
rate, so they had to be changed to a faster component (6, 8, or
10 MHz CPU and SIO).

Now what do we do?

Just make the changes and move on with life, because you will benefit
from the new 38.4 KHz computer serial port data rate.  Yes, it can be
done and much more if you care to take the challenge.

In order to get the 19.2 KHz data rate we had to steal a higher clock
rate at U1.10 to make it work.  The same holds true to achieve the
38.4 KHz data rate, but this time we had to take it from U4.5 to get
the 614.4 KHz signal.  If you divide this by 16 you get the true data
rate of 38.4 Khz.

It may be possible to get a 76.8 KHz data rate from U4.4, but I have
not tried or tested this modification yet.

So, it is possible to get a higher data rate and yes I have tested
the 38.4K data rate using up to 4 nodes on a stack.  Believe me, it
makes a big difference when data is received in all directions from
the radio ports, because the data screams on the node stack and does
not have to wait for those 1200 or 2400 baud signals to clear on the
node ports.

OK, that is enough of this theoretical jargon.  Lets move on to what
it actually takes to make the 38.4K baud computer serial port
modification to work on your TNC-2 compatible.  If some of the
instructions are not clear, maybe I forgot a few steps, but they
should all be correct.

Step by Step Modification instructions for the TNC-2

1. Make sure you are using static protection such as a wrist strap or
   other means of static protection.

2. Turn off the TNC.

2. Remove the power plug and any other cables attached to your TNC.

4. Remove the TNC cover (usually 4 screws).

5. Remove the circuit board from the chassis.  This is usually about
   4 screws.  There may be a screw attaching the regulator to the
   chassis on the front on the TNC case. If so, you will need to
   remove this, before attempting to remove the PCB from the chassis.
   Some MFJ's and most PacComm units have a screw holding the
   regulator to the chassis.

6. Locate U1.7 (that's U1 pin number 7) on the bottom of the PCB.
   This is the side of the board where you cannot see the physical
   components.  Some refer to it as the solder side of the assembly.
   There is a trace going from U1.7 to a VIA.  Using an Exacto knife
   or a Dremel tool, cut the trace between U1.7 and the nearest
   connected VIA.

   What we did here was break the signal path between U1.7 and SW2.4,
   which was the path for the 4800 baud computer serial port
   selectable switch.

   Note: A VIA is a pad with a hole drilled through it with a barrel
   inserted in the hole that attaches a signal from one layer of the
   PCB to another layer.  If you do not understand, then you are in
   this project way over your head.  You may want to stop here and
   have somebody more qualified to finish the project for you or
   proceed at your own risk.

7. Using an Ohm Meter or similar device, measure the resistance
   between U1.7 and SW2.4 to verify that the trace has been properly
   cut in Step 6.  If you measured less than 100 ohms, then the trace
   has not been properly cut or the PCB board may be laid out
   differently than what I have ever seen (unlikely).  Do not proceed
   until the trace has been cut.

8. Now it is time for the soldering iron.  If this scares you, please
   stop and have somebody more qualified to finish the job for you.

   On the bottom of the PCB (Solder side), solder a jumper wire
   between U4.5 and the VIA we mentioned in step number 6.  The VIA
   is just a convenient place to solder the other end of the wire we
   are adding to the circuit.  If you are using a small gauge wire it
   should fit nicely inside the VIA hole for soldering.

9. On the component side of the PCB, remove U3 if it is an LM324 and
   replace it with a TL084.  This is a critical step.  If you leave
   the LM324 in place, this modification will not work and you will
   be lucky to get anything higher than 4800 baud to work properly.
   The LM324 is too slow and must be repaced.

10. Verify that the jumper at JMP2 is in the 4.9152 MHz position.
    Some people or manuals may refer to it as the 5 MHz position.

    This step can be very confusing, because I have found that some
    of the PCB layouts do not match what is stated on the schematic
    or in the TNC manual.  To be sure that JMP2 is configured
    properly, perform the following check with an Ohm Meter:

    4.9152 MHz setting
    Test for continuity between U4.1 and U10.13.  If these 2 points
    are shorted together, then JMP2 is configured for the 4.9152 MHz

    2.4576 MHz setting
    Test for continuity between U4.3 and U10.13.  If these 2 points
    are shorted together, then JMP2 is configured for the 2.4576 MHz

    If you find a short between U4.1 and U10.13 and you also find a
    short between U4.3 and U10.13, then something is terribly wrong
    with the JMP2 configuration.  This needs to be corrected and
    probably explains why your TNC has never worked properly.

    Some of the older Z80 SIO and CPU chips may not work properly
    when JMP2 is set for the 4.9152 MHz operation.  If this is the
    case, you can leave the JMP2 settings in the 2.4576 MHz position
    and hope your computer serial  port modification works at 38.4K
    baud.  I have seen several that would not work at 38.4K baud when
    the JMP2 was not configured to the 4.9152 MHz setting.  You may
    have to replace the Z80 SIO and CPU with a higher speed component
    (6 MHz or higher) to make it work.  Many of the older TNC-2
    compatibles were shipped with 4 Mhz CPU/SIO combinations.  You
    have been warned!

11. Congratulations, you have don it!  The 4800 baud computer serial
    switch on the back of the TNC is now wired for 38.4K baud
    opertaion.  So, make sure you followed all the steps (double check)
    and put your TNC back together to try it out.

Z80 CPU and SIO Identification Information
(just a few chip markngs to idenify component speed)

Z80CPU                     2 MHz
Z80BCPU      (Z8400BB1)    6 MHz
Z80 SIO/0    (Z844006PSC)  6 MHz

ZO844004PSC  (Z80 SIO/0)   4 MHz

Z80A-SIO/0-D (LH0084A)     4 MHz
TMPZ84C00AP  Toshiba       4 MHz

Z80ACPU      (Z84004)      4 MHz
Z80BCPU      (Z84006)      6 MHz
Z80HCPU      (Z84008)      8 MHz

Z84C004                    4 MHz
Z84C006                    6 MHz
Z84C008                    8 MHz
Z84C0010                  10 MHz
Z84C0020                  20 MHz

Z0840004                   4 MHz
Z0840006                   6 MHz
Z0840008                   8 MHz

These Modifications have been tested on MFJ1270, MFJ1270B, and
PACCOMM TNC-200 units.  It may work on others that are TNC-2
compatible.  If you find that it works with other TNC's not listed,
please drop me a note at one of the following addresses:


Internet: n8deu@aol.com

The author of this article cannot be held responsible for anything
resulting from this modification.  Since it is experimental in
nature you may be very satisfied with the results.

Permission is granted for copy and distribution provided it is copied
in its entirety, including this statement.

73' de Tim - N8DEU