| |
SK1DAN > TNC 01.11.20 12:04l 261 Lines 14898 Bytes #999 (0) @ WW
BID : 4815_SK1DAN
Read: GAST
Subj: Tiny2 repair
Path: DBO595<DBX320<DBX233<VB1BOX<NL3VKL<NL3PRC<SK1BBS
Sent: 201101/0942Z 17841@SK1BBS.FRL.NLD.EU [Sneek/Snits] FBB7.00i $:4815_SK1DAN
From: SK1DAN@SK1BBS.FRL.NLD.EU
To : TNC@WW
# Generated by: TstHWin v2.21b - Registered to PA2SNK SNEEK
# On : 1-11-2020 01:29:06
# UTC: 1-11-2020 00:29:06
TNC Repair Techniques
Lessons Learned while Repairing 23 Tiny-2 TNCs
By: Burt VE2BMQ
--------------------------------------------------------------------------------
Note: This is an update to the article in a couple of Reports back about fixing
a lightning damaged PacCommTiny2 TNC. ************************
Recently, I had the "opportunity" to repair 23 Tiny-2 TNCs from the KNOX node
site that were damaged by lightning (that's right - 23 of them!!!!!). With the
help of an assistant, I was able to put every one back into working condition.
A couple of the units showed traces vaporized from the board (those needed
total IC transplants as well as replacing the vaporized traces). However most
of the TNCs did not show any visible signs of damage.
During this operation, we discovered the manufacturing defect in the Tiny-2 Mk2
version that was reported elsewhere in this Report.
The troubleshooting techniques used were pretty much the same as outlined in
the last article, except that we found that some techniques were more useful
than others. With so many units to work on, we found very quickly what
techniques were the most efficient. They were as follows:
*
Examine the circuit board very thoroughly (using a magnifying glass if
necessary) for burnt or vaporized circuits traces. Usually they will be
accompanied by black soot or copper deposit on nearby surfaces. If you find
any, then be prepared for a big job. All the TNCs that I found with blown
traces needed a majority of ICs to be replaced. Two TNCs in this lot had areas
the size of your thumb vaporized. They needed the traces replaced with wire
and ALL of the ICs plus some diodes, transistors and even the clock oscillator
in one case replaced. You may want to reconsider the repair job at this point.
My assistant proved that repair of even the worst case was possible given
enough time and perseverance. I was ready to consign it to the spare parts
pile (what spare parts - we had to replace everything!!) but he took it as a
challenge.
*
Using two known good working TNCs (one for each model of Tiny-2), we swapped
the socketed ICs one at a time from the damaged units, into the known good TNC.
If the TNC still worked, the IC was marked good and put aside. If bad, it was
discarded. That took care of the large ICs, the modem and RS-232 interfaces.
Most of the SIOs and RS-232 interface chips and about 40% of the CPU chips and
modems were bad. Most of the RAM chips were good. The fate of the EPROMs
were unknown as they had already been removed. As most of the small ICs were
soldered in, they had to wait for the next step.
*
The bad TNC is then powered from a regulated 12 volt supply with good
adjustable current limiting (fully adjustable down to zero current) and a
current meter to measure the current drain to the unit being tested. The
current limit should be set to about 300 ma and the TNC turned on. If current
is in the 20-30 ma range (see detailed current drain figures below), there is a
reasonably good chance that the remaining ICs are good. Replace the tested
good ICs into their sockets, adding new ones as required to replace the blown
chips. Turn on the TNC and check if it is working. If it is, then you have
just fixed the unit. If it is not working yet, go to step 5.
For your reference, I have measured the typical current drain of a few Tiny-2
boards and their components. Make sure you add 3 ma for each LED that is lit
on the front panel. Make sure that the board is not touching anything
conductive when measuring the current. Best to suspend the board in mid air
off the surface you are working on (I don't know why but the current went down
when I did this). To get the total current drain expected for an assembly, add
all the current values together for the components used in that assembly (don't
forget the LEDs). Don't take the resulting sum too literally, it may vary by
10% or more either way.
Typical Current Drain for Various Tiny-2 Component Sets
+------------------------------------------------------------------------------+
¦ * Component ¦ * Current Drain ¦
+------------------------------------------------------------+-----------------¦
¦ * Tiny-2 board with only small soldered ICs installed ¦ * 16 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * Tiny-2 Mk2 board with only small soldered ICs installed ¦ * 21 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * LED (lit) ¦ * 2-3 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * TCM3105 modem chip ¦ * 1-3 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * MAX231 RS-232 interface chip (no load) ¦ * 2-6 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * MAX232 RS-232 interface chip (no load) ¦ * 5-12 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * RAM chip ¦ * ¦
+------------------------------------------------------------+-----------------¦
¦ * CMOS EPROM ¦ * ¦
+------------------------------------------------------------+-----------------¦
¦ * NMOS EPROM ¦ * 8-10 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * CMOS SIO (Z84C40) ¦ * ¦
+------------------------------------------------------------+-----------------¦
¦ * NMOS SIO (Z8440) ¦ * 70-80 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * CMOS CPU (Z84C00) ¦ * 2-3 ma ¦
+------------------------------------------------------------+-----------------¦
¦ * NMOS CPU (Z8400) ¦ * 120-160 ma ¦
+------------------------------------------------------------------------------+
Even if your TNC current drain falls into the normal range, it does not
necessarily mean that the chips are all good. There can still be some low
level leakage that disturbs the logic. That will be addressed later.
*
If your TNC is drawing excessive current in the above current drain test with
just the soldered ICs left, then you likely have one or more of the ICs
defective. A good CMOS chip will draw very little current (microamps) unless
it is working at high speed. The only chips that will be working at high
speed are the clock and its divider (U7 in the Tiny-2 and U7, U22 in the Mk2
version). The problem is to find which chips are bad without having to
desolder every one. You can try logic tracing but you will likely be
frustrated very quickly. There is a simpler way.
Any current passing through a resistance will cause heating. Experience shows
that 10 ma through a CMOS chip will raise the temperature about 1 deg C (2 deg
F). Using a suitable sensitive electronic thermometer with a surface probe,
measure the temperature of each of the ICs. Any IC that is a degree or more
warmer than the board or its neighbors will be suspect. Note: a particularly
warm IC will raise the temperature of its neighbors, so test them all and
replace the warmest first, then retest if necessary. Before each test, let the
board stay powered up for 5 minutes or more.
The thermometer I used was a digital readout thermocouple meter with 0.1 deg
resolution. The thermocouple probe was a special surface measuring probe,
spring loaded to make good contact with the surface and having very low mass so
that it stabilizes very quickly. I realize that very few people would have
access to such an instrument. Alternatives would be the temperature probe
available for the Fluke series of DVMs (it works but the small contact surface
slows the response) or simply a very small thermistor on the end of a stick
with a digital ohmmeter to measure it. Some of the consumer indoor/outdoor
electronic thermometers might be suitable provided they read out to 0.1
degrees. Accuracy and linearity is unimportant. What is needed is the ability
to measure small differences in temperature on a surface.
Any ICs that are replaced at this stage should have sockets installed. Even
with the best desoldering tools/techniques, there is no point in having to do
it more than once. Use good quality sockets. See notes below about desoldering
techniques.
*
If after all this, the TNC still does not work, check all the diodes with an
ohmmeter. Also the bipolar transistors (the base-emitter junction will show
one diode drop). Replace the bad or questionable ones. FET transistors will
have to be tested in operation.
*
At this point, if the unit is still not working (we had about 90% of the lot
going at this point) then you are left with some logic tracing. The problems
we encountered at this point (on about 3 units) were mostly due to improper
logic levels due to leakage or missing clock signal. A logic probe proved
useless. We found the problems using a regular digital voltmeter. Also we had
to use a scope to track problems with the clock on two units.
For the record, we took an average time of about 1 hour per unit to do the
repairs. Some went very quickly and were working after a few minutes of
substitute testing. But others like the ones with blown traces took several
hours. The cost of replacement parts depends on where you buy them. We found
most of them at Active Components (1-800-677-8899 in the US ). Digi-Key
(1-800-344-4539) is also a possible source. If all else fails, PacComm can
furnish parts. We started with a half dozen complete sets of spare ICs for this
job. At least one complete set on hand before you start is almost a necessity.
Or at a minimum, one set of the small ICs to supplement the large ICs from a
known working TNC. A complete set of ICs might cost $20-25 US with the RAM
chip being the major cost. The most frequent defective IC was the SIO followed
by the MAX231/232, CPU, 74C14 and TCM3105 ICs.
One problem at this time and in the future is that the modem chip, TCM3105, is
no longer being manufactured. It is questionable how long it will be available
from PacComm. We may be forced to go 9600bps in order to use repaired TNCs in
the future (forcing us to go to 9600 bps is probably a good idea). For the
time being, keep any TCM3105s you have and use any that are in 4800/9600 bps
equipped TNCs you may have.
Finally just a precautionary note about desoldering of components on circuit
boards. The modern circuit boards used in TNCs today are complex precision
components. They are also very sensitive to improper repair techniques. All
of the connections use "plated through holes" and most of the traces are very
narrow. To remove a multilead component, the solder must be completely removed
from all the holes. "Solder wick" will help remove the odd component but must
be used with care. Too much pressure from a hot soldering iron on the wick
will loosen a pad or trace from the plastic base material and cause more
problems. Manual "solder suckers" are worse than useless, they will often
cause damage in unskilled hands. Other techniques include cutting (very
carefully) all the leads of an IC with a sharp diagonal cutter, a grinder or a
saw and removing the leads one at a time from the board. When working with
lightning damaged TNCs, it is not unusual to have to unsolder a half dozen
14/16 pin ICs. There just has to be a better way than solder "wick" or
mutilation of parts.
The better way is the "continuous vacuum" desoldering tool. These tools consist
of a special soldering iron with a hollow hole through the tip connected to a
solder collection chamber and a mechanical vacuum pump. Examples are the
tools made by Pace (the best in my opinion), Ungar or others. They are easy
to use, quick and most important can remove large components with little risk
to the board and the thru-holes. Although they are very expensive when new,
used units can be purchased for about the same price as a good quality (new)
solder station. They will quickly pay for themselves in reduced board damage.
If you have TNCs or similar boards to repair and don't have such a unit, search
out someone who has or encourage your local club to acquire one.
******************
This article was excerpted (with minor modifications) from the March 1997 issue
of the NEDA Report distributed to all NEDA members.
Comments or criticism should be directed to the author by e-mail, or by packet
to:VE2BMQ @ VE2FKB.#MTL.PQ.CAN.NA
We welcome technical articles on networking topics for publication. Information
should be freely disseminated and not hoarded so all can benefit from your
knowledge.
For more information about NEDA. membership, please contact the club by
e-mail to neda@dino.qc.ca or snail-mail to the NEDA address
--------------------------------------------------------------------------------
Return to Index of selected articles
or
Return to the NEDA Home Page
--------------------------------------------------------------------------------
Accesses: 8
Comments and suggestions regarding this web page should be addressed to the
NEDA Web-Master
Please identify which page you are referring to in your comments
Copyright (c) 1997 by North East Digital Association . See NEDA Home Page for
conditions.
html mark up by Burt, VE2BMQ rev: 21/03/97
Þ Þ
Þ Þ
ÞÞÞ ÞÞÞ
ÞÞÞÞ ÞÞÞÞ
ÞÞÞÞÞÞ ÞÞÞÞÞÞ
ÞÞ ÞÞÞÞÞ ÞÞ
Þ ÞÞÞ Þ
ÞÞÞÞÞÞÞ ÞÞÞÞÞÞÞ
ÞÞÞÞÞÞÞÞ ÞÞÞÞÞÞÞÞ
PA2SNK te Sneek/Snits (Friesland)
Lese vorherige Mail | Lese naechste Mail
| |