IBM 3420 User manual

-----.J
..
~
~------1 ------1
------1
------1 ------1
-----.J
-.:......J ~
~
' ------1
------1
----.J
------1 ------1
-----.J
----
"---./
-----!l: .
~ ------1
------1
,
.........:..-:;
--...!..:.J ' ~ ------1 ~
,
I
~
------1 ----'-....J
•-----"--..J
------1
~ ------1 ------1
------1
• __ -..I
------1
~ ------1
------1
------1
------1
,.
I
------1
~
------1
-----------
"----
----1 -----./ -----.J
----1 -----./
-----.J
---
~
----.J -----J
.---l
----1 ----1
......:........
---~
---------
----.J
~
----.J ----.J
~ ---./
~ ~
----.J ~
-' -'
~ ---./ ----.J ----.J
---.--
~
-----0.
----./
----.J
-'
----./
-' -'
~
-' -'
----./
-'
----.J ----.J ~ ~ ----.J
----./
----.J ----.J
----.J ~ ~ ~
---------
---------
'--------'.
.'
========:=~~ ~
--J ----.J ----.J --'
---.J ---'
---.J ---'
~
---'
---.J
---'
---'
---' ---'
---'
---.J
--' --' ---' ---'
---
---' ---'
---"
----,
---'
---'
-----l
---' ---'
---'
---'
---' ---' ---' ---'
---'
--' ---' ---'
~
---' ---' ---' ---'
---'
----'
---' ---'
----'
........:....,
---'
----'
---'
---'
----'
---' ---'
----'
---'
---'
----'
---'
---'
-----
-----
-----
-----
,
----
------ ,
~~ /
-----' '
-----'
-----'
-----'
,
,
~ ---J
-----'
-----'
-'
---J
.---' -----' -----'
-----'
--
~ .
-----'
~ -
~
-----'
---=
-----' -----' -----' -----' -----'
-----'
-----' -----'
-----' -----'
-----' -----'
-----' -----'
IBM 3420/3803
Magnetic Tape Subsystem Reference Guide

'PREFACE
,.This reference~guide and associated video presentation were created by FE Service Plan-
'. 'ning and Support to assist you in maintaining the 3420/3803 Magnetic Tape Subsystem.
Two approaches were taken:
L', New information is included as a result of IR data indicating the areas where more
in-depth knowledge
is needed.
2. Currentinformation, which
is frequently used, has been condensed from the MLM/ :
Pathfinder and included here.
The guide
is divided'into four major sections. Each of these sections contains informa-
tion which logically fits into that section. Take the time to glance through the guide and
get an idea
of what is included.
Pay particular attention to the "NTF Service Checks" portion of each area. It is intended
..
to give you a list of actions to be performed which may help eliminate many call-back
situations.
When you write a no trouble found IR (major unit 960) try to use unit/cause
codes which best describe the reported failure symptom so that FE SP/S may have a
better understanding
of areas of the machines with intermittent problems.
THIRD EDITION (January 1980)
This is a major revision of, but does not make S229-6019-1
obsolete.
Address any comments concerning the contents
of thls publication to: IBM Field
Engineering Service Planning, Department 96C,
BUilding 005, Tucson, Arizona 85744.
© Copyright International Business Machines Corporation 1978, 1980

CONTENTS
GENERAL
3420 Subsystem Characteristics. . . . .
3803 Model 3 and 3803 Modell Differences
Recording Philosophy . . . . . . . .
Understanding EREP1. . . . . . . .
Procedure for Scoping 3420 Tape Unit Sense Bits on Logic Board
3420/3803 Status/Sense Bytes ~ . . . .' . . .
3803-2/3420-3Tiuough 8 Tape Drive Sense Atialysis. .
3803/3420 Online Tests
What
is Where in the MLM?
Common Abends. . . .
Tools and Test Equipment
MEDIA
, IRD PEP Specialist
Magnetic
Tap~
CommOl1 Media Related Problems
3420 "
Read/Write Service Techniques
Read/Write Failure Analysis .
Read/Write Head Service Hints .
3420 Read/Write NTF Ch~ck1ist .
Tape Developing Procedure
Developed Tape Samples . . .
3420 RD/WR Data Flow . . .
,3420 Tape Motion Service Techniques. .
3420 Tape Motion NTF Checklist .
3420 Tape Motion . : . . .
3420 Power Service Techniques
342Q Power NTP Checklist
, 34:20 Power - DC Voltages .
3420/625,0- Conversion-Tips .
34 20 PM Schedule, . . .'
3420 Sales Feature Codes, .
3420 (All Models) Status Identifier.
3420 ECA Checklist ." .
Models3, 5, 7 - AI'
BO'ard ,. .
Models 4, 6, 8 -
Al Board. .
.Card
Jump~rs -Models 3, 5, 7
Card Jumpers -Models 4,6,8 '
'Co1l1mon 3420 Part Numbers.
'FlafBelt Pneuf!1atic Supply
.,
3
5
9
17
19
20
27
31
35
39
45
47
51
53
61
63
65
67
69
71
73
77
79
83
85
87
91
95
99
~
101
103
117
119
, 121
125
129
131

CONTENTS (continued)
3803
3803 Modell Data Flow
3803 Model 2 Data Flow
3803 Microprocessor Instruction Format.
3803 Microprocessor Tips. . . . . .
Current ALU Patches . . . . . . .
Current 3420/3803 Related Software Zaps
3803 Channel Interface . . . . .
3803 DC Voltages . . . . . . .
3803 Mod 1 to 2 - 6250 Conversion Tips
3803 Conversion Tips from Data Bank
3803 Preventive Maintenance Schedule
3803 Sales Feature Codes. . .
3803-1 Feature Identifier . . .
3803-1 Feature Installation Times .
3803-2 Feature Installation Times
3803 ECA Checklist. .
Bibliography . . . .
Index . . . . .
Reader's Comment Form
iv
139
141
143
145
149
153
155
163
165
169
173
175
177
179
. 183
185
197
199

" GENERAL
""----- - --

3420 SUBSYSTEM CHARACTERISTICS
Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
Tape Speed (Read orWrite)
(ips) 75 75 125 125 200 200
(cm/sec) 190,5 190,5 317,5 317,5 508 508
6250 Read Access Time, nominal*(ms) 2.3 1.6 1.1
1600 Read Access Time, nominal*(ms) 4.0 4.0 2.9 2.6 2.0 1.65
6250 Write Access Time, nominal*(ms) 2.1 1.5 0.95
1600 Write
Access Time, nominal*(ms) 4.0 3.0 2.9 2.0 2.0 1.28
Forward Start Time, nominal**(ms) 1.8 1.4 1.4
1.1 1.3 .08
Data Rates (Kb/sec;Kd/sec):
6250 bpi 470/940 780/1560 1250/2500
1600 bpi
PE
120/240 1201240 200/400 200/400 320/640 320/640
800 bpi NRZI (9-Track) 60/120 100/200 160/320
800 bpi NRZI (7-Tra.ck) 60 100 160
556 bpi NRZI (7-Track) 41.7 69.5 111.2
200 bpi NRZI (7-Track) 15.0 25.0 40.0
Passing Times per Byte (usec):
6250 bpi 2.133 1.28 0.80
1600 bpi
PE
8.3 8.3 5.0 5.0 3.1
3.1
800 bpi NRZI 16.7 10.0 6.2
556 bpi NRZI 24.0 14.4 9.0
200 bpi NRZI 66.7 40.0 25.0
Passing Times, IBG (ms):
6250 bpi 4.0 2.4 1.5
9-Track(PE
and NRZI) 8.0 8.0 4.8 4.8 3.0 3.0
7-Track'(NRZI) 10.0 6.0 3.75
Rewin.~· Time (2400-foot reel) 60 60 60 60 45 45
Rewind/Unload Time:
(2400-fQot reel)
(sec)
66
66 66 66 51 51
Load Operation, approximate time (in
sec) to 'tape unit ready' (after
'reel/cartridge
is mounted and 10 10 10 10 7 7
LOAD/REWIND is pressed)
* Read access time is the interval fr.om initiation of a Forward Read command given to the t-ape control
whe~ tape is not at load point, until the first data byte is read when tape is brought up to speed from
stoppe€! status.
,*Write
access time is the interval from the issuance of a Move command given to ~he tape unit when tape
is not at load point, until the first data byte is written on tape when tape is brought up to speed from
stopped status.
**Start time is the int~rval from the issuance of a Move command to the tape unit, until tape attai~s
90% of specified velocity. .
1

3803 MODEL 3 AND 3803 MODEL 1 DIFFERENCES
The Model 3 was designed to attach to 370/115s and 125s. It is basically a Modell with
three major differences. These differences are required because the 115 and 125
don't
use a standard 370 channel.
-ALUI
is modified to interface with the channel on the 115 and 125 and translate the
activity to 370 channel sequences which the rest
of the 3803 understands.
-Hardware logic called AUTO DATA TRANSFER
is provided to handle the data trans-
fer. It includes the byte counter.
- A 32-BYTE DATA BUFFER
is provided for reading and writing. This large a buffet
is required in the event that data crosses a page boundary. When this happens, time
is required for ALUI to set the address of the new page back into the data address con-
trol in the 115 or 125.
The 3803 Model 3 tape subsystem connects to the Magnetic Tape Adaptor (MT A) inter-
face of system 3115 or system 3125. The term MT A interface refers to a set of lines over
which control and data signals are exchanged between the control unit and the
MT A
portion
of the system.
The basic 3803-3 Control Unit consists
of a MT A interface attachment section, a data
buffer section, a microprogram section, a read section, a write section, a tape unit inter-
face and a CE section.
I/O commands received from CPU through MTA are executed with microprograms resi-
dent in two independent read-only storage (ROS) units within the 3803-3 Control Unit.
One ROS controls the MTA interface lines while the other ROS controls the tape unit
interface.
The 3803-3 Control Unit operates in burst mode for all data transfers and executes com-
mands received across the interface, performing one command on a single tape unit at a
time.
When a 3803 Mode13 is taken offline, all functions from the control to the tape unit
become the same
as a Modell 3803.
3

RECORDING PHILOSOPHY
Types of Codes:
NRZI
PE
GCR
Non-Return
to Zero Indicating
Phase Encoding
Group Coded Recording
The progression
of IBM digital recording from NRZI to PE to GRC is the result of IBM's
effort to provide the customer with the lowest cost and highest reliability tape storage
devices.
NRZI coding utilizes a chnage in flux
to indicate a one and no change to represent a zero.
NRZI tapes can be written in densities
of 200, 556 bpi in 7-track format, and 800 bpi in
9-trace format. (See example A.)
PE coding utilizes a change of magnetic flux in the positive direction to indicate a one
and a change in the negative direction
to indicate a zero. PE tapes are written in 1600
bpi 9-track density only. (See example B.)
GCR coding utilizes a combination of NRZI and PE to achieve the highest density avail-
able on
IBM drives. Density for GCR is effectively 6250 bpi. (See Example C.)
Group Coded Recording -
6250 bpi
Group Coded Recording (GCR) offers many advantages over previously used recording
methods. This recording offers higher reliability even with existing tape libraries.
Greatly expanded error correction capability has been engineered into GCR. Higher
data rates and lower access times
give higher throughput and reduced channel time, re-
sulting in higher system performance. Data is compacted on tape, reducing rewind
times, shortening the length
of tape required for a data set, reducing the number of
reels, reducing mounts and dismounts, and improving overall tape handling. Other
advantages exist,
but here we wish to discuss only the recording technique.
The data
is still recorded in blocks, or groups of characters. A block of data may be
a single character or byte, or a number
of bytes as determined by the programming
system used. The significant improvements in the GCR mode are:
1. The information data
is recorded at an effective density of 6250 bytes per inch.
2. The separation between blocks (IBG)
is 0.3 inch, and,
3. That simultaneous errors in any two
of the nine tracks are corrected automatically.
A GCR block consists
of a preamble, data and a postamble. The preamble and post-
amble are each 80 bytes long and serve to synchronize the read detection circuits in a
manner similar
to previous 1600 bpi subsystems. The data portion of the block consists
of the following:
5

RECORDING PHILOSOPHY (continued)
1. For every seven bytes of channel data, an ECC character (error correcting code)
is generated and these eight bytes, called a data group, are encoded into ten
bytes, called a storage group, which are then written on the tape. There will be
as many of these ten byte storage groups as there are multiples of seven channel
data bytes in the record block. The last group
of the record block may contain
up to 158 storage groups.
2. The remainder or last group of the channel data bytes (zero to six bytes) is
encoded with whatever "pad" bytes are necessary, an auxiliary check character,
and the
ECC character generated from these into a ten-byte residual group.
This residual data group
is created for every block recorded, even though no
residual bytes are found in the record and will always be ten bytes long. The
auxiliary check character verifies read and write operations.
3. End
of data is signaled by a unique subgroup of five bytes immediately preceding
the residual group.
4. Following the residual group, a ten-byte
CRC (Cyclic Redundancy Check) group
is encoded. This group, with the auxiliary check character, ensures the integrity
of the read and write operation including verifying any error corrections which
may have taken place.
5. Interleaved into the recorded block, after every 158 storage groups, is a resync
burst. This burst allows the tape control unit
to put back into full operation any
track(s) which may have lost synchronization or dead tracked due to tape
defects. This action limits dead tracking for greater throughput.
6250 bpi does not relate to actual writing density on tape but to effective data density
Actual density (9042 bpi) is greater due to the formatting and enGoding, but this for-
matting and encoding
is transparent to the user. The power of the format and encoding
method
is such that reliable error correction occurs for any two tracks Simultaneously
in error. Also, tracks are not immediately dequeued or 'dead tracked' when an error
occurs
as they were in the past. It is thus conceivable that a block could have errors in
all nine tracks and appear
to the user to be read error-free so long as only two tracks have
errors at any
given instant.
6250 bpi, combined with 0.3 inch IBGs, provides for greatly improved channel data
rates and access times.
6

RECORDING PHILOSOPHY (continued)
NRZI (Example A)
o o o o
--1
u L
Write Signal
Analog Read Signal
Read Digital Output
PE (Example B)
Write Signal
Read Digital Output
7

RECORDING PHILOSOPHY (continued)
GCR (Example C)
o o o o
J I I I..-.-..
________ n _______ 1 L
J I I I I I I I
Physical Size of a 4K Byte Record
NRZI (800)
PE (1600)
D
GCR (6250)
8

UNDERSTANDING EREPI
EREP is designed to help you decide if you have a machine problem or defective media,
as well as giving you a good idea of how your 3420s are performing. EREPI, with its
enhancement package,
is even more useful than previous EREP versions. EREPI guide-
lines are presented here to assist the CE in managing his 3420 account.
IRD (Information Records Division)
is using a criteria of one temporary write error per
5,000 SIOs and one temporary read error per 70,000 SIOs as a reference for an average
running 3420/3803 customer account.
We agree with these numbers with the understand-
ing that they are intended
to be a national average. They are not a certain indication of
reliability to be obtained in individual accounts because of the influence of unmeasured
variables (ie, customer block size).
The IRD criteria
is used in determining temporary read or write error criteria figures on
the following flowcharts. EREPI parameters should be used to produce a 3420/3803
Summary Output with threshold values at one temporary read and fifteen temporary
write errors. These are suggested initial threshold values which
CE may wish to change
later.
All permanent errors should be investigated to an acceptable conclusion and all temporary
error quantities not meeting
criteria figures should be recorded and tracked to determine
cause
of outside-criteria conditions.
Sheet 1
9

UNDERSTANDING EREPI (continued)
Go to
Sheet 7
Temporary
Error
Analysis
Sheet 2
Begin 3420
ER EP Analysis
Look at "permanent
error summary"
section
Yes
Permanent
Write
If the 3420 is now
working properly
approach problem
as a media failure.
Try to recreate
write failure using
same tape and 3420,
if possible. Use pro-
c~dure shown on
Permanent Read
Analysis Sheet 3,
except perform
wri te instead of
read. Also see M LM
5A -105.
10
If several permanent
errors are listed try
to determine if one
3420 or one volume
of tape appears to be
at fault.
I
Permanent
Read
Go to
Sheet 3
Permanent
Read
Analysis
L ______ ,
I
I
Use Sense
Data to Enter
MLM or PF

UNDERSTANDING EREPI (continued)
Permanent Read Error
Permanent Read Error
(From Sheet 2)
No
Yes
Read tape on tape unit
where failure occurred.
No
Locate and mark failure
with felt pen -read tape
on another tape unit
Permanent Read Error Analysis Chart
Reference MLM 00-011
Sheet 3
Online: Run "FRIEND" (OLT T0200A) or
have customer run
"ditto" or equiv-
alent tape-to-tape job using the fail-
ing tape as input.
Offline:
If 3803 is available perform read
command from
CE panel. Turn on
"stop on data flow check" switch.
11

UNDERSTANDING EREPI (continued
Yes
Return to original
tape unit.
Look for physical
damage on tape.
Develop tape.
Refer to section
on tape damage.
No
Permanent Read Error Analysis Chart
Sheet 4
Refer to section on developing tape.
12

UNDERSTANDING EREPl( continued)
No
Yes
Repai r cause of
damage.
Scope failing
block off line.
Permanent
Read Error Analysis Chart
Sheet 5
:Refer to tape samples elsewhere in this book.
.
The purpose of scoping a permanent read
error
is to locate the failure within the
block and determine the cause. Proceed
as follows:
1. Locate the failing block offline.
2. Loop on the failing block:
a. Read forward-backspace on forward
read failures.
b. Read backward-forward space on
backward read failures.
3. Refer to MLM 00-012 for more informa-
tion.
13

UNDERSTANDING EREPI (continued)
Scope failing
block on line.
Call area technical
assistant or branch
office tape specialist.
Permanent Read Error Analysis Chart
Sheet 6
Use the FRIEND program to scope perm-
anent read errors online:
Note: Don't use channel command word
(CCW) chains. Reading the bad
record breaks the chain.
1. Locate the failing block with the control
unit offline.
2. I Set up a program loop to read forward-
I backspace or read backward-forward
space whichever
is applicable.
3.
By altering the channel command word
(CCW) byte count, the command-out
response to service-in will set the stop
trigger in the tape control at any byte
chosen. Sync external on the stop trig-
ger to examine any portion of the block
in detail.
Note: Detailed observation
of a PE block
is difficult due to normal distor-
tion of the PE signal and the high
amount
of skew the PE detection
circuits can tolerate. Examine
several tracks
at comparable points,
looking for noticeable differences
in levels and waveforms. A record
containing excessive phase shift
is
usually indicated by multi-track
error (MTE) or read write register
(R/W Reg) vertical redundancy
checks (YRC). Check the tape unit
that wrote the tape for a possible
defective write driver. Refer
to
MLM 00-012 for more information.
· If you cannot determine the cause of the
permanent read error,
save the tape and any
: information you have gathered for use by
· your branch office or region tape specialist.
· A permanent read error can cost the cus-
: tomer considerable reconstruction and rerun
time. There should be
NO undefined per-
manent read errors.
14

UNDERSTANDING EREPI (continued)
Temporary Read
Temporary Write
Error Analysis
(From Sheet 2)
I
Look at "CUA Statistics"
section of EREP printout.
I
Look at the three
columns pertaining to:
• Temp RDS
• Temp WRTS
• SIO
Count
I
Using the three columns
indicated above, determine
if
temporary error failure
rate exceeds criteria of:
1.
Temp write error
per
5,000 SIOs
2. Temp read error
per
70,000 SIOs
I
Any CUA entry meeting criteria
may be ignored. Those
not
meeting criteria should be
logged in a manner so as to
give the CE a method of
spotting consistently outside
-criteria volumes
or CUAs.
(See Note 1)
Sheet 7
Note 1:
One logging method would be a grid chart
with CUAs plotted along vertical edge and
volume IDs plotted along horizontal edge.
Number
of temporary errors could be
entered at appropriate intersections. A con-
sistent horizontal pattern of error entries
would indicate a possible
tape unit problem-
a vertical
pattern would indicate a bad vol-
ume of tape.
15