CJL_DECtape_20070117

Newsgroups: alt.sys.pdp8
Subject: TU55/TU56 issues; aren't you glad you asked?  [Note: Extremely Long!]
Date: 17 Jan 2007 04:02:23 -0800
From: CJL

Upgrading TU55 and TU56 guides the CORRECT way [and fixing all of the sins of
the past in the process] and some allied DECtape/LINCtape topics.

Foreward: As per the pestering of certain individuals [you all know who you
are!], I have once again been persuaded to give a complete treatise on a
well-focused topic.  In this case, this is everything you [never] wanted to know
about DECtape hardware as it related to the drives. Only minimal mention of the
controllers prevailing will be issued. [CONTROLLER hardware discussions probably
need two or three other writings; SOFTWARE discussions about DECtape and
LINCtape are best left to at least three other writings, perhaps four, or even
six.]

There are no hard numbers in this document, just the concepts that have to be
dealt with and some relevant examples.

All TU55 and TU56 guides are essentially the same except for some detailed
dimensions that would make them not truly interchangeable. [Actually, this could
be untrue!  They might be interchangeable, but that really isn't important; they
are FUNCTIONALLY interchangeable and suffer from the same problems and have the
same solutions.]

The express purpose of the guides is to "guide" the DEC/LINCtape over the area
the tape must be guided.  In the center is the region where the head is beneath
the tape.  In point of fact the tape NEVER [normally!] touches the head in the
manner superficiously the design suggests.  This is because the moving tape
causes an "air bearing" to form beneath it, thus little or no pressure exists
causing the tape to press down on the head [which would inevitably sandpaper
away the head; this has happened to improperly maintained drives!]

Thus, a well adjusted drive and guides should theoretically never wear out the
tape or the heads and this is likely the closest to "forever" tape media we can
hope to get.  [Please note that there are actual tapes written on 1962-era LINC
machines that still exist and read fine!]

Unfortunately, this perfection more or less only applies to the original LINC
and LINC-8 drives!  These earlier drives are low-performance and thus slightly
higher reliability, depending largely on careful machining of a lot of aluminum
surfaces that aren't stressed to the point that not-so-obvious second-order
problems surface, such as in the TU55 and TU56.  [Nominally, DECtape transfers
every 133 microseconds while LINC/LINC-8 drives transfer every 160 microseconds,
thus the motors are that much slower and gentler on the tape and the long
all-aluminum guides.]

Once we start discussing TU55 and TU56, the whole ballgame changes. Due to using
much "beefier" motors in these drives [arguably especially in the case of the
TU56], DEC wisely chose to make some second-order corrections to the design.

Unfortunately, it was a quite bumpy ride from their original notion to somethat
that actually worked!  Because DECtape is such an inherently reliable device, it
wasn't noticed right away just how wrong things were as they wore out as a
natural consequence of the inadequate design.

Not knowing where to start, I'll just pick an obvious starting point: The front
wear plate.  Once you start stressing the tape, you need to have a device to
push the tape towards the rear lest it tend to wander towards and away from the
front leading to misalignment in the same sense as a diskette drive isn't
running right if not having the media in the correct place over the heads.

Thus, the front guide plate is merely a cover placed well away from the tape and
never directly participates in this pushing process.  Instead, it houses a
free-moving roughly rectangular wear plate.  [Note:  Since the shape of the
guides is actually a critically-designed arc to make the air-bearing effect,
cosmetically the front plate matches the same curve.  To minimize the cosmetic
effect of the wear plate being notices, most of the wear plate isn't visible
from the front, just the little peak of it sticking up higher than the guide
front.  However, it just has a trapezoidal cut, not a curve, to vaguely follow
the overall curve in effect, etc.]

Behind the wear plate is a stiff spring [later deemed WAY too stiff and mounted
in the wrong place!] that does the pushing toward the rear. Its appearance and
overall stiffness is vaguely reminiscent of classic ball-point pen springs from
the same era.  What it is NOT is a much larger and floppier spring made of much
finer wire, which is a fair description of the replacement spring that is much,
much less forceful, etc.

Thus, the problem is that this spring and wear plate combination tend to
over-push the tape towards the rear resulting in all sorts of long-term
problems, such as creating a score line in the rear cheek plate due to the tape
always being way-too-much pushed against it. Inevitablly this scratches the
surface of the rear cheek plate making it a tape killer, as the tape sometimes
is cutting the score deeper, and at other times the tape is getting caught in
the already unwelcome and getting worse score line in the plate, etc.

The net effect on the tape media itself is pretty devastating:  The tape tends
to become thinner, meaning literally the tape is less wide than when it was
manufactured.  Clearly, most tape designs aren't considering that the media
would get less wide than nominal which inevitably leads to various reliability
problems, etc.

If the tape were merely damaged only in that regard, this would be correctable,
as is the case in a drive fully updated.  Unfortunately, another more serious
effect occurs where the tape tends to take on a "fluted" appearance, i.e., it
literally will not lie flat.  These tapes are essentially fatally damaged, since
in a properly adjusted drive, there simply isn't enough force available to make
the tape even temporarily flatten out while passing over the head sufficiently
for normal read/write operations.  The root cause is these are mylar tapes and
indeed have been overstretched and will not ever come back, albeit in a strange
"direction" of the distortion, etc.

Please note:  Properly maintained systems did and should continue to ban the use
of these permanently damaged tapes.  Since good tapes are so hard to get, we
really have to spread the word on preventing further loss of the already scarce
resource of good-condition used tape media.

The way DEC eventually handled the thinner tape issue was to change the amount
of travel on the front wear plate, so that it could still maintain effective
contact with the tape as it pushed the media even further towards the rear
plate.  Thus, the range of acceptable tape widths was raised to a much wider
range than could have been anticipated in the original design. [Note: Thinner
tapes cannot work on LINC/LINC-8 drives because the guides have no pusher
element.  The guide is simply a fixed aluminum channel that has to have proper
width tapes; nothing exists to compensate for tapes of insufficient width; it
simply wasn't in the design because those original drives never contributed to
the problem!  Thus, too-thin tapes can easily wander on the older drives and
will be unreliable there, etc.]

To implement this change, depending on the cover design itself, invariably the
fix is to remove the guides from the drive and mill out some number of
additional thousandths of an inch in the guide in specifically the area where
the wear plate travels.  Thus, the new spec for the wear plate allows tapes to
be that number of additional thousandths of an inch less-wide than nominal, etc.

We need people capable of milling the hard steel that most guides are made from.
However, some early TU55 guides are actually made of aluminum and presumably can
be dealt with as well, just with less brute-force effort to accomplish the same
thing.  [Note:  DEC never implemented a retrofit for the non-steel guides; the
orders were to upgrade to the newly dimensioned guides for all TU55 and TU56
drives and toss the old ones.  Embarassingly enough, some of the new guides have
"ugly spots" in that whatever vendor it was, had no pride in their workmanship; 
the areas normally hidden by the front plates shows gross voids in the metal,
which really doesn't give one much confidence in the workmanship, but usually,
these guides are well within the new spec, etc.  In any case, there doesn't seem
to be any reason to change the material as long as the milling work can be
accomplished on these earlier aluminum guides which are dimensionally identical
where it matters.]

A side issue, valid for the vast majority of guide designs as well as some of
the older ones [especially on TU55] has to do with the exact implementation of
the channel that is affected.  In some guides, the front cover plate is flat on
the back and is not a participant in the milling process that needs to be
applied.  However, in other versions, the slot is also partially implemented in
the cover itself, thus additional milling is apparently needed to make these
parts viable as well.  Especially in the case of the early production TU55
drives, some of the guides are designed with different "amounts" of the front
cover parcelled out between the guide and the cover plate.  Regardless of how
implemented, the set covers a compatible original wear plate travel area, and
the modifications would consist of milling out the same amount of material in
any such case.  [Note:  DEC always recommended using only the newest parts to
the newest spec, but this is because DEC field service needed replaceable parts
and was not a machine shop! Experts today can clearly make any of these work
equivalently, etc.]

Even what the cover is made from isn't a constant, as some seem to be made from
a common part-aluminum [part zinc and I don't know what else!] alloy that is
often identified by some three-number industry code.  [Note:  I think it means
something akin to the vague specification for "carburetor metal" meaning that it
must conform to some percentage range of two or three metals with a really wide
slop factor being quite acceptable, etc.]  Some of them are clearly made of some
not-so-hard steel that may even tend to rust.  At the least, any cosmetic
repaint job would have to be adjusted according to the material, etc.  [Perhaps
the steel ones get black-ox'd and the aluminum-based ones get black anodized? 
Or some form of epoxy-based paint?  I am not a metallurgist or paint expert!]

In any case, once the collection of guide and front plate has been milled out
correctly, this allows the front wear plate travel to the newest DEC spec for
acceptable, albeit width-deprived, tape media that can still flatten out and be
usable [as opposed to the heavily "fluted" tape that cannot].

This solves one of the problems.  Now to that spring:

The spring that pushes the wear plate is just way too gross and ill-thought
through.  However, if you merely change it to a floppier design, you don't get a
viable solution, likely instead a disimprovement!

The change involved drilling a small flat-bottom hole in the front cover's
backside.  This repositions the application point of the spring which likely
affects the spring's overall balance, performance and range.  The diameter of
the hole and spring are as consistent as the original ones were to each other,
but this is a far larger diameter spring base and you get a smoother, more
linear spring motion which has to be viable over the now longer travel range to
handle while adequately pushing the wear plate through all of its now wider
range of linear motion.  [I suspect the application point of the original spring
wasn't centered well enough, thus for certain positions, especially the new one
that an extremely width-deprived tape demands, would tend to provide subtle
forces that would tend to make the force not be applied equally across the
entire wear plate.  Remember, this is a SET of changes that interact AND change
the overall design spec to handle a range of conditions that didn't exist in the
original spec of the drives!  And an additional pitfall - All of the changes
must be implemented as a set; picking and choosing likely can have negative
consequences!]

While this change clearly is designed for use with the last version of the guide
and front cover combination, there basically isn't much change if applying to
older guide material; the same relative positioning and depth of the new spring
hole would still have to be created using a flat-bottom-hole-creating drill bit
or equivalent.

There is no need to fill in the original small spring hole; it can be totally
ignored.  I am not even sure if there ever were production parts that had only
the new-purpose hole!

At this point, we are almost done.  The rest of the problem has to do with the
condition of the rear cheek plate in the area that can or has already developed
a bad score line, the destroyer of tapes that makes them fluted, etc.

There sometimes is an easy solution or band-aid, but in other cases only a more
radical approach can solve it.

Starting with the DEC recommendation first:

The standard later-production TU56 part consists of a single part, namely a rear
cheek plate that is made double-sided and can be used on either side of the
guide.  Obviously there is no designated front here; one face is used when
applying on the left front, and flip it for use on the right front, etc.

IF AND ONLY IF the plates have NEVER been used for any position other than as
factory mounted, this implies that the "backs" of the plates as a pair are
virgin.  As such, the plates can be flipped to provide new surfaces without a
score line.

Unfortunately, if the drives have seen much "service" [including and especially
Field Service], they may have already been flipped, and possibly multiple times
at that.  [It seems that some field service personnel believed that this was
analogous to rotation of tires, even though there is no purpose served in
attempting to use already-worn surfaces.  Perhaps it was a matter of a spiraling
down situation, where each reassembly merely gained partial ground back, albeit
inferior at each iteration, etc.]

Because all of the other modifications discussed here are MANDATORY, a change to
a virgin surface should give long service life from here forward.

Unfortunately, this is a hindsight argument compared to all of the drives that
were in use in various stages of bad maintenance inevitably leading to all of
the cumulative problems that make the drive unreliable and/or a tape killer,
etc.

In early production TU56 and virtually all TU55 with original parts, the rear
cheek plate parts are NOT interchangeable as used.  [Of course this didn't stop
some dimwitted field service people not fully paying attention to try to do the
impossible!]

Non-flippable plates are easily identified: The edge on a two-sided plate tapers
to a thinner thickness at the edge in a small effort to taper off the force
applied to the tape - on both sides. Single-sided plates have the back side
merely flat, no taper at all.

Both plate tapes are upgradeable.  The original can be ground down on the back
side to match the dimensions of the two-sided.  This involves grinding a
straight run near the edge of the aluminum plate at a slight angle.  It's fairly
obvious if you examine a two-sided plate to obtain the correct tapered effect.

At this point, all parts are either now "natural" or modified to be elegible for
two-sided service.  The only remaining problem is reconditioning the plate to
eliminate the score line.  [Note: An advantage of the single-sided rear cheek
plate is that clearly field service should have never used it flipped over,
other than for the aforementioned dimwits that, thankfully, were rare!  Thus,
these now-useful "back" sides have no score line at all, and this step is
essentially cosmetic.]

All surfaces front and back now need to be heavily polished totally smooth.  The
result is now a naked-metal bright-finish "raw" aluminum part differing from the
original double-sided part in an irrelevant way - the total thickness of the
plate is slightly less for having worn off all of the original surface perhaps a
few thousandths of an inch.  This is unimportant, as the only consequence is
that the total depth of the drive overall is now a few thousandths less; all of
the relevant parts move toward the rear of the drive as a set, thus no other
consequences, not even picky ones!

The rest involves finishing the parts.  I am not clear just how the parts are
finished, but I suspect it involves something either anodize or some form of
epoxy paint.  It's also possible that two differing processes were used in
various production phases.  I doubt that any one process is superior to the
other because the score factor has been reduced over an order of magnitude as
compared to prior-to-final-spec drives, etc.

At this point, you can reassemble the rear cheek plates and guides knowing you
have no score line to deal with; tapes will not become fluted and ruined, etc.

If you either had double-sided or opted to upgrade old single-sided to
double-sided, you also can rest assured that you also have a "plan B" should the
finish wear out, but this isn't really likely; these DEC-proscribed
modifications and the retrofits I mention should make an eventual flip
unnecessary, etc.  [But remember, in historical practice, they were
re-assembled, but the other mods were never performed in many instances, thus
scoring inevitably killed the front-most surface of the rear cheek plates, so
inadequate was to merely change the rear cheek plates.  Pathetically, this was
often the case where conventional wisdom applied instead of knowledge!]

A side note on guide assembly and reassembly.  Generally the guides of TU55 and
TU56 are held on with a pair of pins each, as well as with cap-socket screws to
make sure they don't fall out.  [Note:  Should a drive appear that does not seem
to have rear holes on the backside of the chassis front plate, holes should be
drilled, and the guides should be drilled and tapped to provide these holes to
conform to the overall spec.]

Early construction involves the use of what are I believed called either roll
pins or split pins.  Whichever they are, they are of a cross-section that looks
like a "C" washer and are springy.  These need to be removed and thrown away!

The replacement is the use of a steel cylindrical rod pin driven into the guide,
as this form of pin is intended to be easily inserted into the chassis to align
the guide where it belongs.  This allows easy assembly and disassembly for
cleaning, inspection and possible repair or upgrade.

If you have one of these early production situation with the split type of pin,
you're in for a real "treat" removing the guides!  Be prepared to invent all
sorts of wooden jigs to extricate the guides without breaking anything, as they
will put up quite a fight until you can get them out.  Once you establish a
small gap between the guide and the chassis, the rest of the removal process
gets easier as you can then apply wooden wedges to finish the job.  You will
likely want to throw the split pins across the room, hoping to aim at the guy
who designed the drives with this stupid original choice of parts!

In any case, the replacement involves what apparently is a standard mechanical
part, or at least something easily made from standard steel rod stock, merely
cut to length and polished, at least on the end that sticks out of the rear of
the guide and into the drive front chassis plate, etc.

Guides should literally just fall right out of the drive from their own meager
weight.  This is why the drilled and tapped holes on the rear are mandatory, as
well as on the chassis to match; they hold the guides in, NOT the springy pins
that are thousands of percent overkill, etc.

We have come full-circle back to the wear plate, another center of conventional
wisdom versus knowledge.

What was not well known in field service circles was that there are actually not
two, but actually THREE versions of the wear plates!

The actual original plates were perfectly fine, but only if used in an otherwise
modified drive.  Unfortunately, these original plates were only ever INSTALLED
in a totally UNMODIFIED drive!

These original wear plates are made of a smooth, double-sided soft steel.  It is
clearly flippable, analogous to flipping the rear cheek plates, albeit with a
whole lot less effort.

They were often flipped over, which of course just leads to plates scored on
both sides, since in these early days, all were ignorant of all of the other
changes, etc.

Eventually, someone invented a band-aid on the problem that lead eventually to
disaster - a replacement part made of ceramic.  However, there are two distinct
generations of this replacement!

The later ones are NOT a problem, and in fact are the designated preferred
replacement part.  The ceramic clearly can withstand the overload caused by not
modifying the pusher spring a whole lot better. All of this, in hindsight, is
irrelevant for a drive with all of the other mods implemented.  It is arguably
overkill to replace the wear plates with ceramic if all of the other changes are
made.  Clearly, the later production ceramic is welcome in a drive fully
modified, etc.

The problem is that there is an EARLIER plate that, while made of the same
ceramic, is in fact a SINGLE-SIDED plate.  As in the case of the much larger
rear cheek plate, you can see that the edge leads to a tapered edge, but on the
other side it is totally flat, and clearly not flippable.

This did not stop MANY field service morons from flipping them anyway, not
understanding they were erroneously applying a technique valid, but NOT for the
parts in their hands!  The net result is an instant tape destruction device
designed to quickly slice to death all tapes mounted on the now-killer drive,
etc.

Needless to say, all era wear plates will score when no other changes were made,
an additional consequence of conventional wisdom, which, loosely translated, is
that it's perfectly OK to be lazy and just implement a SMALL PORTION of the
regimen, namely whatever is the easiest.  This means most lazy personnel changed
only the wear plates to ceramic, without a care for whether they were that
"harder" to order part that somehow was never in stock, or instead the ones that
were all too plentiful in field service stock, namely the one-sided ones, with
total disregard for a fron versus a back.  Better people might at least get the
sides right.  Some would order the two-sided part so that didn't matter.  Some
even dealt with flipping the rear cheek plates after taking off the guides. 
Few, if any, ever understood the mods regarding the newly milled guides and
accompanying springs and replacement covers.  [I suspect because field service
MANAGERS instructed their guys to deemphasize replacing "expensive" parts like
the guides, as opposed to "cheap" parts like wear plates, etc.]

The later production, reversible ceramic wear plates are dimensionally identical
to the original soft steel ones, and clearly a non-scored set is preferred. 
Again, you have the "plan B" notion that in the future, they could be reversed
if need be, but due to all of the other mods, this is more likely than not to
never be the case!

Thus, if you cannot get new replacement parts, at least recondition whatever you
have.  The soft steel parts can clearly be buffed to a score-free finish,
presumably on both sides.  Expect that they will have to be eventually turned
over because this is a softer material. [After all, they do call it a WEAR
plate, meaning it is the part meant to wear out!]  If you are stuck with the
single-sided ceramic, perhaps it can be carefully ground smooth again until
there is no longer a score line.  [And remember to install them with the correct
sides outward!!!]  The newer ceramic should also respond to the same treatment,
and of course on both sides, perhaps one side needing more treatment than the
other depending on the cumulative effect of the sins of the past, etc.]  There
is unfortunately some evidence that the single-sided ceramic plates do not ever
work as well as either material double-sided plates.  Perhaps someone can figure
out how to "grind" the edge of the ceramic to match the double-sided spec, or at
least so at the edge [leaving them still not really fully double-sided
usage-capable, etc.]

Once all of this is done, the results are quite dramatic.  A good way to monitor
the situation is to get a baseline reading on an oscilloscope of looking at the
raw output of the tape head.  There is a convenient test point for this on the
G88x manchester reader/writer cards invariably associated with the controller
side of things.  Well running drives have very smooth tape motion and the signal
has no jerkiness to it.  Worn out systems have a havoc-ridden signal set that
taxes the ability of the system to even read tapes formatted correctly on other
systems.  Tapes formatted on worn drives are extremely unreliable, especially on
the drives that performed the flakey formatting, etc.

In essence, all of this is mandatory for the continued ability for the drive to
be useful, reliable, and tape-friendly instead of tape-killing.

Specifically on the likes of the TD8E, making these changes is considered
mandatory, since, unless the drive is quite new and not yet worn, the TD8E
software-controlled tape controller will invariably fail.  There isn't much
margin for error on the TD8E, since the software is just barely fast enough to
keep up with the tape when all is well.  The effect of a badly maintained drive
is that some words come too late and some too early.  The late ones aren't a
problem, but the early ones come too quickly for the software to handle it;
there just aren't enough machine cycles to keep up, etc.

In fact, hooking the TU56 to a TD8E and running some tape reliability programs
is a good way to determine if the drive is in good overall condition; poorly
maintained systems will get numerous tape errors on the TD8E well before you can
notice it on a TC01 or TC08, etc.

In any case, the main reason to do all of this is to make the device work as
intended, and that includes working with the now shrinking supply of media out
there.

Turning to a related topic:  Tape drive hubs.

There are several generations of tape hubs.  The best overall results will
probably come as a surprise to anyone who believes they know this seemingly
simple tape subject.  I'll leave your shock and dismay until after first
discussing DEC's offerings:

The absolute original tape hubs are the best ones - the ones on the LINC/LINC-8
drives.  They are essentially correctly designed Hartman hubs.  There is no
replacement nor need to replace.  The only maintenance is to periodically remove
the Hartman springs and remove the worn-out rubber bands and replace with new
rubber bands.  Apply as many as it takes to give a nice firm-yet-gentle "snap"
feel to mounting a reel, etc.  Failure to do so will yield slippage of the reel
on the hub, especially if the reel comes from a TU55 system with the stock hubs,
etc. [see below.]

Then we come to the stock Hartman hubs of the TU55.  As manufactured, these are
basically unusable.  Heavily used drives have many screwdriver or other
implements-related gouges in the chassis front plate as frustrated users
attempted to extricate their tapes from these horribly-designed hubs.  According
to DEC, the best thing is to replace them with the TU56 "spider"hubs.  [See
below for an opposing view.]

On the TU56 that has NOT been deliberately mis-maintained by field service, one
finds the so-called "spider" hubs made totally out of plastic.  They generally
work fine, depending on a gradual plastic-on-plastic "grab" effect between the
hub and tape reel.  This generally works fine, and additionally has the
advantage of being quite accomodating to reels that were ruined by the stock
Hartman hubs on the TU55, etc.

However, a casual placement of the reel can lead to local disaster - a spinning
tape can fall off the drive ruining the tape spilled all over the floor, etc.
and likely getting totally ruined, etc.  Firmly shoving the tape onto the reel
should prevent this, but certain reels need more firmness than others to prevent
falling off; user beware!  Only Hartman hubs can lock a reel onto a hub, etc.

One problem of the early production of spider hubs was the ill-advised design of
placing the two metal set screws that hold the hub on placed 180 degrees apart. 
This inevitably led to the spider hub falling off the shaft, or invariably the
destruction of one of the drilled holes in the plastic hub being overstressed by
overzealous field service monkeys attempting to make this not happen by
overtightening the screws to the point of material failure [which is quite easy;
this is a screw thredded into plastic!]

A simple fix would have been to use a "loctite" compound to prevent screw
loosening, but this was simply never widely done.  Even a damaged hub can be
reconditioned if you are careful.

This type of hub was replaced with a more sensible arrangement:  The holes are
tapped 120 degrees apart.  Thus, the force to hold the hub on the shaft
separates to one to align the hub on the D "flat" side of the shaft [which can
be manufactured by filing the shaft in-place if necessary!] while the other one
can get a good "bite" onto the unmodified circular surface 120 degrees apart.

This arrangement invariably holds far better than the 180 degrees spacing of the
former part, etc.

Even broken hubs can be upgraded to the newer arrangement by merely tapping one
[two if necessary!] replacement holes for the set-screws 120 degrees apart.

[Note:  It is possible to fix the worst-case situation where two holes exist and
are both ruined by over-tightening as long as you carefully place the new holes
120 degrees apart and far enough away from the damaged originals, etc.]  And
additionally, it is advised to use a loctite compound as well to avoid the hub
falling off of the drive, etc.

Unethical field service occasionally and unconscionally stuck obsolete TU55
Hartman hubs on TU56 drives!  This represents the worst of all worlds since
there is usually a plastic surface behind the tape hub which will instantly
gouge on even a mild form of "tool" pressure attempting to remove the hub from
these evil parts [see below for their redemption.]

[Side note:  Extremely late production TU56 and TU56-H drives are actually a
step backwards!  Admittedly, this is a sign that all of the relevant guide parts
have been updated [assuming that field service didn't STEAL any of them!],
except that the final mods are actually cheapening the drive for a misplaced
notion of lowering production costs:  The actual change involves making the
plastic panel of the drive, arguably one of its nicest cosmetic parts, a mere
fraction of the normal size.  This leaves a large area of the front chassis
plate "naked" except for a cheap coat of black paint over some quite rough metal
that was never finished well enough to deserve being seen with only a coat of
paint!  One minor "advantage" of the newest production, is that, should field
service steal your good spider hubs and replace with TU55 unmodified Hartman
hubs, at least you now have a metal finish to gouge while attempting to remove
your tape reels!  And a good paint spray can cover those marks, etc.!]

At this point, you smug old-timers are telling yourself that the best fix is
obviously to just get the new-production TU56-style spider hubs with the
120-degree set-screw spacing and be done with it.

But you're all WRONG!

There is a BETTER, albeit non-DEC, solution to the problem.  Use the original
TU55 Hartman hub, but first MODIFY it!

The problem pervasive to the entire tape drive design and production seems that
DEC just didn't quite have its mechanical engineering up to speed, especially
when it comes to specifying springs!  In this particular case, let's review the
original Hartman hub on the LINC/LINC-8 drives, which are arguably the most
elegant of all.

LINC drive hubs need minor maintenance, the business about the shimming with
rubber bands.  Otherwise, the hubs are elegant, gentle to use, harm neither tape
reel, human, nor tape front chassis plate, and never have marks of "tools" being
applied simply because the problem NEVER occurs.  These hubs err on the side of
the tape falling off or at least slipping, NOT on the side of nearly impossible
to remove the reels without "tools" etc.

These nice hubs use extremely finely-wound long springs of a rather floppy
nature.  They are so gentle, you can easily grab a spring and just take it off
the hub with ease, etc. [So that you can then play with the rubber bands!]

Compare this with the dreaded, unmodified TU55-type Hartman hubs. Putting on the
tape, you can feel the tape being pulled out of your hand, and it won't give it
back to you without a fight -- or a screwdriver!

The elegant fix is to design a replacement spring essentially identical to the
original LINC drive's spring [the same material, but perhaps a slightly
different length?]

It should be made to be slightly unable to fully grasp the reel properly as a
matter of design.  This one then allow final tweaking with rubber bands as
necessary to get the desired range of "grab" one desired.  Yes, this is a
personal preference issue.  Those not faint of heart have been known to grab a
reel on a TU55 and literally rip it off, occasionally actually damaging the reel
in the process! [And then of course there are the habitual "tool" users!]

Clearly, this is the extreme we need to avoid.  At the other end of the
desirable spectrum, it is possible to create an idealized situation that
essentially matches that of a well-maintained LINC drive.

Unfortunately, a well-maintained LINC drive CANNOT hold many reels actually
damaged by original TU55 Hartman hubs.  They simply cannot be used on a
LINC/LINC-8; they will likely slip when the motors turn or even actually fall
off the hub!

For those who continue to use most of their damaged tape reels, this is not a
viable solution.  However, for those who discard their most-damaged reels in
favor of only undamaged or at most SLIGHTLY damaged reels, it is possible to
adjust [via rubber bands] to a compromise [even on LINC drives!] where all of
these undamaged and marginally damaged reels will work fine with the proper
amount of rubber bands, albeit to requiring somewhat more force to remove the
tape than could otherwise be achieved with a gentler adjustment [less rubber
bands] but still well in the reasonable range.

In any case, there is nother like the snap of a reliable Hartman hub, both when
mounting the tape and reasonably unmounting the tape.  Spider hubs are just
inferior to this, albeit adequate and having the side advantage of accomodating
reels damaged by unmodified TU55 Hartman hubs, etc.

Unless you have actually used LINC/LINC-8 drives [I own several such machines,
some totally functional] you cannot have an informed opinion on this.  It simply
didn't happen on DEC-designed drives.  [Note:  DEC did NOT design these drives. 
They simply sourced the same drives for the LINC-8 from the vendor who made them
for the LINC; they are totally compatible with each other and are NOT compatible
with anything DEC used.  A side issue of their specs:  All of the drives use
rubber sprocketed belts to rotate the hubs off of independent drive motors,
while all DEC designs are direct-drive to much more powerful motors, which in
the case of the TU55 actually use clutchs and brakes! Note: Software exists to
unload the tape reels on either TU55 or TU56 systems, but then, if TU55,
actually lock the brake stopping the flapping reels.  The TU56 uses a power
amplifier to drive the motor and to electronically DC brake the drive, thus no
mechanics to implement this more advanced form of direct-drive, etc., but the
drive motors must spin down since they cannot actually lock.  The TU55, but not
the TU56, supports an earlier spec of drive logic voltage swings, usually known
as "relay logic" signals.  This is to allow possible compatibility with the
older 55x series of drives when a TU55 is suitably modified with Wxxx buss
conversion cards.  The TU56 only supports negative buss, such as is standard for
TU55, positive buss such as is standard on the PDP-11, and a command cable, such
as is used on the TD8E.  In the latter case, the G888 Manchester reader/writer
cards are installed directly in the drive, as opposed to G882 [for TC01] or G888
[for TC08] installed in the controller.  Note: There are a total of five such
cards.  The LINC/LINC-8 drives are totally implemented with actual relays, and
likely are at least similar to the so-called "relay logic" levels associated
with the 55x drives DEC used later, etc.]

As often things happen in life, ironically a lot of the remedies that apply to
the TU5x drives involve UNDOING the consequences of half-baked notions, for
which said consequences were not dealt with properly at the time.  In hindsight,
the best solution would have been to have introduced far less change than was
done; a lot of the misadventures associated with band-aiding the problem would
have never happened.

The best drives would be based on the TU55 motor, clutch, brake assembly, but
with properly implemented Hartman hubs.  All of the other guide-related mods
make sense, but took years of incremental and sometimes counter-intuitive
measures to get finally properly implemented.

Had all of this been done, the TU56 as we know it, simply would have never been
implemented and the world would never had seen spider hubs.

And one can ask the question:  Considering the gentle tape handling of the
LINC-8 drives with their much simpler guide design, albeit slightly slower
transfer rate, etc., was all of this speed improvement really worth it?

Arguably no, considering all of the grief just to get the transfers 20 percent
faster.  Direct drive is a good idea, but the beefing up is debatable
considering all of the second-order problems, etc.  [Note:  A 20% slower TU56
electronics would be a blessing for the TD8E!]  Better to just use gentler
direct-drive motors, etc. and not to change the guide system at all!

One more obscure point:  Due to overall size limitations, it is somewhat
difficult to pack what you have in the TU55 into a dual drive. Compare that to
the TU56-H situation which is really just a fully implemented TU56 with some of
the parts removed [but most of the weight!].

Yet, there actually is a quite rare beast.  While I do not know what a BE02 is
[or even if it ever existed!], I have had contact with a BE01 and a BE03!

A BE01 is a standard TC01 with several modifications [one real nasty!]. First of
all, the rack mounting metalwork is sawed off, so it must be specially mounted
in a mini-rack.  As a standard offering for some special custom product, it was
essentially housed in an industrial-type case with a door.  However, since they
didn't implement any form of see-through effect, the entire status light panel
and associated lamp driver modules are removed.

Additionally, the FORMAT switch [the one you have to throw to format DECtapes]
is omitted.  If you implement the switch yourself [it merely involves grounding
a format enable signal], it will NOT format tapes!

The reason for this is the BE01 also deletes ONE MINOR SHORT JUMPER WIRE from
the wrap which disables the ability for the switch to work even if added in! 
[Note: Why bother changing the wrap wire-list?] After many hours of sleuthing,
the ability to format using a miniature switch hung near the module chassis was
restored.  [Note:  We also mounted the racks on a wooden base to allow removal
from the somewhat revolting industrial-type case.  While not as strong, since
there was no rack-mount "ears" to attach to, this was still adequate to allow
functionality, although no easy way to properly mount into a standard DEC rack
cabinet.  The only way would be to attempt to tap and drill holes in the ends of
the sawed off rack chassis members to allow some add-on "ear"equivalents to be
screwed on, etc., but we never went beyond the wooden mount, etc.]

Now turning to the BE03.  This is actually a DUAL TU55 in the space a TU56 would
take!  The chassis is REALLY tight with closely spaced parts!  All is what one
would expect.  However, since this is a TU55-era device, the switches are all of
the TU55 style, meaning they are rather large switches.  As such, much
functionality was removed:

1)  The drives are arbitrarily unit 0 on the left and unit 1 on the right, and
fixed as such.  [Note:  A common user mod to a LINC/LINC-8 drive is to add a
mini-toggle switch which implements the left 0, right 1 OR right 0, left 1
possibilities.]  There are no unit select switches available on a standard BE03.
Note:  There isn't really any viable space to considering doing even the LINC
trick to reverse the drive selects!]

2)  The drives do NOT implement write-protect.  [This is probably something that
should be modified using mini-switches if possible!]

Yet, all else is there, including the individual ability to manually move the
tapes either direction.  Both drives will lock after proper UNLOAD software is
executed, etc. [which, again, a TU56 cannot do!]. All of this fits in a standard
rack size and requires 110v power in the same manner as a TU55 drive, etc.

So, amazingly enough, all three eras of drives have a dual-drive model
available!

cjl (Mike Ross: Are you reading this?)