Re: [TUHS] Spider

Re: [TUHS] Spider

[Paul Ruizendaal]

> Ugh. Memory lane has a lot of potholes. This was a really long time ago. 

Many thanks for that post - really interesting!

I had to look up "Pierce Network", and found it described in the Bell Journal:
https://ia801903.us.archive.org/31/items/bstj51-6-1133/bstj51-6-1133_text.pdf

In my reading the Spider network is a type of Pierce network.

However, the network that you remember is indeed most likely different from Spider:
- it was coax based, whereas the Spider line was a twisted pair
- there was more than one, whereas Spider only ever had one (operational) loop

Condon and Weller are acknowledged in the report about Spider as having done many of its hardware details. The report discusses learnings from the project and having to tune repeaters is not among them (but another operational issue with its 'line access modules’ is discussed).

All in all, maybe these coax loops were pre-cursors to the Spider network, without a switch on the loop (“C” nodes in the Pierce paper). It makes sense to first try out the electrical and line data protocol before starting work on higher level functions.

I have no idea what a GLANCE G is...

Re: [TUHS] Spider

[Jon Steinhart]

Paul Ruizendaal writes:
> > Ugh. Memory lane has a lot of potholes. This was a really long time ago. 
>
> Many thanks for that post - really interesting!
>
> I had to look up "Pierce Network", and found it described in the Bell Journal:
> https://ia801903.us.archive.org/31/items/bstj51-6-1133/bstj51-6-1133_text.pdf
>
> In my reading the Spider network is a type of Pierce network.
>
> However, the network that you remember is indeed most likely different from Spider:
> - it was coax based, whereas the Spider line was a twisted pair
> - there was more than one, whereas Spider only ever had one (operational) loop
>
> Condon and Weller are acknowledged in the report about Spider as having done
> many of its hardware details. The report discusses learnings from the project
> and having to tune repeaters is not among them (but another operational issue
> with its 'line access modules’ is discussed).
>
> All in all, maybe these coax loops were pre-cursors to the Spider network,
> without a switch on the loop (“C” nodes in the Pierce paper). It makes sense
> to first try out the electrical and line data protocol before starting work
> on higher level functions.
>
> I have no idea what a GLANCE G is...

Again, very fuzzy memory here but there are a few folks on the list who might
remember more like Heinz.  As I said earlier, I recall Sandy being in our group
so our network may have been the first try.  I don't know who designed it.

The GLANCE G was an experimental graphics terminal developed in our group.  I
think that John Camlet did a bunch of the hardware design as I have a strong
memory of a long afternoon with him after having weird squiggles show up in
short vectors which turned out to be a microcode bug.  The thing had 4K
resolution in at least one axis, it may have been a square screen.  It used
1K DRAMs and was built around a 74S181 bit slice machine.  A lot of interesting
stuff was done on it;  I mentioned Dick Hause's Display Terminal Editor in an
earlier post.  I also remember a really nice graphical demo of how the phone
network routed around outages which of course no longer works due to lack of
redundancy.

Oh, yeah, the real use for these comes back to me :-)  A number of the folks
in the group including Carl Christensen (who was one of the two people who
interviewed Ken for his job at the labs) had purchased property up in
Vermont for ski cabins.  These lots had ancient surveys, the old "from the big
tree near the large rock" sort of property descriptions.  So the main use of
the Glance G's for a while was to draw out the surveys and figure out what to
do about closure errors.  If you don't know about it, and this is the only
reason why I do, a closure error is when you follow the property boundary
described by the survey and the end isn't at the beginning.

Anyway, these were very cool machines for their time.  I don't recall any other
graphical devices at the time except for glass tty terminals, Tektronix storage
tubes, and the GRIN-2 on the PDP-15 that we used for space war as per earlier
posts.

Dave Hagelbarger designed the keyboard and to this date it's still one of the
best keyboards that I ever used as far as feel goes.  It had plastic cups under
the keys that gave resistance and then accelerated after a certain point so you
pushed on a key and then it would shoot the rest of the way.  And there was a
solenoid on the bottom of the circuit board so there was a little click that
you felt in your fingers.  Great feedback.

There were some extra parts left over from building these.  I remember that Joe
Condon had the glass shop do some work on an extra or dead CRT, and one of the
terminals in his office was actually a fish tank.  When you turned it on a light
would come on behind the tube and you could see the fish swimming around in it.
Maybe Joe deserves credit for the original fish screen saver.

Oh yeah, another thing that I remember being tried was that someone had built an
array of LEDs and photodiodes surrounding the screen that could be used as an
input device.  It was low res but you could point at things with your fingers.

One of the projects that I did was to write software for nicely plotting graphs
on these.  Was used a bunch by Jim Kaiser and crowd for their digital filter
work.  I remember a really nice graph that Jim did that showed just how carefully
the touch tone frequencies were chosen to have no harmonics in common.

This isn't really UNIX stuff but I suppose that it passes for prehistory.  Again,
I seem to recall that one of these was up in the UNIX lab but have no memory of
what it was used for.  If I had to guess, it would be chess.

Oh yeah, I think that you can see the GLANCE G instruction set if you have a
first edition copy of Newman and Sproul.  I remember getting a copy of that book
and having a weird sense of deja vu during their discussion of vector graphics.
Then I realized that it was because it was the instruction set that was burned
into my memory.

Jon

Re: [TUHS] Spider

[Gregg Levine]

Hello!
GACK! (Excuse me.) Hello!
That description of how someone back there used a batch of LEDs and
photodiodes to form an input device triggered a memory of my own. One
Steve Ciarcia who wrote the Circuit Cellar column for Byte magazine
described that same method. He even included a photo of the prototype
using a picture frame to hold them. The code to manage it was written
in BASIC of all things for one of his custom cards, and it was mounted
on his equally custom system. He also described a similar method of
generating graphics using two DAC devices and a scope itself switched
into X-Y mode. (Code was also in BASIC for starters.)

Let us now return to the wonders of UNIX and as it relates to the
early networking ideas.
-----
Gregg C Levine gregg.drwho8@gmail.com
"This signature fought the Time Wars, time and again."

On Thu, Jan 23, 2020 at 9:38 PM Jon Steinhart <jon@fourwinds.com> wrote:
>
> Paul Ruizendaal writes:
> > > Ugh. Memory lane has a lot of potholes. This was a really long time ago.
> >
> > Many thanks for that post - really interesting!
> >
> > I had to look up "Pierce Network", and found it described in the Bell Journal:
> > https://ia801903.us.archive.org/31/items/bstj51-6-1133/bstj51-6-1133_text.pdf
> >
> > In my reading the Spider network is a type of Pierce network.
> >
> > However, the network that you remember is indeed most likely different from Spider:
> > - it was coax based, whereas the Spider line was a twisted pair
> > - there was more than one, whereas Spider only ever had one (operational) loop
> >
> > Condon and Weller are acknowledged in the report about Spider as having done
> > many of its hardware details. The report discusses learnings from the project
> > and having to tune repeaters is not among them (but another operational issue
> > with its 'line access modules’ is discussed).
> >
> > All in all, maybe these coax loops were pre-cursors to the Spider network,
> > without a switch on the loop (“C” nodes in the Pierce paper). It makes sense
> > to first try out the electrical and line data protocol before starting work
> > on higher level functions.
> >
> > I have no idea what a GLANCE G is...
>
> Again, very fuzzy memory here but there are a few folks on the list who might
> remember more like Heinz.  As I said earlier, I recall Sandy being in our group
> so our network may have been the first try.  I don't know who designed it.
>
> The GLANCE G was an experimental graphics terminal developed in our group.  I
> think that John Camlet did a bunch of the hardware design as I have a strong
> memory of a long afternoon with him after having weird squiggles show up in
> short vectors which turned out to be a microcode bug.  The thing had 4K
> resolution in at least one axis, it may have been a square screen.  It used
> 1K DRAMs and was built around a 74S181 bit slice machine.  A lot of interesting
> stuff was done on it;  I mentioned Dick Hause's Display Terminal Editor in an
> earlier post.  I also remember a really nice graphical demo of how the phone
> network routed around outages which of course no longer works due to lack of
> redundancy.
>
> Oh, yeah, the real use for these comes back to me :-)  A number of the folks
> in the group including Carl Christensen (who was one of the two people who
> interviewed Ken for his job at the labs) had purchased property up in
> Vermont for ski cabins.  These lots had ancient surveys, the old "from the big
> tree near the large rock" sort of property descriptions.  So the main use of
> the Glance G's for a while was to draw out the surveys and figure out what to
> do about closure errors.  If you don't know about it, and this is the only
> reason why I do, a closure error is when you follow the property boundary
> described by the survey and the end isn't at the beginning.
>
> Anyway, these were very cool machines for their time.  I don't recall any other
> graphical devices at the time except for glass tty terminals, Tektronix storage
> tubes, and the GRIN-2 on the PDP-15 that we used for space war as per earlier
> posts.
>
> Dave Hagelbarger designed the keyboard and to this date it's still one of the
> best keyboards that I ever used as far as feel goes.  It had plastic cups under
> the keys that gave resistance and then accelerated after a certain point so you
> pushed on a key and then it would shoot the rest of the way.  And there was a
> solenoid on the bottom of the circuit board so there was a little click that
> you felt in your fingers.  Great feedback.
>
> There were some extra parts left over from building these.  I remember that Joe
> Condon had the glass shop do some work on an extra or dead CRT, and one of the
> terminals in his office was actually a fish tank.  When you turned it on a light
> would come on behind the tube and you could see the fish swimming around in it.
> Maybe Joe deserves credit for the original fish screen saver.
>
> Oh yeah, another thing that I remember being tried was that someone had built an
> array of LEDs and photodiodes surrounding the screen that could be used as an
> input device.  It was low res but you could point at things with your fingers.
>
> One of the projects that I did was to write software for nicely plotting graphs
> on these.  Was used a bunch by Jim Kaiser and crowd for their digital filter
> work.  I remember a really nice graph that Jim did that showed just how carefully
> the touch tone frequencies were chosen to have no harmonics in common.
>
> This isn't really UNIX stuff but I suppose that it passes for prehistory.  Again,
> I seem to recall that one of these was up in the UNIX lab but have no memory of
> what it was used for.  If I had to guess, it would be chess.
>
> Oh yeah, I think that you can see the GLANCE G instruction set if you have a
> first edition copy of Newman and Sproul.  I remember getting a copy of that book
> and having a weird sense of deja vu during their discussion of vector graphics.
> Then I realized that it was because it was the instruction set that was burned
> into my memory.
>
> Jon

Re: [TUHS] Spider

[Heinz Lycklama]

Interesting history. Carl Christensen was also one of about 4 people
who interviewed me for a job at Bell Labs during a snowstorm in
Summit and Murray Hill in early 1969. I did use the early Glance terminals
during my programming days at Bell Labs, but had nothing to do
with their design. Carl and I taught a group of Explorer Scouts on
Monday evenings at Murray Hill in the early 1970's. Jon was one
of those scouts that we taught UNIX and its programming languages.
And we introduced them (including Jon) to  spelunking! in the summer.

Heinz

On 1/23/2020 6:37 PM, Jon Steinhart wrote:
> Paul Ruizendaal writes:
>>> Ugh. Memory lane has a lot of potholes. This was a really long time ago.
>> Many thanks for that post - really interesting!
>>
>> I had to look up "Pierce Network", and found it described in the Bell Journal:
>> https://ia801903.us.archive.org/31/items/bstj51-6-1133/bstj51-6-1133_text.pdf
>>
>> In my reading the Spider network is a type of Pierce network.
>>
>> However, the network that you remember is indeed most likely different from Spider:
>> - it was coax based, whereas the Spider line was a twisted pair
>> - there was more than one, whereas Spider only ever had one (operational) loop
>>
>> Condon and Weller are acknowledged in the report about Spider as having done
>> many of its hardware details. The report discusses learnings from the project
>> and having to tune repeaters is not among them (but another operational issue
>> with its 'line access modules’ is discussed).
>>
>> All in all, maybe these coax loops were pre-cursors to the Spider network,
>> without a switch on the loop (“C” nodes in the Pierce paper). It makes sense
>> to first try out the electrical and line data protocol before starting work
>> on higher level functions.
>>
>> I have no idea what a GLANCE G is...
> Again, very fuzzy memory here but there are a few folks on the list who might
> remember more like Heinz.  As I said earlier, I recall Sandy being in our group
> so our network may have been the first try.  I don't know who designed it.
>
> The GLANCE G was an experimental graphics terminal developed in our group.  I
> think that John Camlet did a bunch of the hardware design as I have a strong
> memory of a long afternoon with him after having weird squiggles show up in
> short vectors which turned out to be a microcode bug.  The thing had 4K
> resolution in at least one axis, it may have been a square screen.  It used
> 1K DRAMs and was built around a 74S181 bit slice machine.  A lot of interesting
> stuff was done on it;  I mentioned Dick Hause's Display Terminal Editor in an
> earlier post.  I also remember a really nice graphical demo of how the phone
> network routed around outages which of course no longer works due to lack of
> redundancy.
>
> Oh, yeah, the real use for these comes back to me :-)  A number of the folks
> in the group including Carl Christensen (who was one of the two people who
> interviewed Ken for his job at the labs) had purchased property up in
> Vermont for ski cabins.  These lots had ancient surveys, the old "from the big
> tree near the large rock" sort of property descriptions.  So the main use of
> the Glance G's for a while was to draw out the surveys and figure out what to
> do about closure errors.  If you don't know about it, and this is the only
> reason why I do, a closure error is when you follow the property boundary
> described by the survey and the end isn't at the beginning.
>
> Anyway, these were very cool machines for their time.  I don't recall any other
> graphical devices at the time except for glass tty terminals, Tektronix storage
> tubes, and the GRIN-2 on the PDP-15 that we used for space war as per earlier
> posts.
>
> Dave Hagelbarger designed the keyboard and to this date it's still one of the
> best keyboards that I ever used as far as feel goes.  It had plastic cups under
> the keys that gave resistance and then accelerated after a certain point so you
> pushed on a key and then it would shoot the rest of the way.  And there was a
> solenoid on the bottom of the circuit board so there was a little click that
> you felt in your fingers.  Great feedback.
>
> There were some extra parts left over from building these.  I remember that Joe
> Condon had the glass shop do some work on an extra or dead CRT, and one of the
> terminals in his office was actually a fish tank.  When you turned it on a light
> would come on behind the tube and you could see the fish swimming around in it.
> Maybe Joe deserves credit for the original fish screen saver.
>
> Oh yeah, another thing that I remember being tried was that someone had built an
> array of LEDs and photodiodes surrounding the screen that could be used as an
> input device.  It was low res but you could point at things with your fingers.
>
> One of the projects that I did was to write software for nicely plotting graphs
> on these.  Was used a bunch by Jim Kaiser and crowd for their digital filter
> work.  I remember a really nice graph that Jim did that showed just how carefully
> the touch tone frequencies were chosen to have no harmonics in common.
>
> This isn't really UNIX stuff but I suppose that it passes for prehistory.  Again,
> I seem to recall that one of these was up in the UNIX lab but have no memory of
> what it was used for.  If I had to guess, it would be chess.
>
> Oh yeah, I think that you can see the GLANCE G instruction set if you have a
> first edition copy of Newman and Sproul.  I remember getting a copy of that book
> and having a weird sense of deja vu during their discussion of vector graphics.
> Then I realized that it was because it was the instruction set that was burned
> into my memory.
>
> Jon

Re: [TUHS] Spider

[Jon Steinhart]

Paul Ruizendaal writes:
> 
>
> > I have vague memories here that maybe Heinz can help with if his are any better.
> > I believe that Sandy played a part in "the loop" or "the ring" or whatever it
> > was called that we had connecting our Honeywell 516 to peripherals.  I do
> > remember the 74S00 repeaters because of the amount of time that Dave Weller
> > spent tuning them when the error rate got high.  Also, being a loop, Joe
> > Condon used to pull his connectors out of the wall whenever people weren't
> > showing up to a meeting on time.  I don't know whether our network was a
> > forerunner to the spider network.
>
> It most likely was Spider - it became operational in 1972. The vist report that I linked to earlier also says:
>
> "The current system contains just one loop with the switching computer (TEMPO I),
> four PDP-11/45 computers, two Honeywell 516 computers, two DDP 224 computers,
> and one each of Honeywell 6070, PDP-8 and PDP-11/20. In fact many of these are
> connected in turn to other items of digital equipment.”
>
> It would be interesting to know more about the H516’s and Spider, any other recollections?

Ugh.  Memory lane has a lot of potholes.  This was a really long time ago.

I believe that the 516 network predated your quote above and was most likely
designed for the 516 as I'm pretty sure that Sandy was in our department; I
have some recollection of his office being in our block of rooms and of
spending some time talking to him.

I am guessing that one of these was installed in the UNIX lab but I
really don't know.  My guess is based on a recollection that there was a
GLANCE G up there, and the only interface to one of these was the network.
I can't imagine them not having a cool toy like the GLANCE G up there.
Maybe Doug or Ken remembers.

I also recall that the network was technically a Pierce Loop, named after
John Pierce of course.  My best recollection is that this ran off of the
DMA controller on the 516.  I think that it regularly sent a big block of
data around the loop.  That block had "holes" in it in which any "terminal"
device could say "hey, gimme some attention" or "here's some data".  I'm
using terminal here to mean any device other than the 516 that was moving
the bits around.  I have no memory of how collisions were handled or whether
there was any sort of bit-per-terminal device to avoid collisions.

The bits were sent around on coax.  They didn't go very far, and there were
repeater boards everywhere.  I don't recall any separate power supplies for
the repeater boards so maybe there was some DC on the coax for that.  The
repeater boards used 74S00s for drivers, and there was some sort of tuned
circuit that had an adjustment.  Every once in a while Dave Weller would
make the circuit with a scope and a little green screwdriver to keep things
working.

A day sticks out in my memory.  We were still in building 2 before buildings
6 and 7 were built and we moved to building 7.  The error rate on the network
went through the roof and became unusable, and I think when that happened the
516 crashed.  Wasn't a repeater, wasn't Joe pulling the plugs out of the wall.
Someone in a nearby lab had left a cover off of their cyclotron.  Can't express
how comforting that was; on par with discovering that the track on which I ran
in junior high school was a radioactive toxic waste dump or coming across family
photos of me in an open-pit Quebec asbestos mine as a kid.

I know that there were other instances of this network around the labs.  I did
a project that supported integrated circuit test systems in labs driven by a
single cost computer.  Yes folks, there was a day in which a computer cost more
than a wafer stepper :-)  This system was based on 516-TSS and the network.  It
sticks in my mind because I was working on it during the big Western Electric
debacle around the memory chip failures in the big new ESS installed in the
basement of the Chicago Sears Tower.

Anyway, that's all I can remember about this at the moment.  Heinz should know
something about it but his memory may be as bad as mine.  Another person who
would know is John Camlet.  Don't know if he's still around; a quick search
shows someone of about the right age living in Plainfield, NJ so if someone
lives out there try knocking on the door.

Jon

Re: [TUHS] Spider

[Paul Ruizendaal]

> I have vague memories here that maybe Heinz can help with if his are any better.
> I believe that Sandy played a part in "the loop" or "the ring" or whatever it
> was called that we had connecting our Honeywell 516 to peripherals.  I do
> remember the 74S00 repeaters because of the amount of time that Dave Weller
> spent tuning them when the error rate got high.  Also, being a loop, Joe
> Condon used to pull his connectors out of the wall whenever people weren't
> showing up to a meeting on time.  I don't know whether our network was a
> forerunner to the spider network.

It most likely was Spider - it became operational in 1972. The vist report that I linked to earlier also says:

"The current system contains just one loop with the switching computer (TEMPO I),
four PDP-11/45 computers, two Honeywell 516 computers, two DDP 224 computers,
and one each of Honeywell 6070, PDP-8 and PDP-11/20. In fact many of these are
connected in turn to other items of digital equipment.”

It would be interesting to know more about the H516’s and Spider, any other recollections?