18 Jul
2018
18 Jul
'18
3:15 a.m.
I'm trying to model the PDP-1 DEC system models, first with SPICE
(ngspice on Linux), then with VHDL.
I'm struggling to understand the properties of various transformers in
the system module circuits, most importantly (at least, the ones I see
in schematics I want to model)
2 winding T2003 (used in the 1304 delay, 1410 pulse generator, 1540
sense amp, 1607 pulse amplifier)
2 winding T2026 (used in the 1410)
4 winding T2029 (used in the 1201 flip-flop)
3 winding T2033 (used in 1204 flip-flop)
2 winding T2048 (used in the 1607 pulse amplifier, but the maintenance
manual suggests this might be substituted with a T2010?)
The maintenance manual also suggests uses for T2006, T2012, T2017,
T2018, T2019, T2020, T2021, T2023, T2024 are used in modules I don't
have schematics for yet.
Is there any source for information about these? Basic specs? I don't
even know what the turns ratios are for these parts, much less
plausible inductances, so my spice models misbehave pretty badly.
I've also posted a question on Reddit about this
https://www.reddit.com/r/AskElectronics/comments/8zsyuq/reverse_engineering…
--Joe
18 Jul
18 Jul
9:39 a.m.
From: Joseph Oswald: Wednesday, July 18, 2018 12:15 AM
I'm trying to model the PDP-1 DEC system models,
first with SPICE
(ngspice on Linux), then with VHDL.
The spice modeling of flip-chip modules has been an interest area
for me. I imagine that's similar, though perhaps less daunting, than
modeling the earlier system modules. Flip chip schematics, etc. can
be found here:
http://www.so-much-stuff.com/pdp8/flipchip/flipchip.php
and some of the spice model stuff here:
http://svn.so-much-stuff.com/svn/trunk/spice/
Particularly a parts library and the Perl to extract information (netlists)
from the Eagle schematics. There are sub-directories
with results
from the Perl for red and blue handled modules, and a few of the
green. Finding
data to fill in the component model library is "fun",
as the parts are long obsolete, and data sheets are hard to find.
I also lack the EE skills to be effective at converting the data sheets
into spice models, even when I can find them.
Your first "fun" step might be to locate module schematics and
convert them to a useful form.
I'm struggling to understand the properties of
various transformers in
the system module circuits, most importantly (at least, the ones I see
in schematics I want to model)
Those are all earlier than the info that I have, which is mostly for
the T2037 and T2052. Perhaps the earlier ones are wound on
a similar core, or the measurement technique Josh and I used
can be used to begin to reverse engineer them:
http://www.so-much-stuff.com/pdp8/repair/pulse.php
Is there any source for information about these? Basic
specs? I don't
even know what the turns ratios are for these parts, much less
plausible inductances, so my spice models misbehave pretty badly.
Unfortunately, the schematics don't usually state the relevant specs
(just part numbers), so it may be necessary to measure up originals.
(Hopefully in-circuit.)
Vince
12:44 p.m.
Thanks for the response. I had seen your info on the two other parts.
Unfortunately, I have no idea if these system modules exist outside a
museum.
I might have to back off even further and work out a technique on
something like the B301 or B611 modules, which seem to be obtainable
and at least vaguely similar to the transformer circuits I am
interested in.
Can you detail a little more what your measurement method is? Do you
have an LRC meter or something, and measure the resistance +
inductance open-circuit? At what frequency?
--Joe
On Wed, Jul 18, 2018 at 9:39 PM, Vincent Slyngstad
<v.slyngstad at frontier.com> wrote:
From: Joseph Oswald: Wednesday, July 18, 2018 12:15
AM
I'm trying to model the PDP-1 DEC system models, first with SPICE
(ngspice on Linux), then with VHDL.
The spice modeling of flip-chip modules has been an interest area for me. I
imagine that's similar, though perhaps less daunting, than modeling the
earlier system modules. Flip chip schematics, etc. can be found here:
http://www.so-much-stuff.com/pdp8/flipchip/flipchip.php
and some of the spice model stuff here:
http://svn.so-much-stuff.com/svn/trunk/spice/
Particularly a parts library and the Perl to extract information (netlists)
from the Eagle schematics. There are sub-directories with results from the
Perl for red and blue handled modules, and a few of the green. Finding data
to fill in the component model library is "fun",
as the parts are long obsolete, and data sheets are hard to find.
I also lack the EE skills to be effective at converting the data sheets into
spice models, even when I can find them.
Your first "fun" step might be to locate module schematics and convert them
to a useful form.
I'm struggling to understand the properties
of various transformers in
the system module circuits, most importantly (at least, the ones I see
in schematics I want to model)
Those are all earlier than the info that I have, which is mostly for the
T2037 and T2052. Perhaps the earlier ones are wound on a similar core, or
the measurement technique Josh and I used can be used to begin to reverse
engineer them:
http://www.so-much-stuff.com/pdp8/repair/pulse.php
Is there any source for information about these?
Basic specs? I don't
even know what the turns ratios are for these parts, much less
plausible inductances, so my spice models misbehave pretty badly.
Unfortunately, the schematics don't usually state the relevant specs
(just part numbers), so it may be necessary to measure up originals.
(Hopefully in-circuit.)
Vince
2:01 p.m.
From: Joseph Oswald: Wednesday, July 18, 2018 9:44 AM
Thanks for the response. I had seen your info on the
two other parts.
Unfortunately, I have no idea if these system modules exist outside a
museum.
You're welcome.
I've seen them occasionally, but not often. Like the flip-chip modules, I
fear much of the information about the less common modules may not
be recoverable.
I might have to back off even further and work out a
technique on
something like the B301 or B611 modules, which seem to be obtainable
and at least vaguely similar to the transformer circuits I am
interested in.
Possibly quite similar, though with a lot less germanium and such,
due to being of a later generation.
Can you detail a little more what your measurement
method is? Do you
have an LRC meter or something, and measure the resistance +
inductance open-circuit? At what frequency?
It has been a while, I'm working from memory, and not an EE to begin
with, so I likely have the details wrong or at least a little fuzzy.
My understanding is that the gauge and length of the wire determine
it's DC resistance. An accurate resistance measurement in the milliohm
range is tricky, as there are usually oxides, etc. In theory, you can get
a number that constrains the type and amount of wire in a winding.
There is an additional issue: sometimes your exemplar is actually
defective, so for the T2052, I tossed the measurements for the
transformer that seemed atypical. If you have at best one example,
I guess you'll just have to hope the results are confirmed by the
simulation.
Measuring the inductance is a little easier, and as I understand things,
much more important anyway. Knowing the inductance and the core
material, one can determine the number of turns. I use a tool called
"mini Ring Core Calculator" by Wilfried Burmeister to explore the
likely possibilities. Knowing or postulating an FT-37-77 core constrains
the problem nicely. For the actual measurements, I borrowed a friend's
inductance meter.
Your emphasis is not really on physically re-creating the part, so it
seems likely that all Spice really needs is the inductance of the
windings and some simple assumptions about their coupling.
Vince
2230
days inactive
2230
days old
3 comments
2 participants
participants (2)
-
josephoswald＠gmail.com -
v.slyngstad＠frontier.com