On Fri, 16 Jul 2021, Sergey Kubushyn wrote:

On Thu, 15 Jul 2021, John Miles wrote:

I'm almost 100% sure it is just a snafu that happens all the time and was
quite widespread back then when CAD didn't exist yet but leave 0.00001%
doubt of missing something.

Agreed, that is very weird. Same with those resistors... usually you try
to
minimize the number of BOM line items by consolidating resistors, but they
seem to have done the opposite here.

The LC BPF at the output of the A69 post-VR amp is also funky. On the
schematic, the inductors are 0.652 uH, 0.568 uH, 0.652 uH, 0.94 uH, 1.047
uH, 1.047 uH, and 0.94 uH, but in the parts list they are more
reasonable-looking values. Meanwhile, the capacitors are bog-standard 220
pF and 470 pF parts.

The filter could be explained by the engineering department getting a
little
too carried away with their new computer, or maybe they just measured the
parts they used and transcribed the values to the BOM without rounding.
Likely true for the "1.047 uH" toroids, but it doesn't explain the others,
and nothing explains CR3068 as far as I can tell. I wonder if it's an IP
protection feature. If Hickok or Lavoie or somebody like that were to
clone
the instrument, things would get awkward if they had to explain what that
diode was for in court. Similar to the practice of adding fictitious
towns
to a map to guard against commercial copying.

I doubt that IP protection guess. Back then they didn't have CADs and those
designs were not just computer files. It was a bunch of paper drawings,
mylar tape PCB designs and so on. Not even starting on somebody willing to
make verbatim copies, it was totally impossible even if somebody decided to
do that. There was no way to just COPY something back then. Yes, one could
take the Tektronix schematics (that were not secret and were included in
service manuals) and make a copy of their instrument. You still had to go
through entire cycle of at least re-drawing their schematics, then do
mechanical design, PCB layout (with Mylar tape or something like that) and
much more. There is no way one could do this using just dumb monkeys -- it
would require engineers. And such things like that diode or multiplying BOM
lines is almost impossible to miss -- it would've been noticed and fixed in
the process. That is not even starting on their proprietary chips and other
stuff that should've been somehow re-implemented.

It is not like you can just blindly copy [stolen] design files and do
something with your own label as it is possible today. It was way more labor
intensive back then and it required brains. Brains were an absolute
necessity back then. It is a handicap nowadays but back then things were
totally different...

BTW, I forgot to mention that actual board matches schematic exactly. Don't
know about those inductance values -- have absolutely no interest
desoldering and measuring those -- but actual parts are ALL there. There is
a CR3068 (something like 1N4148) right next to 'LS145 pins, that 2.2K
resistor in ONE place instead of 3.3K everywhere is actually a 2.2K one and
the rest is actually 3.3K and so on...

---
******************************************************************
* KSI@home KOI8 Net < > The impossible we do immediately. *
* Las Vegas NV, USA < > Miracles require 24-hour notice. *
******************************************************************

On Thu, 15 Jul 2021, John Miles wrote:

I'm almost 100% sure it is just a snafu that happens all the time and was
quite widespread back then when CAD didn't exist yet but leave 0.00001%
doubt of missing something.

Agreed, that is very weird. Same with those resistors... usually you try to
minimize the number of BOM line items by consolidating resistors, but they
seem to have done the opposite here.

The LC BPF at the output of the A69 post-VR amp is also funky. On the
schematic, the inductors are 0.652 uH, 0.568 uH, 0.652 uH, 0.94 uH, 1.047
uH, 1.047 uH, and 0.94 uH, but in the parts list they are more
reasonable-looking values. Meanwhile, the capacitors are bog-standard 220
pF and 470 pF parts.

The filter could be explained by the engineering department getting a little
too carried away with their new computer, or maybe they just measured the
parts they used and transcribed the values to the BOM without rounding.
Likely true for the "1.047 uH" toroids, but it doesn't explain the others,
and nothing explains CR3068 as far as I can tell. I wonder if it's an IP
protection feature. If Hickok or Lavoie or somebody like that were to clone
the instrument, things would get awkward if they had to explain what that
diode was for in court. Similar to the practice of adding fictitious towns
to a map to guard against commercial copying.

I doubt that IP protection guess. Back then they didn't have CADs and those
designs were not just computer files. It was a bunch of paper drawings,
mylar tape PCB designs and so on. Not even starting on somebody willing to
make verbatim copies, it was totally impossible even if somebody decided to
do that. There was no way to just COPY something back then. Yes, one could
take the Tektronix schematics (that were not secret and were included in
service manuals) and make a copy of their instrument. You still had to go
through entire cycle of at least re-drawing their schematics, then do
mechanical design, PCB layout (with Mylar tape or something like that) and
much more. There is no way one could do this using just dumb monkeys -- it
would require engineers. And such things like that diode or multiplying BOM
lines is almost impossible to miss -- it would've been noticed and fixed in
the process. That is not even starting on their proprietary chips and other
stuff that should've been somehow re-implemented.

It is not like you can just blindly copy [stolen] design files and do
something with your own label as it is possible today. It was way more labor
intensive back then and it required brains. Brains were an absolute
necessity back then. It is a handicap nowadays but back then things were
totally different...

---
******************************************************************
* KSI@home KOI8 Net < > The impossible we do immediately. *
* Las Vegas NV, USA < > Miracles require 24-hour notice. *
******************************************************************

Or the famous Cadence vs. Avanti EDA code lawsuit.  Very hard for Avanti to explain the exact same misspellings in the comments!  Sure you didn't steal Cadence's IP, Avanti!  Case closed.          Jim Ford Sent from my Verizon, Samsung Galaxy smartphone

On Thu, 15 Jul 2021, Ed Breya via groups.io wrote:

Eh, there is NO logic needed. Wired OR is done without any diodes using
OC/OD outputs. And it is done like that at the very bottom of the schematic.
On the same 'LS145, just different 2 outputs.

There is no different biasing -- 2N2907 emitter is connected to +15V, 10K
from emitter to base, 10K to 'LS145 OPEN COLLECTOR output. When that output
is high, 2N2907 base is connected to its emitter and that 10K to 'LS145 OPEN
COLLECTOR output is just hanging in the air. Transistor is closed, no
collector current. When 'LS145 output is active (low), that 10K is grounded
and there is base current that makes transistor open fully, to saturation.
2N2907A has beta at least a hundred, so that base current is more than
sufficient for full saturation. Emitter is at +15V wrt ground so a diode
drop makes no difference.

The schematic is classical RF diode switch. No level shifting with diodes.
No special biasing -- it is either ~+15V or ~-15V. Plus/minus couple of
volts makes absolutely no difference.

Please, don't play a guessing game with trivial advices. Just look at the
schematic.

I'm 99.9999999% sure it is just a snafu caused by multiple people and even
departments working on that schematic before it reached manufacturing. There
is no sacred knowledge or black magic there, the thing is trivial.

If it is purely a (TTL) logic section, it could be that a little
diode-logic was added to save a gate package or such, and required a
little different biasing to assure proper levels. Often, a Ge or Schottky
diode could be used with a saturated transistor output to get a valid
TTL-low when isolation is needed, or a regular Si diode could work with
some biasing help.

---
******************************************************************
* KSI@home KOI8 Net < > The impossible we do immediately. *
* Las Vegas NV, USA < > Miracles require 24-hour notice. *
******************************************************************

I'm almost 100% sure it is just a snafu that happens all the time and was
quite widespread back then when CAD didn't exist yet but leave 0.00001%
doubt of missing something.

Agreed, that is very weird. Same with those resistors... usually you try to
minimize the number of BOM line items by consolidating resistors, but they
seem to have done the opposite here.

The LC BPF at the output of the A69 post-VR amp is also funky. On the
schematic, the inductors are 0.652 uH, 0.568 uH, 0.652 uH, 0.94 uH, 1.047
uH, 1.047 uH, and 0.94 uH, but in the parts list they are more
reasonable-looking values. Meanwhile, the capacitors are bog-standard 220
pF and 470 pF parts.

The filter could be explained by the engineering department getting a little
too carried away with their new computer, or maybe they just measured the
parts they used and transcribed the values to the BOM without rounding.
Likely true for the "1.047 uH" toroids, but it doesn't explain the others,
and nothing explains CR3068 as far as I can tell. I wonder if it's an IP
protection feature. If Hickok or Lavoie or somebody like that were to clone
the instrument, things would get awkward if they had to explain what that
diode was for in court. Similar to the practice of adding fictitious towns
to a map to guard against commercial copying.

-- john, KE5FX

If it is purely a (TTL) logic section, it could be that a little diode-logic was added to save a gate package or such, and required a little different biasing to assure proper levels. Often, a Ge or Schottky diode could be used with a saturated transistor output to get a valid TTL-low when isolation is needed, or a regular Si diode could work with some biasing help.

Ed

On Thu, 15 Jul 2021, Ed Breya via groups.io wrote:

It is pure logic, ON/OFF. 'LS145 pin active (low), +15V forward bias, pass
diodes in full conductance, those shorting to ground reverse biased i.e.
OFF. 'LS145 pin inactive, -15V reverse bias, pass diodes OFF, those shorting
to ground forward biased i.e. input (after pass diode) shorted to ground,
output (before pass diode) also shorted. Both sides of switches DC blocked
by 10nF ceramics.

There is nothing fancy there. Those are just switches that are either ON or
OFF connecting one of 5 filters between IN and OUT. No fancy p-i-n diodes,
no active parts besides those switches. All diodes are just dumb rectifiers,
not even fast ones (think 1N4001). All filters are totally passive, either
LC or crystal. 10/100Hz has +15V power for thermal stabilization (oven). 4
out of 5 have a trivial gain control (trimpot as variable divider). The
fifth one doesn't have even that, its output level is used as reference to
set other filters' gain against.

There is a dual stage gain block after that (post-VR amplifier) on a
separate board but there is also nothing fancy there -- fixed gain medium
power output stage with 8th order LC filter after it preceeded by a single
variable gain stage. Both are regular common emitter amplifiers with
variable stage using a trimpot in parallel with emitter degeneration
resistor to set required gain.

There are fancy p-i-n diodes and whole lot of other interesting stuff in
494AP but not on that board.

I'm almost 100% sure it is just a snafu that happens all the time and was
quite widespread back then when CAD didn't exist yet but leave 0.00001%
doubt of missing something.

It is probably like e.g. very good vintage Kenwood KA-9100 amplifier where a
small droplet of ink made its way to one channel 10K resistor on the
schematic and made it 1.0K. This went all the way to production and all
those KA-9100 have 10K resistor on one channel final power amplifier board
and 1K on other channel.

BTW, it is not that I need this as a repair hint; it is just out of
curiosity.

There is another fancy thing -- IN has a ground bias return make of series
connected 64uH inductor and TWO 330 Ohm resistors while OUT has a single 680
Ohm resistor instead of those two 330 Ohm ones. Once again there is no
particular reason for that. There is no high voltage there (-2dBm level AC
signal, DC blocked by a 10nF ceramic capacitor), no high power, no
especially high frequencies (it is 10 MHz IF) and those resistors are crappy
1/4W 5% carbon film ones.

I don't know the circuit here, but you may want to look at the logic and
PIN control circuits and circuit descriptions in more detail. Sometimes a
couple extra dB attenuation or passing, depending on the situation, makes
things work much better. It may boil down to turning on a PIN harder, to
short signals to ground, or to get a little more through signal, in
certain spots. Also note that in many setups, the Q (hence BW) of the
crystal RBW IF filters is set by variable PIN RF resistance (DC control
current), so that aspect has to be handled in the overall biasing scheme.

---
******************************************************************
* KSI@home KOI8 Net < > The impossible we do immediately. *
* Las Vegas NV, USA < > Miracles require 24-hour notice. *
******************************************************************

I don't know the circuit here, but you may want to look at the logic and PIN control circuits and circuit descriptions in more detail. Sometimes a couple extra dB attenuation or passing, depending on the situation, makes things work much better. It may boil down to turning on a PIN harder, to short signals to ground, or to get a little more through signal, in certain spots. Also note that in many setups, the Q (hence BW) of the crystal RBW IF filters is set by variable PIN RF resistance (DC control current), so that aspect has to be handled in the overall biasing scheme.

Ed

Can somebody come out with any reason for the CR3068 diode on VR 2nd Filter
Select board (left bottom corner on the schematic on page 185 of 494A/AP
Vol.2 Service Manual)?

There are 5 filters there, each having same exact switch on input and output
made from couple of diodes and a 2N2907 switch applying either +15V for ON
or -15V for OFF via 64uH inductor thus forward/reverse biasing those diodes.
Two switches per filter, on input and output so all inactive filters are
disconnected from both ends.

The switches are driven from an open collector 74LS145. Same filter is
inserted for 1MHz and 3MHz (it is a pass-through with only gain adjustment
potentiometer -- it is SECOND stage filters so actual filter is implemented
in the first stage). To make it happen they tied 'LS145 pins 2 and 9
together so either one would pull the 2N2907 switch bases low thus inserting
that bypass "filter" for corresponding BCD code.

The 100kHz/10kHz/1kHz filters are all separate so 'LS145 3/4/5 pins each
control their own pair of switches.

Then, 10Hz/100Hz share the same filter, exactly like 1MHz/3MHz, but this
time it is an actual filter. That means 'LS145 pins 6 and 7 should be tied
together to implement wired OR exactly like it is done with pins 2 and 9 for
the 1MHz/3MHz one.

Tying OC outputs into wired OR is perfectly fine when active level is LOW,
it is done all the time and this is what 1MHz/3MHz have. However for
10Hz/100Hz they put a DIODE between corresponding pins that is equivalent to
a short pulling the Q2020 and Q8035 bases to ground when either pin 6 or 7
is low.

But that is not all -- all switches are identical with 3.3K resistor to -15V
EXCEPT the one implemented with Q2020 that has 2.2K instead of 3.3K with
everything else EXACTLY identical. That also doesn't change anything as all
those switches are DC-blocked by 10nF capacitors on both sides of each
diodes/transistor switches. Diodes are also identical so there is absolutely
no reason for 10Hz/100Hz part to be any different from 1MHz/3MHz one.

Am I missing something?

---
******************************************************************
* KSI@home KOI8 Net < > The impossible we do immediately. *
* Las Vegas NV, USA < > Miracles require 24-hour notice. *
******************************************************************