Topics

A 2465 teaser...

Chuck Harris
 

I'm coming to think it is the power circuitry around the
CH1 and CH2 preamps.

I soldered a wire onto one of the pads where the delay line,
attaches to the PCB, and carefully adjusted the hold off,
on my bench scope, so that I got a clean picture of that
output... which is the direct output of the Channel Switch.

I connected the CH1 Gate signal from the back panel to my
bench scope's CH2, and used it for the trigger. The CH2
display of the gate gives me a clear picture of the beginning
and end of each channel's time slot.

I can easily see each of the channels as it outputs from
the Channel Switch in succession. Hook on CH1 and CH2 and
everything. OBTW, that Hook is the clue, as it is *always*
there on CH1 and CH2.

If you look at the power distribution circuitry around the
preamps, you will see that CH3/CH4 use only +5V and -5V, and
share common RF isolation filter networks to clean up their
sources of power.

CH1/CH2 share a common RF isolation filter networks on +5V,
and common RF isolation filter networks on -5V and -8V.

Ok, why is the hook the clue?

Simple, the hook is the vertical preamplifier's way of saying
that it has squishy power, and doesn't like the channel switch
turning its load on and off when CH1 or CH2 is selected.

When the channel switch connects to CH1 or CH2, it is pulling
the quiescent output level of the preamp towards the load
potential of the channel switch... Oh so slowly. Forming the Hook.

It could also be a bad output amplifier on both CH1 and CH2, but
what are the odds of that happening simultaneously?

I think it is more likely that one of the power supply isolation
RF filters that CH1 and CH2 share is squishy, and I will find
one supply has a nice healthy 5us pulse on it. A mirror image
of the hook that shows on the leading edge (left) of the CH1 and
CH2 traces.

I will check that soon and report back.

-Chuck Harris

machine guy @Mac [TekScopes] wrote:

Chuck, I think its from the readout. Most likely the readout DO circuitry. Probably in U485, U475, or U975. There is a complicated switch using emitter coupled logic implemented by U485, U475, and U975. If this switch and logic fails, readout is coupled to the vertical channels of the channel switch. I have never tested to see if it affects only Ch1 and Ch2. But I have seen it. Check waveforms 50 and 51 on U975.
Mac

On ‎Thursday‎, ‎November‎ ‎9‎, ‎2017‎ ‎04‎:‎51‎:‎43‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:


A hook is a trace that goes from vertical to horizontal
with a little curve between. If you could see the whole
thing, it would look like a shark's tooth, or a lazy
saw tooth pulse.

With display off, it looks like this:

(_____________________________________________CH1

(_____________________________________________CH2

_____________________________________________CH3
_____________________________________________CH4

When you turn display on it looks like:

(__(____(____(______(________(_______(((___(__CH1

(_(__(_____(________(___(_____(_((_____(_(((__CH2

_____________________________________________CH3
_____________________________________________CH4

The "(" sections are actually a sharp rise with
a slower fall, and look like random grass. They
appear only on CH1 and CH2, not at all on CH3 and CH4.

The "(" sections after the trigger are at least
0.75 div high, and unaffected by vertical input
condition, or attenuator setting. The initial pulse
hook, is undoubtedly just one of the "herd"... only
it appears with or without the display being turned
on.

I haven't tried replacing the CH1 and CH2 preamp
yet. It would surprise me to find that both failed
in exactly the same way at the exactly the same time.

[Note that CH1 and CH2 are otherwise performing normally,
and in calibration, and that the "(" stuff does not
appear in the CH2 output connector on the back panel.
I would think that would rule out the preamps.]

It would seem more likely to me that it would be
a bad bypass cap on a filter section that feeds both
CH1 and CH2 in parallel. But the power looks pretty
clean, and nothing larger than 1 or 2 mv.

-Chuck Harris

Chuck Harris
 

Another possibility.

The complexity of the circuitry going to CH5 has made me lift an
eyebrow a couple of times now. I even bought a stash of new CA3046's,
and replaced U475, to no effect, at the time.

I did check 48, 49, 50, 51, 52, 53, 54 and 55 waveforms, and they all
nominally matched what the book shows, when the SUT was set the way the
book said to. But notice I said nominally, the hook was obviously
there on 54 and 55.. 52 and 53 in the manual are just plain wrong as
far as amplitude goes... 150mv, not 5v p-p. I think the rest were ok,
though I may have missed some other clue.

-Chuck Harris

machine guy @Mac [TekScopes] wrote:

Here is what I think, and I'm pretty confident. The channel switch (U400) design "leaks" Ch 5 into Ch1 and Ch2. Tek decided (for some reason) to fix this by adding logic external to the channel switch. That external logic is implemented in U485, U475 and U975. These parts function so that any time horizontal A or horizontal B (A sweep or B sweep) are operating the readout into U400 is clamped to zero at the emitter outputs of U475C and U475D. In your case, they are not functioning, most likely a fault in U975. So, the readout vertical reaches U400 and leaks into Ch1 and Ch2, just like Tek always knew it would.

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎11‎:‎47‎:‎44‎ ‎AM‎ ‎CST, machine guy @Mac [TekScopes] <TekScopes@...> wrote:


Chuck, I think its from the readout. Most likely the readout DO circuitry. Probably in U485, U475, or U975. There is a complicated switch using emitter coupled logic implemented by U485, U475, and U975. If this switch and logic fails, readout is coupled to the vertical channels of the channel switch. I have never tested to see if it affects only Ch1 and Ch2. But I have seen it. Check waveforms 50 and 51 on U975.
Mac

On ‎Thursday‎, ‎November‎ ‎9‎, ‎2017‎ ‎04‎:‎51‎:‎43‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:


A hook is a trace that goes from vertical to horizontal
with a little curve between. If you could see the whole
thing, it would look like a shark's tooth, or a lazy
saw tooth pulse.

With display off, it looks like this:

(_____________________________________________CH1

(_____________________________________________CH2

_____________________________________________CH3
_____________________________________________CH4

When you turn display on it looks like:

(__(____(____(______(________(_______(((___(__CH1

(_(__(_____(________(___(_____(_((_____(_(((__CH2

_____________________________________________CH3
_____________________________________________CH4

The "(" sections are actually a sharp rise with
a slower fall, and look like random grass. They
appear only on CH1 and CH2, not at all on CH3 and CH4.

The "(" sections after the trigger are at least
0.75 div high, and unaffected by vertical input
condition, or attenuator setting. The initial pulse
hook, is undoubtedly just one of the "herd"... only
it appears with or without the display being turned
on.

I haven't tried replacing the CH1 and CH2 preamp
yet. It would surprise me to find that both failed
in exactly the same way at the exactly the same time.

[Note that CH1 and CH2 are otherwise performing normally,
and in calibration, and that the "(" stuff does not
appear in the CH2 output connector on the back panel.
I would think that would rule out the preamps.]

It would seem more likely to me that it would be
a bad bypass cap on a filter section that feeds both
CH1 and CH2 in parallel. But the power looks pretty
clean, and nothing larger than 1 or 2 mv.

-Chuck Harris

hahi@... [TekScopes] wrote:
I am puzzled. Anyone have any ideas?
Not sure exactly what you mean by hooks ...

This may not be the same but I have seen a few scopes with noise
on the trace(s) looking like about one minor div of digital noise
randomly across the trace(s). All fixed by replacing the preamp(s).

/Håkan





------------------------------------
Posted by: hahi@...
------------------------------------


------------------------------------

Yahoo Groups Links












------------------------------------
Posted by: machine guy <@Mac>
------------------------------------


------------------------------------

Yahoo Groups Links



 

I have a 2465A that once did something I would describe exactly as you have for your 2465.  I puzzled out the circuit involving U475, U485 and U975 enough to understand what appears to be its purpose.  That purpose is to block readout "noise" from the channel switch when either Ch1 or Ch2 are running.  I checked the Ch1 and Ch2 inputs to the channel switch and they were clean, the outputs (waveforms 54 and 55) were not.  I tried a different channel switch and got the same results.  It seemed to be associated with the readout (turning off readout would stop the noise the "running part" of the trace).  So I studied the complex circuit at the input to Ch5 and concluded its purpose was to block the readout when Ch1 or Ch2 are running.  In my case, one of the pins of U975 was stuck.  I changed U975 and everything got well for me.  U975 isn't the only part that is critical to the operation.  Its just the easiest to diagnose.  A simple test of my theory would be to firmly ground pin 2 of U485.  This will prevent all readout vertical from reaching the channel switch.  Then if the noise on the trace goes away, the complex circuit is failing.  Cant say how without more tests.

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎01‎:‎12‎:‎28‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:

 
Another possibility.

The complexity of the circuitry going to CH5 has made me lift an
eyebrow a couple of times now. I even bought a stash of new CA3046's,
and replaced U475, to no effect, at the time.

I did check 48, 49, 50, 51, 52, 53, 54 and 55 waveforms, and they all
nominally matched what the book shows, when the SUT was set the way the
book said to. But notice I said nominally, the hook was obviously
there on 54 and 55.. 52 and 53 in the manual are just plain wrong as
far as amplitude goes... 150mv, not 5v p-p. I think the rest were ok,
though I may have missed some other clue.

-Chuck Harris

machine guy @Mac [TekScopes] wrote:
Here is what I think, and I'm pretty confident. The channel switch (U400) design "leaks" Ch 5 into Ch1 and Ch2. Tek decided (for some reason) to fix this by adding logic external to the channel switch. That external logic is implemented in U485, U475 and U975. These parts function so that any time horizontal A or horizontal B (A sweep or B sweep) are operating the readout into U400 is clamped to zero at the emitter outputs of U475C and U475D. In your case, they are not functioning, most likely a fault in U975. So, the readout vertical reaches U400 and leaks into Ch1 and Ch2, just like Tek always knew it would.

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎11‎:‎47‎:‎44‎ ‎AM‎ ‎CST, machine guy @Mac [TekScopes] <TekScopes@...> wrote:


Chuck, I think its from the readout. Most likely the readout DO circuitry. Probably in U485, U475, or U975. There is a complicated switch using emitter coupled logic implemented by U485, U475, and U975. If this switch and logic fails, readout is coupled to the vertical channels of the channel switch. I have never tested to see if it affects only Ch1 and Ch2. But I have seen it. Check waveforms 50 and 51 on U975.
Mac

On ‎Thursday‎, ‎November‎ ‎9‎, ‎2017‎ ‎04‎:‎51‎:‎43‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:


A hook is a trace that goes from vertical to horizontal
with a little curve between. If you could see the whole
thing, it would look like a shark's tooth, or a lazy
saw tooth pulse.

With display off, it looks like this:

(_____________________________________________CH1

(_____________________________________________CH2

_____________________________________________CH3
_____________________________________________CH4

When you turn display on it looks like:

(__(____(____(______(________(_______(((___(__CH1

(_(__(_____(________(___(_____(_((_____(_(((__CH2

_____________________________________________CH3
_____________________________________________CH4

The "(" sections are actually a sharp rise with
a slower fall, and look like random grass. They
appear only on CH1 and CH2, not at all on CH3 and CH4.

The "(" sections after the trigger are at least
0.75 div high, and unaffected by vertical input
condition, or attenuator setting. The initial pulse
hook, is undoubtedly just one of the "herd"... only
it appears with or without the display being turned
on.

I haven't tried replacing the CH1 and CH2 preamp
yet. It would surprise me to find that both failed
in exactly the same way at the exactly the same time.

[Note that CH1 and CH2 are otherwise performing normally,
and in calibration, and that the "(" stuff does not
appear in the CH2 output connector on the back panel.
I would think that would rule out the preamps.]

It would seem more likely to me that it would be
a bad bypass cap on a filter section that feeds both
CH1 and CH2 in parallel. But the power looks pretty
clean, and nothing larger than 1 or 2 mv.

-Chuck Harris

hahi@... [TekScopes] wrote:
I am puzzled. Anyone have any ideas?
Not sure exactly what you mean by hooks ...

This may not be the same but I have seen a few scopes with noise
on the trace(s) looking like about one minor div of digital noise
randomly across the trace(s). All fixed by replacing the preamp(s).

/Håkan

[Non-text portions of this message have been removed]



------------------------------------
Posted by: hahi@...
------------------------------------


------------------------------------

Yahoo Groups Links












------------------------------------
Posted by: machine guy <@Mac>
------------------------------------


------------------------------------

Yahoo Groups Links





[Non-text portions of this message have been removed]

Chuck Harris
 

According to tektronix, U485A, U485B, U485C, and U475, form an
analog switch that routes either the readout at the base of U485A,
or the ground at the base of U485C to the output at the emitter of
U475C. The selection voltages are developed by voltage dividers
on the complementary outputs of U975C, and U975A.

Which pretty much fits your description exactly...

Also interesting is readout can only happen when the CH3 and CH4
inputs are deselected (eg. VS3 and VS4 are high). So, readout only
steals the beam from CH1 and CH2, or when no channels are being
displayed.... never from CH3 or CH4.

Why didn't I notice that phrase earlier?

I checked all of the signals once again, and they match what the manual
describes, however, the signal examples are only for when the readout
is turned off. Hmmm?

The nand gate in U975A, that creates signals 51 and 50 are indeed inverses
of each other, and look clean... and they match the waveform examples.

Hmmm? Hmmm?

Ok, why am I in AC mode, and why didn't I notice that the DC signals
at 52 and 53 were straddling ground?

Ah! U485 can do that if the "C" section is broken... Let's replace
it. Now where are my CA3046's? I only bought them a couple of years
ago... Seems like only yesterday....

Found them, replaced U485, and.......

The problem is gone!

Seems to me like I am always tagging a 2N3904 on the side of an open
circuited CA3046.

One of these days I should start recording which section of the
CA3046 pops.

Thanks machineguy for forcing me to think about that ugly mess
tagged on the side of the Channel Switch hybrid.

And thanks to everyone else for their hints, and being a sounding
board.

-Chuck Harris

machine guy @Mac [TekScopes] wrote:

I have a 2465A that once did something I would describe exactly as you have for your 2465. I puzzled out the circuit involving U475, U485 and U975 enough to understand what appears to be its purpose. That purpose is to block readout "noise" from the channel switch when either Ch1 or Ch2 are running. I checked the Ch1 and Ch2 inputs to the channel switch and they were clean, the outputs (waveforms 54 and 55) were not. I tried a different channel switch and got the same results. It seemed to be associated with the readout (turning off readout would stop the noise the "running part" of the trace). So I studied the complex circuit at the input to Ch5 and concluded its purpose was to block the readout when Ch1 or Ch2 are running. In my case, one of the pins of U975 was stuck. I changed U975 and everything got well for me. U975 isn't the only part that is critical to the operation. Its just the easiest to diagnose. A simple test of my theory would be to firmly ground pin 2 of U485. This will prevent all readout vertical from reaching the channel switch. Then if the noise on the trace goes away, the complex circuit is failing. Cant say how without more tests.

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎01‎:‎12‎:‎28‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:


Another possibility.

The complexity of the circuitry going to CH5 has made me lift an
eyebrow a couple of times now. I even bought a stash of new CA3046's,
and replaced U475, to no effect, at the time.

I did check 48, 49, 50, 51, 52, 53, 54 and 55 waveforms, and they all
nominally matched what the book shows, when the SUT was set the way the
book said to. But notice I said nominally, the hook was obviously
there on 54 and 55.. 52 and 53 in the manual are just plain wrong as
far as amplitude goes... 150mv, not 5v p-p. I think the rest were ok,
though I may have missed some other clue.

-Chuck Harris

 

Glad you got it fixed.  Glad I could help.  Only the returning the many times you have helped me.


Mac

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎04‎:‎53‎:‎00‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:

 
According to tektronix, U485A, U485B, U485C, and U475, form an
analog switch that routes either the readout at the base of U485A,
or the ground at the base of U485C to the output at the emitter of
U475C. The selection voltages are developed by voltage dividers
on the complementary outputs of U975C, and U975A.

Which pretty much fits your description exactly...

Also interesting is readout can only happen when the CH3 and CH4
inputs are deselected (eg. VS3 and VS4 are high). So, readout only
steals the beam from CH1 and CH2, or when no channels are being
displayed.... never from CH3 or CH4.

Why didn't I notice that phrase earlier?

I checked all of the signals once again, and they match what the manual
describes, however, the signal examples are only for when the readout
is turned off. Hmmm?

The nand gate in U975A, that creates signals 51 and 50 are indeed inverses
of each other, and look clean... and they match the waveform examples.

Hmmm? Hmmm?

Ok, why am I in AC mode, and why didn't I notice that the DC signals
at 52 and 53 were straddling ground?

Ah! U485 can do that if the "C" section is broken... Let's replace
it. Now where are my CA3046's? I only bought them a couple of years
ago... Seems like only yesterday....

Found them, replaced U485, and.......

The problem is gone!

Seems to me like I am always tagging a 2N3904 on the side of an open
circuited CA3046.

One of these days I should start recording which section of the
CA3046 pops.

Thanks machineguy for forcing me to think about that ugly mess
tagged on the side of the Channel Switch hybrid.

And thanks to everyone else for their hints, and being a sounding
board.

-Chuck Harris

machine guy @Mac [TekScopes] wrote:
I have a 2465A that once did something I would describe exactly as you have for your 2465. I puzzled out the circuit involving U475, U485 and U975 enough to understand what appears to be its purpose. That purpose is to block readout "noise" from the channel switch when either Ch1 or Ch2 are running. I checked the Ch1 and Ch2 inputs to the channel switch and they were clean, the outputs (waveforms 54 and 55) were not. I tried a different channel switch and got the same results. It seemed to be associated with the readout (turning off readout would stop the noise the "running part" of the trace). So I studied the complex circuit at the input to Ch5 and concluded its purpose was to block the readout when Ch1 or Ch2 are running. In my case, one of the pins of U975 was stuck. I changed U975 and everything got well for me. U975 isn't the only part that is critical to the operation. Its just the easiest to diagnose. A simple test of my theory would be to firmly ground pin 2 of U485. This will prevent all readout vertical from reaching the channel switch. Then if the noise on the trace goes away, the complex circuit is failing. Cant say how without more tests.

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎01‎:‎12‎:‎28‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:


Another possibility.

The complexity of the circuitry going to CH5 has made me lift an
eyebrow a couple of times now. I even bought a stash of new CA3046's,
and replaced U475, to no effect, at the time.

I did check 48, 49, 50, 51, 52, 53, 54 and 55 waveforms, and they all
nominally matched what the book shows, when the SUT was set the way the
book said to. But notice I said nominally, the hook was obviously
there on 54 and 55.. 52 and 53 in the manual are just plain wrong as
far as amplitude goes... 150mv, not 5v p-p. I think the rest were ok,
though I may have missed some other clue.

-Chuck Harris


[Non-text portions of this message have been removed]

 

Chuck, I am sure you understand that "complex circuit" on Ch5 input to U400, the Channel Switch.  I think I do but I will document it for others and I hope you will correct any errors. 
U485A and U485B form a differential amplifier whose operation depends on current to their common emitters through R483.  Diode CR484 is fitted to stop the amplifier by "stealing" the emitter current when U975 pin 1 is high.  U485C and U485D form a similar amplifier that is disabled when the Pin 3 of U975 is high.  The outputs from these two differential amplifers is joined by the common collectors of U485B and U485D.  U975A acts to select one or the other (not both) of these two amplifiers.  Amplifier One (U485A and U485B) carries the readout signal as presented on pin 2 of U485A.  Amplifier Two (U485C and U485D) "amplifies" the ground that is present on pin 6 of U485C.  Thus, the signal at pin 6 of U475C is either Readout Vertical or ground, depending on the logic state of U975A.  U475C is a simple emitter follower to provide the current that drives the input to Ch5+ on U400 and other circuitry.  That other circuitry includes another differential amplifier (unswitched this time) comprised of U475A and U475B that inverts the Ch5+ signal for input to U475D and on to Ch5-.  U475D is another emitter follower to drive current into Ch5- on U400.  
Bottom line, a somewhat complex yet straight forward "mux" to present either Readout Vertical or Ground to the inputs of U400 Ch5. 
I was surprised by the number of components Tek used to accomplish such a simple task and aks the question: "why doesn't the Channel Switch ignore Ch5 inputs when Ch1 or Ch2 is selected?"  I concluded that late in the development of the 2465 they discovered U400 "leaks" Ch5 into Ch1 and Ch2.  The dies for U400 were proved and mass production had likely begun.  The only solution was to block the "leak" by adding a switch in the pathway for Ch5 input so it is "grounded" when Ch1 or Ch2 is running.  No time for special IC's and fidelity was important.  So Tek built a kludge to mitigate the problem.  It was cheap, and it worked, and it prevented a complex redesign of U400, so it stayed for the 2465A and 2465B.  
Mac

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎07‎:‎25‎:‎58‎ ‎PM‎ ‎CST, machine guy @Mac [TekScopes] <TekScopes@...> wrote:

 
Glad you got it fixed.  Glad I could help.  Only the returning the many times you have helped me.

Mac

On ‎Friday‎, ‎November‎ ‎10‎, ‎2017‎ ‎04‎:‎53‎:‎00‎ ‎PM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:

 
According to tektronix, U485A, U485B, U485C, and U475, form an
analog switch that routes either the readout at the base of U485A,
or the ground at the base of U485C to the output at the emitter of
U475C. The selection voltages are developed by voltage dividers
on the complementary outputs of U975C, and U975A.

Which pretty much fits your description exactly...

Also interesting is readout can only happen when the CH3 and CH4
inputs are deselected (eg. VS3 and VS4 are high). So, readout only
steals the beam from CH1 and CH2, or when no channels are being
displayed.... never from CH3 or CH4.

Why didn't I notice that phrase earlier?

I checked all of the signals once again, and they match what the manual
describes, however, the signal examples are only for when the readout
is turned off. Hmmm?

The nand gate in U975A, that creates signals 51 and 50 are indeed inverses
of each other, and look clean... and they match the waveform examples.

Hmmm? Hmmm?

Ok, why am I in AC mode, and why didn't I notice that the DC signals
at 52 and 53 were straddling ground?

Ah! U485 can do that if the "C" section is broken... Let's replace
it. Now where are my CA3046's? I only bought them a couple of years
ago... Seems like only yesterday....

Found them, replaced U485, and.......

The problem is gone!

Seems to me like I am always tagging a 2N3904 on the side of an open
circuited CA3046.

One of these days I should start recording which section of the
CA3046 pops.

Thanks machineguy for forcing me to think about that ugly mess
tagged on the side of the Channel Switch hybrid.

And thanks to everyone else for their hints, and being a sounding
board.

-Chuck Harris



[Non-text portions of this message have been removed]




[Non-text portions of this message have been removed]

Chuck Harris
 

Hi Mac,

Your description looks to be ok. But what is wrong with the description
in the manual? It is available for our use, and seems pretty clear to me.

My shame is I should have read it earlier as it contained lots of clues
that would have gotten me off of the bad power boondoggle.

2 or 3 years ago, I did read the description, and concluded that U475 or
U485 was likely bad. I replaced U475, but hesitated to change U485 when
replacing U475 had no effect. I don't like replacing parts shotgun style,
much preferring learning over efficiency. Shotgunning when done rationally
is very fast, but you never really learn what the failure was, just the
area where the failure occurred.

I think the difference in the way I approach repairs is because I am an
electrical engineer, not a technician. The failures a technician fixes
are circuits that once worked. Replace the broken part, and operation
will be restored. The failures an engineer fixes may never have worked
correctly. You have to replace the broken idea to fix an engineering
failure.

I don't believe the complexity in feeding CH5 is a mistake, but rather
an issue of partitioning... I believe the channel switch hybrid works
exactly the way it was supposed to work.

Partitioning was the job of the 2465 system engineer.

What I think they did is reason that the display needed to be updated
often, without fail, or it would flicker, or wink out. You can tolerate
a little of that in a trace, but in the display it would look ridiculously
broken. The display has to be stable. Display has to be priority 1.

So, what they did is make a 4 channel MUX for CH1-4, and give it individual
selection lines that mostly mirrored the state of the channel selection
buttons on the front panel.

CH1 and CH2 were special in that tradition required there be a function
where the MUX output is CH1 + CH2... the "ADD" function. That means there
is an adder gate before the CH1/CH2 MUX. I think the designer simply made
that adder gate a 3 input adder that added CH1+CH2+CH5, and presented
the sum to the channel switch MUX.

For this to work, the way you deselect an input is to change it to a
zero, or ground (Ch1 + CH2 + 0= no display...). The VS1 - VS4 selection
lines automatically do that for CH1 - CH4 input, but CH5 has no such
selection line, so the user of the Channel Switch is required to do it
himself.

Earlier I said it was a partitioning problem. I believe the mistake
that happened was the Channel Switch hybrid designer, and the Display
hybrid designer each thought the other was responsible for grounding
out the display input when data was not being displayed. The possibility
is also pretty good that neither designed their part with the 2465, or
the other designer, in mind. They designed to spec.

The simple fix would have been to add VS5 to deselect the display channel.

-Chuck Harris

machine guy @Mac [TekScopes] wrote:

Chuck, I am sure you understand that "complex circuit" on Ch5 input to U400, the Channel Switch. I think I do but I will document it for others and I hope you will correct any errors.
U485A and U485B form a differential amplifier whose operation depends on current to their common emitters through R483. Diode CR484 is fitted to stop the amplifier by "stealing" the emitter current when U975 pin 1 is high. U485C and U485D form a similar amplifier that is disabled when the Pin 3 of U975 is high. The outputs from these two differential amplifers is joined by the common collectors of U485B and U485D. U975A acts to select one or the other (not both) of these two amplifiers. Amplifier One (U485A and U485B) carries the readout signal as presented on pin 2 of U485A. Amplifier Two (U485C and U485D) "amplifies" the ground that is present on pin 6 of U485C. Thus, the signal at pin 6 of U475C is either Readout Vertical or ground, depending on the logic state of U975A. U475C is a simple emitter follower to provide the current that drives the input to Ch5+ on U400 and other circuitry. That other circuitry includes another differential amplifier (unswitched this time) comprised of U475A and U475B that inverts the Ch5+ signal for input to U475D and on to Ch5-. U475D is another emitter follower to drive current into Ch5- on U400.
Bottom line, a somewhat complex yet straight forward "mux" to present either Readout Vertical or Ground to the inputs of U400 Ch5.
I was surprised by the number of components Tek used to accomplish such a simple task and aks the question: "why doesn't the Channel Switch ignore Ch5 inputs when Ch1 or Ch2 is selected?" I concluded that late in the development of the 2465 they discovered U400 "leaks" Ch5 into Ch1 and Ch2. The dies for U400 were proved and mass production had likely begun. The only solution was to block the "leak" by adding a switch in the pathway for Ch5 input so it is "grounded" when Ch1 or Ch2 is running. No time for special IC's and fidelity was important. So Tek built a kludge to mitigate the problem. It was cheap, and it worked, and it prevented a complex redesign of U400, so it stayed for the 2465A and 2465B.
Mac

Siggi
 

On Sun, 12 Nov 2017 at 11:05 Chuck Harris cfharris@... [TekScopes] <
TekScopes@...> wrote:



I don't believe the complexity in feeding CH5 is a mistake, but rather
an issue of partitioning... I believe the channel switch hybrid works
exactly the way it was supposed to work.
You can see the data sheet for the channel switch die here <
http://w140.com/tek_made_sm.pdf>, it's "named" 203-0211-90 and appears on
page 6-29 (or 271). It seems pretty clear that this is behaving as
intended, and that the #5 input is designed for a much lower bandwidth
signal than #1/2 and #3/4.
I'm guessing the original intent would have been for the readout signal to
be in a quiescent state when not displayed, or maybe it was simply less
expensive to do the low-bandwidth switching externally to the hybrid.

 

Chuck,
The only thing I see wrong with that description is I have to read it more carefully.  I am embarrassed to admit that I only skimmed that description in the past and wrote my description without even looking at the manual (except for the schematic).  So I agree, lets use the manual.  Duhh
I, too, am an engineer, retired after 35+ years.  Worked during my college years as a radio/TV technician and repairman.  So nearly 40 years around oscilloscopes before assembling my home lab with three Tek scopes and a parts mule.  Ive been retired 18 years now.  I hate shotgunning for repair.  It leaves me wondering if I fixed the problem or covered it up.  Worse, like you said, I never learn how it was designed to work.  That's why the amplifier/switch arrangement for Ch5 input to U400 stands out to me.  It just doesn't fit with my expectation for quality design work.   Your explanation of why its that way (someone wrote the spec wrong) is as good as any. 
Anyway, glad you got it fixed.
Mac

On ‎Sunday‎, ‎November‎ ‎12‎, ‎2017‎ ‎10‎:‎05‎:‎39‎ ‎AM‎ ‎CST, Chuck Harris cfharris@... [TekScopes] <TekScopes@...> wrote:

 
Hi Mac,

Your description looks to be ok. But what is wrong with the description
in the manual? It is available for our use, and seems pretty clear to me.

My shame is I should have read it earlier as it contained lots of clues
that would have gotten me off of the bad power boondoggle.

2 or 3 years ago, I did read the description, and concluded that U475 or
U485 was likely bad. I replaced U475, but hesitated to change U485 when
replacing U475 had no effect. I don't like replacing parts shotgun style,
much preferring learning over efficiency. Shotgunning when done rationally
is very fast, but you never really learn what the failure was, just the
area where the failure occurred.

I think the difference in the way I approach repairs is because I am an
electrical engineer, not a technician. The failures a technician fixes
are circuits that once worked. Replace the broken part, and operation
will be restored. The failures an engineer fixes may never have worked
correctly. You have to replace the broken idea to fix an engineering
failure.

I don't believe the complexity in feeding CH5 is a mistake, but rather
an issue of partitioning... I believe the channel switch hybrid works
exactly the way it was supposed to work.

Partitioning was the job of the 2465 system engineer.

What I think they did is reason that the display needed to be updated
often, without fail, or it would flicker, or wink out. You can tolerate
a little of that in a trace, but in the display it would look ridiculously
broken. The display has to be stable. Display has to be priority 1.

So, what they did is make a 4 channel MUX for CH1-4, and give it individual
selection lines that mostly mirrored the state of the channel selection
buttons on the front panel.

CH1 and CH2 were special in that tradition required there be a function
where the MUX output is CH1 + CH2... the "ADD" function. That means there
is an adder gate before the CH1/CH2 MUX. I think the designer simply made
that adder gate a 3 input adder that added CH1+CH2+CH5, and presented
the sum to the channel switch MUX.

For this to work, the way you deselect an input is to change it to a
zero, or ground (Ch1 + CH2 + 0= no display...). The VS1 - VS4 selection
lines automatically do that for CH1 - CH4 input, but CH5 has no such
selection line, so the user of the Channel Switch is required to do it
himself.

Earlier I said it was a partitioning problem. I believe the mistake
that happened was the Channel Switch hybrid designer, and the Display
hybrid designer each thought the other was responsible for grounding
out the display input when data was not being displayed. The possibility
is also pretty good that neither designed their part with the 2465, or
the other designer, in mind. They designed to spec.

The simple fix would have been to add VS5 to deselect the display channel.

-Chuck Harris

machine guy @Mac [TekScopes] wrote:
Chuck, I am sure you understand that "complex circuit" on Ch5 input to U400, the Channel Switch. I think I do but I will document it for others and I hope you will correct any errors.
U485A and U485B form a differential amplifier whose operation depends on current to their common emitters through R483. Diode CR484 is fitted to stop the amplifier by "stealing" the emitter current when U975 pin 1 is high. U485C and U485D form a similar amplifier that is disabled when the Pin 3 of U975 is high. The outputs from these two differential amplifers is joined by the common collectors of U485B and U485D. U975A acts to select one or the other (not both) of these two amplifiers. Amplifier One (U485A and U485B) carries the readout signal as presented on pin 2 of U485A. Amplifier Two (U485C and U485D) "amplifies" the ground that is present on pin 6 of U485C. Thus, the signal at pin 6 of U475C is either Readout Vertical or ground, depending on the logic state of U975A. U475C is a simple emitter follower to provide the current that drives the input to Ch5+ on U400 and other circuitry. That other circuitry includes another differential amplifier (unswitched this time) comprised of U475A and U475B that inverts the Ch5+ signal for input to U475D and on to Ch5-. U475D is another emitter follower to drive current into Ch5- on U400.
Bottom line, a somewhat complex yet straight forward "mux" to present either Readout Vertical or Ground to the inputs of U400 Ch5.
I was surprised by the number of components Tek used to accomplish such a simple task and aks the question: "why doesn't the Channel Switch ignore Ch5 inputs when Ch1 or Ch2 is selected?" I concluded that late in the development of the 2465 they discovered U400 "leaks" Ch5 into Ch1 and Ch2. The dies for U400 were proved and mass production had likely begun. The only solution was to block the "leak" by adding a switch in the pathway for Ch5 input so it is "grounded" when Ch1 or Ch2 is running. No time for special IC's and fidelity was important. So Tek built a kludge to mitigate the problem. It was cheap, and it worked, and it prevented a complex redesign of U400, so it stayed for the 2465A and 2465B.
Mac


[Non-text portions of this message have been removed]

 

We will probably never know for sure.  But it certainly isn't the way I or my associates would have designed it.  The function is basically an Op Amp inverter with an input mux.  And why would Ch5 ever be added to anything?  I'm not being critical and I don't claim to know, just saying that as an engineer I would have used something more simple.

On ‎Sunday‎, ‎November‎ ‎12‎, ‎2017‎ ‎10‎:‎42‎:‎03‎ ‎AM‎ ‎CST, Sigurður Ásgeirsson siggi@... [TekScopes] <TekScopes@...> wrote:

 
On Sun, 12 Nov 2017 at 11:05 Chuck Harris cfharris@... [TekScopes] <
TekScopes@...> wrote:



I don't believe the complexity in feeding CH5 is a mistake, but rather
an issue of partitioning... I believe the channel switch hybrid works
exactly the way it was supposed to work.
You can see the data sheet for the channel switch die here <
http://w140.com/tek_made_sm.pdf>, it's "named" 203-0211-90 and appears on
page 6-29 (or 271). It seems pretty clear that this is behaving as
intended, and that the #5 input is designed for a much lower bandwidth
signal than #1/2 and #3/4.
I'm guessing the original intent would have been for the readout signal to
be in a quiescent state when not displayed, or maybe it was simply less
expensive to do the low-bandwidth switching externally to the hybrid.






[Non-text portions of this message have been removed]