2467B Trace flicker


thespin@...
 

Hey all,

I have a 2467B whose psu has been recapped. The power supplies have been verified to have in spec ripple and voltages. I find however that the trace, and the trace only, flickers. The readout looks like it has stable intensity. Additionally, the intensity adjustment is very nonlinear and hysteretic, but the pot has been verified clean and functioning well. I understand that the pot’s position is digitized and then reexpressed elsewhere. I just am having trouble understanding this signal path. Any guesses as to what’s going on?

Evan


Michael A. Terrell
 

Disable the On Screen Display' and it won't flicker if the scope is OK. The OSD steals time from the normal display to function.

Michael A. Terrell
--

-----Original Message-----
From: thespin@gmail.com<reply@

Hey all,

I have a 2467B whose psu has been recapped. The power supplies have been verified to have in spec ripple and voltages. I find however that the trace, and the trace only, flickers. The readout looks like it has stable intensity. Additionally, the intensity adjustment is very nonlinear and hysteretic, but the pot has been verified clean and functioning well. I understand that the pot’s position is digitized and then reexpressed elsewhere. I just am having trouble understanding this signal path. Any guesses as to what’s going on?

Evan


Chuck Harris <cfharris@...>
 

You are describing normal behavior for the 2465B family
scopes. The reason they look choppy and erratic like that
is they have the same slow microprocessor as the much older,
and less lavish, 2465, but they are taxing it with a plenitude
of additional whiz bang features.

They are at a threshold where it is arguable that they have
exceeded the 6802's capabilities.

Basically, everything that happens in the 2465B is passed through
the A5 controller's 6802 microprocessor. The front panel is
entirely fly-by-wire. Pots, switches, and signals are selected
using the CPU and its MUX's, and digitized using the CPU, its DAC,
and the trigger hybrid. The CPU takes these digital signals, and
turns them back to analog using its DAC, and sample-and-hold cells,
and feeds them to the various circuits and hybrids that need the
signal.

Frequently, the CPU is too busy to get to digitizing a pot when
it should, and misses or delays that operation. This appears
to the user as the pot stuttering, or not responding.

-Chuck Harris

thespin@gmail.com wrote:

Hey all,

I have a 2467B whose psu has been recapped. The power supplies have been verified to have in spec ripple and voltages. I find however that the trace, and the trace only, flickers. The readout looks like it has stable intensity. Additionally, the intensity adjustment is very nonlinear and hysteretic, but the pot has been verified clean and functioning well. I understand that the pot’s position is digitized and then reexpressed elsewhere. I just am having trouble understanding this signal path. Any guesses as to what’s going on?

Evan




thespin@...
 

Apologies, I may be using the wrong terminology. The “flicker” isn’t high frequency harmonic content, but low frequency noise. I mean flicker not like the look of low frame rate but flicker like a candle. I suppose the former is what the term usually means.

This low frequency noise presents even when the OSD is disabled.

On Sun, Jan 13, 2019 at 11:15 PM, Chuck Harris wrote:


You are describing normal behavior for the 2465B family
scopes. The reason they look choppy and erratic like that
is they have the same slow microprocessor as the much older,
and less lavish, 2465, but they are taxing it with a plenitude
of additional whiz bang features.

They are at a threshold where it is arguable that they have
exceeded the 6802's capabilities.

Basically, everything that happens in the 2465B is passed through
the A5 controller's 6802 microprocessor. The front panel is
entirely fly-by-wire. Pots, switches, and signals are selected
using the CPU and its MUX's, and digitized using the CPU, its DAC,
and the trigger hybrid. The CPU takes these digital signals, and
turns them back to analog using its DAC, and sample-and-hold cells,
and feeds them to the various circuits and hybrids that need the
signal.

Frequently, the CPU is too busy to get to digitizing a pot when
it should, and misses or delays that operation. This appears
to the user as the pot stuttering, or not responding.

-Chuck Harris

thespin@gmail.com wrote:
Hey all,

I have a 2467B whose psu has been recapped. The power supplies have been
verified to have in spec ripple and voltages. I find however that the trace,
and the trace only, flickers. The readout looks like it has stable intensity.
Additionally, the intensity adjustment is very nonlinear and hysteretic, but
the pot has been verified clean and functioning well. I understand that the
pot’s position is digitized and then reexpressed elsewhere. I just am having
trouble understanding this signal path. Any guesses as to what’s going on?

Evan




thespin@...
 

I should also note that I scoped the output of U950 (Z axis hybrid)... I think either VG Out or VZ Out, not sure I was in a bit of a rush this morning. Turning the OSD off and trace on, the amplitude of this signal appeared to fluctuate with the display intensity. Monitoring this signal at startup, when only an error message is displaying, the amplitude of the signal appeared more stable.

On Mon, Jan 14, 2019 at 06:26 AM, <thespin@gmail.com> wrote:


Apologies, I may be using the wrong terminology. The “flicker” isn’t
high frequency harmonic content, but low frequency noise. I mean flicker not
like the look of low frame rate but flicker like a candle. I suppose the
former is what the term usually means.

This low frequency noise presents even when the OSD is disabled.

On Sun, Jan 13, 2019 at 11:15 PM, Chuck Harris wrote:


You are describing normal behavior for the 2465B family
scopes. The reason they look choppy and erratic like that
is they have the same slow microprocessor as the much older,
and less lavish, 2465, but they are taxing it with a plenitude
of additional whiz bang features.

They are at a threshold where it is arguable that they have
exceeded the 6802's capabilities.

Basically, everything that happens in the 2465B is passed through
the A5 controller's 6802 microprocessor. The front panel is
entirely fly-by-wire. Pots, switches, and signals are selected
using the CPU and its MUX's, and digitized using the CPU, its DAC,
and the trigger hybrid. The CPU takes these digital signals, and
turns them back to analog using its DAC, and sample-and-hold cells,
and feeds them to the various circuits and hybrids that need the
signal.

Frequently, the CPU is too busy to get to digitizing a pot when
it should, and misses or delays that operation. This appears
to the user as the pot stuttering, or not responding.

-Chuck Harris

thespin@gmail.com wrote:
Hey all,

I have a 2467B whose psu has been recapped. The power supplies have been
verified to have in spec ripple and voltages. I find however that the trace,
and the trace only, flickers. The readout looks like it has stable
intensity.
Additionally, the intensity adjustment is very nonlinear and hysteretic, but
the pot has been verified clean and functioning well. I understand that the
pot’s position is digitized and then reexpressed elsewhere. I just am
having
trouble understanding this signal path. Any guesses as to what’s going on?

Evan




Siggi
 

On Mon, Jan 14, 2019 at 7:21 AM <thespin@gmail.com> wrote:

I should also note that I scoped the output of U950 (Z axis hybrid)... I
think either VG Out or VZ Out, not sure I was in a bit of a rush this
morning. Turning the OSD off and trace on, the amplitude of this signal
appeared to fluctuate with the display intensity. Monitoring this signal at
startup, when only an error message is displaying, the amplitude of the
signal appeared more stable.
The brighness signals originate in the A5 board and are dependent on the
DAC and the sample-and-hold circuits on the A5. A common failure in the
2400B scopes is the SMD electrolytics on the A5 board leaking. The
electrolyte tends to leach into the components that make the DACs reference
and foul everything that uses the DAC.
If you haven't already attended these caps, that'd be the first port of
call.

You can conclusively verify the state of the DACs reference by measuring
the +1.36V and -1.25V references generated on the A5 board. I don't think
your version of the board will bring those out to test points, but if you
find these flicker with your display, there's your smoking gun.

The next thing to check might be the state of the relevant 4051 MUX
(U2530?) on the A5 board, the hold-up caps and op-amps, see whether that's
all hunky-dory.


Siggi
 

On Mon, Jan 14, 2019 at 7:21 AM <thespin@gmail.com> wrote:

Monitoring this signal at startup, when only an error message is
displaying, the amplitude of the signal appeared more stable.
Hold on, the scope is failing power-on tests? Not with a Test 05 error?
Those are typically the first sign that the DACs reference is horked.


thespin@...
 

Well ok, the error is because I have the options board removed and haven't installed the relevant jumpers to make everyone happy. When the board is properly installed, there are no startup errors and the flicker issue remains.

My A5 is all through-hole, and I've replaced all electrolytics on this board. The problem remains.

On Mon, Jan 14, 2019 at 10:43 AM, Siggi wrote:


On Mon, Jan 14, 2019 at 7:21 AM <thespin@gmail.com> wrote:

Monitoring this signal at startup, when only an error message is
displaying, the amplitude of the signal appeared more stable.
Hold on, the scope is failing power-on tests? Not with a Test 05 error?
Those are typically the first sign that the DACs reference is horked.


Siggi
 

On Mon, Jan 14, 2019 at 11:08 AM <thespin@gmail.com> wrote:

Well ok, the error is because I have the options board removed and haven't
installed the relevant jumpers to make everyone happy.
Right, I had the same on my 2467 when the CTT option was out. Had me
running in circles for a spell.


My A5 is all through-hole, and I've replaced all electrolytics on this
board. The problem remains.
In that case check the 1.36/-1.25V reference voltages - it's unlikely that
they're bad, but this is quick and easy to do. If those voltages are
spot-on, then you've effectively eliminated the DAC, it's reference and a
whole lot of other machinery.

Next port of call would be the 4051 MUX, its sampling cap, the op-amp and
its output. I assume you have N+1 scopes, so you can take a look at one
with the others?

If you put a scope on the input and the relevant output of the MUX, you
should see a periodic waveform on the input, and the output should follow
one of the levels on the input. As you wiggle the brightness pot, you
should see a corresponding change in a section of the input signal.

When you take a look a the op-amp, just make sure it's not railed. There
are decoupling caps on the brightness line on the A1 board, and I remember
a case or two where those went leaky and pulled on the line.

There's only one other easy possibility I can think of, which is to pull,
clean and re-seat the Z-axis hybrid. My 2467 was all kinds of wonky until I
did this for the hybrids. Poor thing had been warehoused for god knows how
long, so all the pots and buttons and sockets were scratchy or intermittent
and weird.


thespin@...
 

It sounds like as per usual, the thing to do is find the problem by continuously bisecting the signal path... but okay, now I understand the signal path better thanks to your explanations and a more careful reading of the manual.

The Z-Axis hybrid and sequencer, though, I understand one should be VERY careful with these. I'm worried about putting the right amount of torque on the nuts without destroying something. Is there a torque spec somewhere? And yeah, this is my 7th Tek scope. I can stop whenever I want I tells ya!

On Mon, Jan 14, 2019 at 11:26 AM, Siggi wrote:


On Mon, Jan 14, 2019 at 11:08 AM <thespin@gmail.com> wrote:

Well ok, the error is because I have the options board removed and haven't
installed the relevant jumpers to make everyone happy.
Right, I had the same on my 2467 when the CTT option was out. Had me
running in circles for a spell.


My A5 is all through-hole, and I've replaced all electrolytics on this
board. The problem remains.
In that case check the 1.36/-1.25V reference voltages - it's unlikely that
they're bad, but this is quick and easy to do. If those voltages are
spot-on, then you've effectively eliminated the DAC, it's reference and a
whole lot of other machinery.

Next port of call would be the 4051 MUX, its sampling cap, the op-amp and
its output. I assume you have N+1 scopes, so you can take a look at one
with the others?

If you put a scope on the input and the relevant output of the MUX, you
should see a periodic waveform on the input, and the output should follow
one of the levels on the input. As you wiggle the brightness pot, you
should see a corresponding change in a section of the input signal.

When you take a look a the op-amp, just make sure it's not railed. There
are decoupling caps on the brightness line on the A1 board, and I remember
a case or two where those went leaky and pulled on the line.

There's only one other easy possibility I can think of, which is to pull,
clean and re-seat the Z-axis hybrid. My 2467 was all kinds of wonky until I
did this for the hybrids. Poor thing had been warehoused for god knows how
long, so all the pots and buttons and sockets were scratchy or intermittent
and weird.


Siggi
 

On Mon, Jan 14, 2019 at 11:32 AM <thespin@gmail.com> wrote:

The Z-Axis hybrid and sequencer, though, I understand one should be VERY
careful with these. I'm worried about putting the right amount of torque on
the nuts without destroying something. Is there a torque spec somewhere?

Not that I'm aware of, though I don't think the nuts bear on the actual
hybrid at any point. I believe the nuts just pull the aluminum
heatsink/casing down until it bears on the plastic rim of the socket.
Meanwhile the ceramic hybrid only touches against the spring-loaded
contacts in the socket.
If you gronk on this enough to distort the aluminum casing, I could see
there being consequences, but you'd be ripping the retaining screws out of
the PCB at that point :).
I don't know how ESD sensitive the trigger hybrid is - but I didn't take
any chances with mine.

I don't remember re-seating the sequencer in my 2467, though it's been a
long time since I fixed'er up.


And yeah, this is my 7th Tek scope. I can stop whenever I want I tells ya!
Yeah, me too. Whenever I want.


 

On Mon, Jan 14, 2019 at 06:57 PM, Siggi wrote:



And yeah, this is my 7th Tek scope. I can stop whenever I want I tells ya!

Yeah, me too. Whenever I want.
I stop every single day!

Raymond


thespin@...
 

DIA line is very noisy from the S/H circuit. Seems to be correlated with display brightness, but it's hard to see. I reckon the input to the opamp is similarly noisy, but it's hard to see with the gain of my scope. Guess I need a 1A7? Not sure if I should blame the mux, the sampling caps, the opamp, or the resistors. Any thoughts?

On Mon, Jan 14, 2019 at 01:00 PM, Raymond Domp Frank wrote:


On Mon, Jan 14, 2019 at 06:57 PM, Siggi wrote:



And yeah, this is my 7th Tek scope. I can stop whenever I want I tells ya!

Yeah, me too. Whenever I want.
I stop every single day!

Raymond


 

The DIA line is the control processors "estimate" of the DI (Display Intensity) command from the front panel.  DI can be measured on either side of R2702.  If R2702 signal (DI) is not noisy and DIA is noisy then the problem is in the MUX/DEMUX circuitry or the sample and hold output.   If DI is noisy the problem is in the front panel pot or its power supply.  Its hard for the MUX/DEMUX to fail on just one signal, DIA unless it is a specific failure of a specific part (U2501, U2530, U2630, or the supporting passive components of these three devices). 
Its possible to observe the signal on pin 5 of U2630 if you have a high impedance test probe.  That signal should be 1/2 of DIA.

On ‎Tuesday‎, ‎January‎ ‎15‎, ‎2019‎ ‎10‎:‎51‎:‎35‎ ‎AM‎ ‎CST, thespin@gmail.com <thespin@gmail.com> wrote:

DIA line is very noisy from the S/H circuit. Seems to be correlated with display brightness, but it's hard to see. I reckon the input to the opamp is similarly noisy, but it's hard to see with the gain of my scope. Guess I need a 1A7? Not sure if I should blame the mux, the sampling caps, the opamp, or the resistors. Any thoughts?

On Mon, Jan 14, 2019 at 01:00 PM, Raymond Domp Frank wrote:


On Mon, Jan 14, 2019 at 06:57 PM, Siggi wrote:



And yeah, this is my 7th Tek scope. I can stop whenever I want I tells ya!

Yeah, me too. Whenever I want.
I stop every single day!

Raymond


thespin@...
 

I’m not sure how I missed this earlier, but DI looks noisy. It has lots of fast spikes in comparison to the OSD brightness pot.

On Tue, Jan 15, 2019 at 12:52 PM, machineguy59 wrote:


The DIA line is the control processors "estimate" of the DI (Display
Intensity) command from the front panel.  DI can be measured on either side
of R2702.  If R2702 signal (DI) is not noisy and DIA is noisy then the
problem is in the MUX/DEMUX circuitry or the sample and hold output.   If DI
is noisy the problem is in the front panel pot or its power supply.  Its hard
for the MUX/DEMUX to fail on just one signal, DIA unless it is a specific
failure of a specific part (U2501, U2530, U2630, or the supporting passive
components of these three devices). 
Its possible to observe the signal on pin 5 of U2630 if you have a high
impedance test probe.  That signal should be 1/2 of DIA. 
On ‎Tuesday‎, ‎January‎ ‎15‎, ‎2019‎
‎10‎:‎51‎:‎35‎ ‎AM‎ ‎CST, thespin@gmail.com
<thespin@gmail.com> wrote:

DIA line is very noisy from the S/H circuit. Seems to be correlated with
display brightness, but it's hard to see. I reckon the input to the opamp is
similarly noisy, but it's hard to see with the gain of my scope. Guess I need
a 1A7? Not sure if I should blame the mux, the sampling caps, the opamp, or
the resistors. Any thoughts?

On Mon, Jan 14, 2019 at 01:00 PM, Raymond Domp Frank wrote:


On Mon, Jan 14, 2019 at 06:57 PM, Siggi wrote:



And yeah, this is my 7th Tek scope. I can stop whenever I want I tells
ya!

Yeah, me too. Whenever I want.
I stop every single day!

Raymond


 

Those fast spikes on DI are sampled by the control processor at about a 10 Hz rate and become aliased down to about a 10 Hz signal on DIA.  So it seems to me your culprit is the control pot itself or a power feed (not likely).

On ‎Wednesday‎, ‎January‎ ‎16‎, ‎2019‎ ‎07‎:‎24‎:‎46‎ ‎AM‎ ‎CST, <thespin@gmail.com> wrote:

I’m not sure how I missed this earlier, but DI looks noisy. It has lots of fast spikes in comparison to the OSD brightness pot.

On Tue, Jan 15, 2019 at 12:52 PM, machineguy59 wrote:


  The DIA line is the control processors "estimate" of the DI (Display
Intensity) command from the front panel.  DI can be measured on either side
of R2702.  If R2702 signal (DI) is not noisy and DIA is noisy then the
problem is in the MUX/DEMUX circuitry or the sample and hold output.   If DI
is noisy the problem is in the front panel pot or its power supply.  Its hard
for the MUX/DEMUX to fail on just one signal, DIA unless it is a specific
failure of a specific part (U2501, U2530, U2630, or the supporting passive
components of these three devices). 
Its possible to observe the signal on pin 5 of U2630 if you have a high
impedance test probe.  That signal should be 1/2 of DIA. 
    On ‎Tuesday‎, ‎January‎ ‎15‎, ‎2019‎
‎10‎:‎51‎:‎35‎ ‎AM‎ ‎CST, thespin@gmail.com
<thespin@gmail.com> wrote:

  DIA line is very noisy from the S/H circuit. Seems to be correlated with
display brightness, but it's hard to see. I reckon the input to the opamp is
similarly noisy, but it's hard to see with the gain of my scope. Guess I need
a 1A7? Not sure if I should blame the mux, the sampling caps, the opamp, or
the resistors. Any thoughts?

On Mon, Jan 14, 2019 at 01:00 PM, Raymond Domp Frank wrote:


On Mon, Jan 14, 2019 at 06:57 PM, Siggi wrote:



And yeah, this is my 7th Tek scope. I can stop whenever I want I tells
ya!

Yeah, me too. Whenever I want.
I stop every single day!

Raymond


thespin@...
 

I'm afraid I don't really understand where the spikes could be coming from. The supply is happy and in tolerance. Unplugging the flex board that plugs into A1 so I can check the pot with a multimeter, it looks like the resistance is in spec, and it doesn't look like the wiper is noisy or anything. The spikes don't occur when turning the pot. They appear spontaneously. I'll try replacing it but even if this does fix it, I'd like to understand why.

On Wed, Jan 16, 2019 at 09:00 AM, machineguy59 wrote:


Those fast spikes on DI are sampled by the control processor at about a 10 Hz
rate and become aliased down to about a 10 Hz signal on DIA.  So it seems to
me your culprit is the control pot itself or a power feed (not likely).
On ‎Wednesday‎, ‎January‎ ‎16‎, ‎2019‎
‎07‎:‎24‎:‎46‎ ‎AM‎ ‎CST, <thespin@gmail.com> wrote:

I’m not sure how I missed this earlier, but DI looks noisy. It has lots of
fast spikes in comparison to the OSD brightness pot.

On Tue, Jan 15, 2019 at 12:52 PM, machineguy59 wrote:


  The DIA line is the control processors "estimate" of the DI (Display
Intensity) command from the front panel.  DI can be measured on either side
of R2702.  If R2702 signal (DI) is not noisy and DIA is noisy then the
problem is in the MUX/DEMUX circuitry or the sample and hold output.   If
DI
is noisy the problem is in the front panel pot or its power supply.  Its
hard
for the MUX/DEMUX to fail on just one signal, DIA unless it is a specific
failure of a specific part (U2501, U2530, U2630, or the supporting passive
components of these three devices). 
Its possible to observe the signal on pin 5 of U2630 if you have a high
impedance test probe.  That signal should be 1/2 of DIA. 
    On ‎Tuesday‎, ‎January‎ ‎15‎, ‎2019‎
‎10‎:‎51‎:‎35‎ ‎AM‎ ‎CST, thespin@gmail.com
<thespin@gmail.com> wrote:

  DIA line is very noisy from the S/H circuit. Seems to be correlated with
display brightness, but it's hard to see. I reckon the input to the opamp is
similarly noisy, but it's hard to see with the gain of my scope. Guess I
need
a 1A7? Not sure if I should blame the mux, the sampling caps, the opamp, or
the resistors. Any thoughts?

On Mon, Jan 14, 2019 at 01:00 PM, Raymond Domp Frank wrote:


On Mon, Jan 14, 2019 at 06:57 PM, Siggi wrote:



And yeah, this is my 7th Tek scope. I can stop whenever I want I tells
ya!

Yeah, me too. Whenever I want.
I stop every single day!

Raymond



Siggi
 

On Thu, Jan 17, 2019 at 2:08 PM <thespin@gmail.com> wrote:

I'm afraid I don't really understand where the spikes could be coming
from. The supply is happy and in tolerance. Unplugging the flex board that
plugs into A1 so I can check the pot with a multimeter, it looks like the
resistance is in spec, and it doesn't look like the wiper is noisy or
anything. The spikes don't occur when turning the pot. They appear
spontaneously. I'll try replacing it but even if this does fix it, I'd like
to understand why.

Hey Evan,
have you tried looking at the MUX input and output at the same time? If
this is a 10Hz refresh rate, then you might need to do this with a digital
scope or at least a storage scope of some description.
The MUX is supposed to only "open" to each S/H at one specific time of the
cyclic input signal, but if it's randomly (or determinstically) opening at
other times, you'd see what you're observing. If you look "across it" like
I'm suggesting, you'd see a correlation between the input signal and the
spike if it's opening at inopportune times.

If you're inclined to go shotgun on this, in your shoes I'd replace the MUX
to see whether that cleans it up. Those MUXes do fail, though the other
possibility is that the IO register that sets its destination address is
wonking out. Have you looked at the MUXes CS/CE and addressing lines?

Siggi


 

I would start with the easy stuff first.  With the scope operating, observe the noise on The +1.36 line at C2420 (the most convenient spot).  If that is noisy you have a noise reference for all panel controls.  Time to check why.  If +1.36 is quiet, check the DI signal at R2702.  See how it behaves as the intensity pot is turned up and down.  Try to learn the nature of the noise on the pot side of R2702.  Compare that with the noise on the U2501 side of R2702.  If noise at both points is roughly equal, place a largish (say 10 mfd.) paralleled with a 0.02 mfd capacitor between R2702 pot side and ground (clip leads are acceptable for this task).  Now evaluate the noise on both sides of R2702.  If there is still noise on the U2501 side of R2702 then the noise is likely coming from U2401.  Otherwise, it comes from the pot.  Pot wiring goes through the mainboard, A1, on its way to R2702.  Perhaps there is a loose or shorted connection in that pathway or in the pathway for +1.36 to the pot..

On ‎Thursday‎, ‎January‎ ‎17‎, ‎2019‎ ‎03‎:‎21‎:‎45‎ ‎PM‎ ‎CST, Siggi <siggi@undo.com> wrote:

On Thu, Jan 17, 2019 at 2:08 PM <thespin@gmail.com> wrote:

I'm afraid I don't really understand where the spikes could be coming
from. The supply is happy and in tolerance. Unplugging the flex board that
plugs into A1 so I can check the pot with a multimeter, it looks like the
resistance is in spec, and it doesn't look like the wiper is noisy or
anything. The spikes don't occur when turning the pot. They appear
spontaneously. I'll try replacing it but even if this does fix it, I'd like
to understand why.

Hey Evan,
have you tried looking at the MUX input and output at the same time? If
this is a 10Hz refresh rate, then you might need to do this with a digital
scope or at least a storage scope of some description.
The MUX is supposed to only "open" to each S/H at one specific time of the
cyclic input signal, but if it's randomly (or determinstically) opening at
other times, you'd see what you're observing. If you look "across it" like
I'm suggesting, you'd see a correlation between the input signal and the
spike if it's opening at inopportune times.

If you're inclined to go shotgun on this, in your shoes I'd replace the MUX
to see whether that cleans it up. Those MUXes do fail, though the other
possibility is that the IO register that sets its destination address is
wonking out. Have you looked at the MUXes CS/CE and addressing lines?

Siggi


victor.silva
 

Issues like you describe are usually related to a display sequencer (U650).
In my opinion I believe this is selected part. I see many 2465Bs with the display sequencer chip with a green thick magic marker line on it.
2445Bs do not normally have that mark on them.

--Victor